[stats wf] indicators across stats dbs & updates in the org ids #248

Closed
dimitris.pierrakos wants to merge 1742 commits from beta into beta2master_sept_2022
28 changed files with 1537 additions and 994 deletions
Showing only changes of commit 8e9493fad9 - Show all commits

View File

@ -40,6 +40,7 @@ public class Constants {
public static final String SDG_CLASS_NAME = "Sustainable Development Goals"; public static final String SDG_CLASS_NAME = "Sustainable Development Goals";
public static final String NULL = "NULL"; public static final String NULL = "NULL";
public static final String NA = "N/A";
public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper(); public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
@ -61,10 +62,16 @@ public class Constants {
.map((MapFunction<String, R>) value -> OBJECT_MAPPER.readValue(value, clazz), Encoders.bean(clazz)); .map((MapFunction<String, R>) value -> OBJECT_MAPPER.readValue(value, clazz), Encoders.bean(clazz));
} }
public static Subject getSubject(String sbj, String classid, String classname, public static Subject getSubject(String sbj, String classid, String classname, String diqualifierclassid,
String diqualifierclassid) { Boolean split) {
if (sbj == null || sbj.equals(NULL)) if (sbj == null || sbj.equals(NULL) || sbj.startsWith(NA))
return null; return null;
String trust = "";
String subject = sbj;
if (split) {
sbj = subject.split("@@")[0];
trust = subject.split("@@")[1];
}
Subject s = new Subject(); Subject s = new Subject();
s.setValue(sbj); s.setValue(sbj);
s s
@ -89,9 +96,14 @@ public class Constants {
UPDATE_CLASS_NAME, UPDATE_CLASS_NAME,
ModelConstants.DNET_PROVENANCE_ACTIONS, ModelConstants.DNET_PROVENANCE_ACTIONS,
ModelConstants.DNET_PROVENANCE_ACTIONS), ModelConstants.DNET_PROVENANCE_ACTIONS),
"")); trust));
return s; return s;
}
public static Subject getSubject(String sbj, String classid, String classname,
String diqualifierclassid) {
return getSubject(sbj, classid, classname, diqualifierclassid, false);
} }

View File

@ -75,9 +75,12 @@ public class GetFOSSparkJob implements Serializable {
fosData.map((MapFunction<Row, FOSDataModel>) r -> { fosData.map((MapFunction<Row, FOSDataModel>) r -> {
FOSDataModel fosDataModel = new FOSDataModel(); FOSDataModel fosDataModel = new FOSDataModel();
fosDataModel.setDoi(r.getString(0).toLowerCase()); fosDataModel.setDoi(r.getString(0).toLowerCase());
fosDataModel.setLevel1(r.getString(1)); fosDataModel.setLevel1(r.getString(2));
fosDataModel.setLevel2(r.getString(2)); fosDataModel.setLevel2(r.getString(3));
fosDataModel.setLevel3(r.getString(3)); fosDataModel.setLevel3(r.getString(4));
fosDataModel.setLevel4(r.getString(5));
fosDataModel.setScoreL3(String.valueOf(r.getDouble(6)));
fosDataModel.setScoreL4(String.valueOf(r.getDouble(7)));
return fosDataModel; return fosDataModel;
}, Encoders.bean(FOSDataModel.class)) }, Encoders.bean(FOSDataModel.class))
.write() .write()

View File

@ -1,178 +0,0 @@
package eu.dnetlib.dhp.actionmanager.createunresolvedentities;
import static eu.dnetlib.dhp.actionmanager.Constants.*;
import static eu.dnetlib.dhp.actionmanager.Constants.UPDATE_CLASS_NAME;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
import java.io.Serializable;
import java.util.Arrays;
import java.util.List;
import java.util.Optional;
import java.util.stream.Collectors;
import org.apache.commons.io.IOUtils;
import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.JavaSparkContext;
import org.apache.spark.api.java.function.MapFunction;
import org.apache.spark.sql.Encoders;
import org.apache.spark.sql.SaveMode;
import org.apache.spark.sql.SparkSession;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.fasterxml.jackson.databind.ObjectMapper;
import eu.dnetlib.dhp.actionmanager.bipmodel.BipScore;
import eu.dnetlib.dhp.actionmanager.bipmodel.score.deserializers.BipResultModel;
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
import eu.dnetlib.dhp.common.HdfsSupport;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.Instance;
import eu.dnetlib.dhp.schema.oaf.KeyValue;
import eu.dnetlib.dhp.schema.oaf.Measure;
import eu.dnetlib.dhp.schema.oaf.Result;
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
import eu.dnetlib.dhp.utils.DHPUtils;
public class PrepareBipFinder implements Serializable {
private static final Logger log = LoggerFactory.getLogger(PrepareBipFinder.class);
private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
public static void main(String[] args) throws Exception {
String jsonConfiguration = IOUtils
.toString(
PrepareBipFinder.class
.getResourceAsStream(
"/eu/dnetlib/dhp/actionmanager/createunresolvedentities/prepare_parameters.json"));
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
parser.parseArgument(args);
Boolean isSparkSessionManaged = Optional
.ofNullable(parser.get("isSparkSessionManaged"))
.map(Boolean::valueOf)
.orElse(Boolean.TRUE);
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
final String sourcePath = parser.get("sourcePath");
log.info("sourcePath {}: ", sourcePath);
final String outputPath = parser.get("outputPath");
log.info("outputPath {}: ", outputPath);
SparkConf conf = new SparkConf();
runWithSparkSession(
conf,
isSparkSessionManaged,
spark -> {
HdfsSupport.remove(outputPath, spark.sparkContext().hadoopConfiguration());
prepareResults(spark, sourcePath, outputPath);
});
}
private static void prepareResults(SparkSession spark, String inputPath, String outputPath) {
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<BipResultModel> bipDeserializeJavaRDD = sc
.textFile(inputPath)
.map(item -> OBJECT_MAPPER.readValue(item, BipResultModel.class));
spark
.createDataset(bipDeserializeJavaRDD.flatMap(entry -> entry.keySet().stream().map(key -> {
BipScore bs = new BipScore();
bs.setId(key);
bs.setScoreList(entry.get(key));
return bs;
}).collect(Collectors.toList()).iterator()).rdd(), Encoders.bean(BipScore.class))
.map((MapFunction<BipScore, Result>) v -> {
Result r = new Result();
final String cleanedPid = CleaningFunctions.normalizePidValue(DOI, v.getId());
r.setId(DHPUtils.generateUnresolvedIdentifier(v.getId(), DOI));
Instance inst = new Instance();
inst.setMeasures(getMeasure(v));
inst
.setPid(
Arrays
.asList(
OafMapperUtils
.structuredProperty(
cleanedPid,
OafMapperUtils
.qualifier(
DOI, DOI_CLASSNAME,
ModelConstants.DNET_PID_TYPES,
ModelConstants.DNET_PID_TYPES),
null)));
r.setInstance(Arrays.asList(inst));
r
.setDataInfo(
OafMapperUtils
.dataInfo(
false, null, true,
false,
OafMapperUtils
.qualifier(
ModelConstants.PROVENANCE_ENRICH,
null,
ModelConstants.DNET_PROVENANCE_ACTIONS,
ModelConstants.DNET_PROVENANCE_ACTIONS),
null));
return r;
}, Encoders.bean(Result.class))
.write()
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.json(outputPath + "/bip");
}
private static List<Measure> getMeasure(BipScore value) {
return value
.getScoreList()
.stream()
.map(score -> {
Measure m = new Measure();
m.setId(score.getId());
m
.setUnit(
score
.getUnit()
.stream()
.map(unit -> {
KeyValue kv = new KeyValue();
kv.setValue(unit.getValue());
kv.setKey(unit.getKey());
kv
.setDataInfo(
OafMapperUtils
.dataInfo(
false,
UPDATE_DATA_INFO_TYPE,
true,
false,
OafMapperUtils
.qualifier(
UPDATE_MEASURE_BIP_CLASS_ID,
UPDATE_CLASS_NAME,
ModelConstants.DNET_PROVENANCE_ACTIONS,
ModelConstants.DNET_PROVENANCE_ACTIONS),
""));
return kv;
})
.collect(Collectors.toList()));
return m;
})
.collect(Collectors.toList());
}
}

View File

@ -78,12 +78,20 @@ public class PrepareFOSSparkJob implements Serializable {
HashSet<String> level1 = new HashSet<>(); HashSet<String> level1 = new HashSet<>();
HashSet<String> level2 = new HashSet<>(); HashSet<String> level2 = new HashSet<>();
HashSet<String> level3 = new HashSet<>(); HashSet<String> level3 = new HashSet<>();
addLevels(level1, level2, level3, first); HashSet<String> level4 = new HashSet<>();
it.forEachRemaining(v -> addLevels(level1, level2, level3, v)); addLevels(level1, level2, level3, level4, first);
it.forEachRemaining(v -> addLevels(level1, level2, level3, level4, v));
List<Subject> sbjs = new ArrayList<>(); List<Subject> sbjs = new ArrayList<>();
level1.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID))); level1
level2.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID))); .forEach(l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID)));
level3.forEach(l -> sbjs.add(getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID))); level2
.forEach(l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID)));
level3
.forEach(
l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID, true)));
level4
.forEach(
l -> add(sbjs, getSubject(l, FOS_CLASS_ID, FOS_CLASS_NAME, UPDATE_SUBJECT_FOS_CLASS_ID, true)));
r.setSubject(sbjs); r.setSubject(sbjs);
r r
.setDataInfo( .setDataInfo(
@ -106,11 +114,18 @@ public class PrepareFOSSparkJob implements Serializable {
.json(outputPath + "/fos"); .json(outputPath + "/fos");
} }
private static void add(List<Subject> sbsjs, Subject sbj) {
if (sbj != null)
sbsjs.add(sbj);
}
private static void addLevels(HashSet<String> level1, HashSet<String> level2, HashSet<String> level3, private static void addLevels(HashSet<String> level1, HashSet<String> level2, HashSet<String> level3,
HashSet<String> level4,
FOSDataModel first) { FOSDataModel first) {
level1.add(first.getLevel1()); level1.add(first.getLevel1());
level2.add(first.getLevel2()); level2.add(first.getLevel2());
level3.add(first.getLevel3()); level3.add(first.getLevel3() + "@@" + first.getScoreL3());
level4.add(first.getLevel4() + "@@" + first.getScoreL4());
} }
} }

View File

@ -69,9 +69,9 @@ public class SparkSaveUnresolved implements Serializable {
.mapGroups((MapGroupsFunction<String, Result, Result>) (k, it) -> { .mapGroups((MapGroupsFunction<String, Result, Result>) (k, it) -> {
Result ret = it.next(); Result ret = it.next();
it.forEachRemaining(r -> { it.forEachRemaining(r -> {
if (r.getInstance() != null) { // if (r.getInstance() != null) {
ret.setInstance(r.getInstance()); // ret.setInstance(r.getInstance());
} // }
if (r.getSubject() != null) { if (r.getSubject() != null) {
if (ret.getSubject() != null) if (ret.getSubject() != null)
ret.getSubject().addAll(r.getSubject()); ret.getSubject().addAll(r.getSubject());

View File

@ -11,21 +11,43 @@ public class FOSDataModel implements Serializable {
private String doi; private String doi;
@CsvBindByPosition(position = 1) @CsvBindByPosition(position = 1)
// @CsvBindByName(column = "doi")
private String oaid;
@CsvBindByPosition(position = 2)
// @CsvBindByName(column = "level1") // @CsvBindByName(column = "level1")
private String level1; private String level1;
@CsvBindByPosition(position = 2) @CsvBindByPosition(position = 3)
// @CsvBindByName(column = "level2") // @CsvBindByName(column = "level2")
private String level2; private String level2;
@CsvBindByPosition(position = 3) @CsvBindByPosition(position = 4)
// @CsvBindByName(column = "level3") // @CsvBindByName(column = "level3")
private String level3; private String level3;
@CsvBindByPosition(position = 5)
// @CsvBindByName(column = "level3")
private String level4;
@CsvBindByPosition(position = 6)
private String scoreL3;
@CsvBindByPosition(position = 7)
private String scoreL4;
public FOSDataModel() { public FOSDataModel() {
} }
public FOSDataModel(String doi, String level1, String level2, String level3, String level4, String l3score,
String l4score) {
this.doi = doi;
this.level1 = level1;
this.level2 = level2;
this.level3 = level3;
this.level4 = level4;
this.scoreL3 = l3score;
this.scoreL4 = l4score;
}
public FOSDataModel(String doi, String level1, String level2, String level3) { public FOSDataModel(String doi, String level1, String level2, String level3) {
this.doi = doi; this.doi = doi;
this.level1 = level1; this.level1 = level1;
@ -33,8 +55,41 @@ public class FOSDataModel implements Serializable {
this.level3 = level3; this.level3 = level3;
} }
public static FOSDataModel newInstance(String d, String level1, String level2, String level3) { public static FOSDataModel newInstance(String d, String level1, String level2, String level3, String level4,
return new FOSDataModel(d, level1, level2, level3); String scorel3, String scorel4) {
return new FOSDataModel(d, level1, level2, level3, level4, scorel3, scorel4);
}
public String getOaid() {
return oaid;
}
public void setOaid(String oaid) {
this.oaid = oaid;
}
public String getLevel4() {
return level4;
}
public void setLevel4(String level4) {
this.level4 = level4;
}
public String getScoreL3() {
return scoreL3;
}
public void setScoreL3(String scoreL3) {
this.scoreL3 = scoreL3;
}
public String getScoreL4() {
return scoreL4;
}
public void setScoreL4(String scoreL4) {
this.scoreL4 = scoreL4;
} }
public String getDoi() { public String getDoi() {

View File

@ -5,11 +5,6 @@
<name>fosPath</name> <name>fosPath</name>
<description>the input path of the resources to be extended</description> <description>the input path of the resources to be extended</description>
</property> </property>
<property>
<name>bipScorePath</name>
<description>the path where to find the bipFinder scores</description>
</property>
<property> <property>
<name>outputPath</name> <name>outputPath</name>
<description>the path where to store the actionset</description> <description>the path where to store the actionset</description>
@ -77,35 +72,10 @@
<fork name="prepareInfo"> <fork name="prepareInfo">
<path start="prepareBip"/>
<path start="getFOS"/> <path start="getFOS"/>
<path start="getSDG"/> <path start="getSDG"/>
</fork> </fork>
<action name="prepareBip">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
<mode>cluster</mode>
<name>Produces the unresolved from BIP! Finder</name>
<class>eu.dnetlib.dhp.actionmanager.createunresolvedentities.PrepareBipFinder</class>
<jar>dhp-aggregation-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
</spark-opts>
<arg>--sourcePath</arg><arg>${bipScorePath}</arg>
<arg>--outputPath</arg><arg>${workingDir}/prepared</arg>
</spark>
<ok to="join"/>
<error to="Kill"/>
</action>
<action name="getFOS"> <action name="getFOS">
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master> <master>yarn</master>
@ -125,6 +95,7 @@
</spark-opts> </spark-opts>
<arg>--sourcePath</arg><arg>${fosPath}</arg> <arg>--sourcePath</arg><arg>${fosPath}</arg>
<arg>--outputPath</arg><arg>${workingDir}/input/fos</arg> <arg>--outputPath</arg><arg>${workingDir}/input/fos</arg>
<arg>--delimiter</arg><arg>${delimiter}</arg>
</spark> </spark>
<ok to="prepareFos"/> <ok to="prepareFos"/>
<error to="Kill"/> <error to="Kill"/>
@ -213,7 +184,7 @@
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master> <master>yarn</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Saves the result produced for bip and fos by grouping results with the same id</name> <name>Save the unresolved entities grouping results with the same id</name>
<class>eu.dnetlib.dhp.actionmanager.createunresolvedentities.SparkSaveUnresolved</class> <class>eu.dnetlib.dhp.actionmanager.createunresolvedentities.SparkSaveUnresolved</class>
<jar>dhp-aggregation-${projectVersion}.jar</jar> <jar>dhp-aggregation-${projectVersion}.jar</jar>
<spark-opts> <spark-opts>

View File

@ -13,10 +13,7 @@ import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaRDD; import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.JavaSparkContext; import org.apache.spark.api.java.JavaSparkContext;
import org.apache.spark.sql.SparkSession; import org.apache.spark.sql.SparkSession;
import org.junit.jupiter.api.AfterAll; import org.junit.jupiter.api.*;
import org.junit.jupiter.api.Assertions;
import org.junit.jupiter.api.BeforeAll;
import org.junit.jupiter.api.Test;
import org.slf4j.Logger; import org.slf4j.Logger;
import org.slf4j.LoggerFactory; import org.slf4j.LoggerFactory;
@ -68,6 +65,7 @@ public class GetFosTest {
} }
@Test @Test
@Disabled
void test3() throws Exception { void test3() throws Exception {
final String sourcePath = getClass() final String sourcePath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos_sbs.tsv") .getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos_sbs.tsv")
@ -96,4 +94,37 @@ public class GetFosTest {
tmp.foreach(t -> Assertions.assertTrue(t.getLevel3() != null)); tmp.foreach(t -> Assertions.assertTrue(t.getLevel3() != null));
} }
@Test
void test4() throws Exception {
final String sourcePath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos_sbs2.csv")
.getPath();
final String outputPath = workingDir.toString() + "/fos.json";
GetFOSSparkJob
.main(
new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", sourcePath,
"--delimiter", ",",
"-outputPath", outputPath
});
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<FOSDataModel> tmp = sc
.textFile(outputPath)
.map(item -> OBJECT_MAPPER.readValue(item, FOSDataModel.class));
tmp.foreach(t -> Assertions.assertTrue(t.getDoi() != null));
tmp.foreach(t -> Assertions.assertTrue(t.getLevel1() != null));
tmp.foreach(t -> Assertions.assertTrue(t.getLevel2() != null));
tmp.foreach(t -> Assertions.assertTrue(t.getLevel3() != null));
tmp.foreach(t -> Assertions.assertTrue(t.getLevel4() != null));
tmp.foreach(t -> Assertions.assertTrue(t.getScoreL3() != null));
tmp.foreach(t -> Assertions.assertTrue(t.getScoreL4() != null));
}
} }

View File

@ -67,92 +67,6 @@ public class PrepareTest {
spark.stop(); spark.stop();
} }
@Test
void bipPrepareTest() throws Exception {
final String sourcePath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/bip/bip.json")
.getPath();
PrepareBipFinder
.main(
new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", sourcePath,
"--outputPath", workingDir.toString() + "/work"
});
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<Result> tmp = sc
.textFile(workingDir.toString() + "/work/bip")
.map(item -> OBJECT_MAPPER.readValue(item, Result.class));
Assertions.assertEquals(86, tmp.count());
String doi1 = "unresolved::10.0000/096020199389707::doi";
Assertions.assertEquals(1, tmp.filter(r -> r.getId().equals(doi1)).count());
Assertions.assertEquals(1, tmp.filter(r -> r.getId().equals(doi1)).collect().get(0).getInstance().size());
Assertions
.assertEquals(
3, tmp.filter(r -> r.getId().equals(doi1)).collect().get(0).getInstance().get(0).getMeasures().size());
Assertions
.assertEquals(
"6.34596412687e-09", tmp
.filter(r -> r.getId().equals(doi1))
.collect()
.get(0)
.getInstance()
.get(0)
.getMeasures()
.stream()
.filter(sl -> sl.getId().equals("influence"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"0.641151896994", tmp
.filter(r -> r.getId().equals(doi1))
.collect()
.get(0)
.getInstance()
.get(0)
.getMeasures()
.stream()
.filter(sl -> sl.getId().equals("popularity_alt"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
Assertions
.assertEquals(
"2.33375102921e-09", tmp
.filter(r -> r.getId().equals(doi1))
.collect()
.get(0)
.getInstance()
.get(0)
.getMeasures()
.stream()
.filter(sl -> sl.getId().equals("popularity"))
.collect(Collectors.toList())
.get(0)
.getUnit()
.get(0)
.getValue());
final String doi2 = "unresolved::10.3390/s18072310::doi";
Assertions.assertEquals(1, tmp.filter(r -> r.getId().equals(doi2)).count());
Assertions.assertEquals(1, tmp.filter(r -> r.getId().equals(doi2)).collect().get(0).getInstance().size());
}
@Test @Test
void fosPrepareTest() throws Exception { void fosPrepareTest() throws Exception {
final String sourcePath = getClass() final String sourcePath = getClass()
@ -222,6 +136,76 @@ public class PrepareTest {
} }
@Test
void fosPrepareTest2() throws Exception {
final String sourcePath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos_sbs_2.json")
.getPath();
PrepareFOSSparkJob
.main(
new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", sourcePath,
"-outputPath", workingDir.toString() + "/work"
});
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<Result> tmp = sc
.textFile(workingDir.toString() + "/work/fos")
.map(item -> OBJECT_MAPPER.readValue(item, Result.class));
String doi1 = "unresolved::10.1016/j.revmed.2006.07.012::doi";
assertEquals(13, tmp.count());
assertEquals(1, tmp.filter(row -> row.getId().equals(doi1)).count());
Result result = tmp
.filter(r -> r.getId().equals(doi1))
.first();
result.getSubject().forEach(s -> System.out.println(s.getValue() + " trust = " + s.getDataInfo().getTrust()));
Assertions.assertEquals(6, result.getSubject().size());
assertTrue(
result
.getSubject()
.stream()
.anyMatch(
s -> s.getValue().contains("03 medical and health sciences")
&& s.getDataInfo().getTrust().equals("")));
assertTrue(
result
.getSubject()
.stream()
.anyMatch(
s -> s.getValue().contains("0302 clinical medicine") && s.getDataInfo().getTrust().equals("")));
assertTrue(
result
.getSubject()
.stream()
.anyMatch(
s -> s
.getValue()
.contains("030204 cardiovascular system & hematology")
&& s.getDataInfo().getTrust().equals("0.5101401805877686")));
assertTrue(
result
.getSubject()
.stream()
.anyMatch(
s -> s
.getValue()
.contains("03020409 Hematology/Coagulopathies")
&& s.getDataInfo().getTrust().equals("0.0546871414174914")));
}
@Test @Test
void sdgPrepareTest() throws Exception { void sdgPrepareTest() throws Exception {
final String sourcePath = getClass() final String sourcePath = getClass()
@ -268,57 +252,4 @@ public class PrepareTest {
} }
// @Test
// void test3() throws Exception {
// final String sourcePath = "/Users/miriam.baglioni/Downloads/doi_fos_results_20_12_2021.csv.gz";
//
// final String outputPath = workingDir.toString() + "/fos.json";
// GetFOSSparkJob
// .main(
// new String[] {
// "--isSparkSessionManaged", Boolean.FALSE.toString(),
// "--sourcePath", sourcePath,
//
// "-outputPath", outputPath
//
// });
//
// final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
//
// JavaRDD<FOSDataModel> tmp = sc
// .textFile(outputPath)
// .map(item -> OBJECT_MAPPER.readValue(item, FOSDataModel.class));
//
// tmp.foreach(t -> Assertions.assertTrue(t.getDoi() != null));
// tmp.foreach(t -> Assertions.assertTrue(t.getLevel1() != null));
// tmp.foreach(t -> Assertions.assertTrue(t.getLevel2() != null));
// tmp.foreach(t -> Assertions.assertTrue(t.getLevel3() != null));
//
// }
//
// @Test
// void test4() throws Exception {
// final String sourcePath = "/Users/miriam.baglioni/Downloads/doi_sdg_results_20_12_21.csv.gz";
//
// final String outputPath = workingDir.toString() + "/sdg.json";
// GetSDGSparkJob
// .main(
// new String[] {
// "--isSparkSessionManaged", Boolean.FALSE.toString(),
// "--sourcePath", sourcePath,
//
// "-outputPath", outputPath
//
// });
//
// final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
//
// JavaRDD<SDGDataModel> tmp = sc
// .textFile(outputPath)
// .map(item -> OBJECT_MAPPER.readValue(item, SDGDataModel.class));
//
// tmp.foreach(t -> Assertions.assertTrue(t.getDoi() != null));
// tmp.foreach(t -> Assertions.assertTrue(t.getSbj() != null));
//
// }
} }

View File

@ -340,18 +340,7 @@ public class ProduceTest {
} }
private JavaRDD<Result> getResultJavaRDD() throws Exception { private JavaRDD<Result> getResultJavaRDD() throws Exception {
final String bipPath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/bip/bip.json")
.getPath();
PrepareBipFinder
.main(
new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", bipPath,
"--outputPath", workingDir.toString() + "/work"
});
final String fosPath = getClass() final String fosPath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos.json") .getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos.json")
.getPath(); .getPath();
@ -379,6 +368,40 @@ public class ProduceTest {
.map(item -> OBJECT_MAPPER.readValue(item, Result.class)); .map(item -> OBJECT_MAPPER.readValue(item, Result.class));
} }
@Test
public JavaRDD<Result> getResultFosJavaRDD() throws Exception {
final String fosPath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos_sbs_2.json")
.getPath();
PrepareFOSSparkJob
.main(
new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", fosPath,
"-outputPath", workingDir.toString() + "/work"
});
SparkSaveUnresolved.main(new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", workingDir.toString() + "/work",
"-outputPath", workingDir.toString() + "/unresolved"
});
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
JavaRDD<Result> tmp = sc
.textFile(workingDir.toString() + "/unresolved")
.map(item -> OBJECT_MAPPER.readValue(item, Result.class));
tmp.foreach(r -> System.out.println(new ObjectMapper().writeValueAsString(r)));
return tmp;
}
@Test @Test
void prepareTest5Subjects() throws Exception { void prepareTest5Subjects() throws Exception {
final String doi = "unresolved::10.1063/5.0032658::doi"; final String doi = "unresolved::10.1063/5.0032658::doi";
@ -415,18 +438,7 @@ public class ProduceTest {
} }
private JavaRDD<Result> getResultJavaRDDPlusSDG() throws Exception { private JavaRDD<Result> getResultJavaRDDPlusSDG() throws Exception {
final String bipPath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/bip/bip.json")
.getPath();
PrepareBipFinder
.main(
new String[] {
"--isSparkSessionManaged", Boolean.FALSE.toString(),
"--sourcePath", bipPath,
"--outputPath", workingDir.toString() + "/work"
});
final String fosPath = getClass() final String fosPath = getClass()
.getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos.json") .getResource("/eu/dnetlib/dhp/actionmanager/createunresolvedentities/fos/fos.json")
.getPath(); .getPath();
@ -483,14 +495,6 @@ public class ProduceTest {
.filter(row -> row.getSubject() != null) .filter(row -> row.getSubject() != null)
.count()); .count());
Assertions
.assertEquals(
85,
tmp
.filter(row -> !row.getId().equals(doi))
.filter(r -> r.getInstance() != null && r.getInstance().size() > 0)
.count());
} }
@Test @Test

View File

@ -0,0 +1,26 @@
DOI,OAID,level1,level2,level3,level4,score_for_L3,score_for_L4
10.1016/j.anucene.2006.02.004,doi_________::00059d9963edf633bec756fb21b5bd72,02 engineering and technology,"0202 electrical engineering, electronic engineering, information engineering",020209 energy,02020908 Climate change policy/Ethanol fuel,0.5,0.5
10.1016/j.anucene.2006.02.004,doi_________::00059d9963edf633bec756fb21b5bd72,02 engineering and technology,0211 other engineering and technologies,021108 energy,02110808 Climate change policy/Ethanol fuel,0.5,0.5
10.1016/j.revmed.2006.07.010,doi_________::0026476c1651a92c933d752ff12496c7,03 medical and health sciences,0302 clinical medicine,030220 oncology & carcinogenesis,N/A,0.5036656856536865,0.0
10.1016/j.revmed.2006.07.010,doi_________::0026476c1651a92c933d752ff12496c7,03 medical and health sciences,0302 clinical medicine,030212 general & internal medicine,N/A,0.4963343143463135,0.0
10.20965/jrm.2006.p0312,doi_________::0028336a2f3826cc83c47dbefac71543,02 engineering and technology,0209 industrial biotechnology,020901 industrial engineering & automation,02090104 Robotics/Robots,0.6111094951629639,0.5053805979936855
10.20965/jrm.2006.p0312,doi_________::0028336a2f3826cc83c47dbefac71543,01 natural sciences,0104 chemical sciences,010401 analytical chemistry,N/A,0.3888905048370361,0.0
10.1111/j.1747-7379.2006.040_1.x,doi_________::002c7077e7c114a8304eb90f59e45fa4,05 social sciences,0506 political science,050602 political science & public administration,05060202 Ethnic groups/Ethnicity,0.6159052848815918,0.7369035568037298
10.1111/j.1747-7379.2006.040_1.x,doi_________::002c7077e7c114a8304eb90f59e45fa4,05 social sciences,0502 economics and business,050207 economics,N/A,0.3840946555137634,0.0
10.1007/s10512-006-0049-9,doi_________::003f29f9254819cf4c78558b1bc25f10,02 engineering and technology,"0202 electrical engineering, electronic engineering, information engineering",020209 energy,02020908 Climate change policy/Ethanol fuel,0.5,0.5
10.1007/s10512-006-0049-9,doi_________::003f29f9254819cf4c78558b1bc25f10,02 engineering and technology,0211 other engineering and technologies,021108 energy,02110808 Climate change policy/Ethanol fuel,0.5,0.5
10.1111/j.1365-2621.2005.01045.x,doi_________::00419355b4c3e0646bd0e1b301164c8e,04 agricultural and veterinary sciences,0404 agricultural biotechnology,040401 food science,04040102 Food science/Food industry,0.5,0.5
10.1111/j.1365-2621.2005.01045.x,doi_________::00419355b4c3e0646bd0e1b301164c8e,04 agricultural and veterinary sciences,0405 other agricultural sciences,040502 food science,04050202 Food science/Food industry,0.5,0.5
10.1002/chin.200617262,doi_________::004c8cef80668904961b9e62841793c8,01 natural sciences,0104 chemical sciences,010405 organic chemistry,01040508 Functional groups/Ethers,0.5566747188568115,0.5582916736602783
10.1002/chin.200617262,doi_________::004c8cef80668904961b9e62841793c8,01 natural sciences,0104 chemical sciences,010402 general chemistry,01040207 Chemical synthesis/Total synthesis,0.4433253407478332,0.4417082965373993
10.1016/j.revmed.2006.07.012,doi_________::005b1d0fb650b680abaf6cfe26a21604,03 medical and health sciences,0302 clinical medicine,030204 cardiovascular system & hematology,03020409 Hematology/Coagulopathies,0.5101401805877686,0.0546871414174914
10.1016/j.revmed.2006.07.012,doi_________::005b1d0fb650b680abaf6cfe26a21604,03 medical and health sciences,0301 basic medicine,030105 genetics & heredity,N/A,0.4898599088191986,0.0
10.4109/jslab.17.132,doi_________::00889baa06de363e37930daaf8e800c0,03 medical and health sciences,0301 basic medicine,030104 developmental biology,N/A,0.5,0.0
10.4109/jslab.17.132,doi_________::00889baa06de363e37930daaf8e800c0,03 medical and health sciences,0303 health sciences,030304 developmental biology,N/A,0.5,0.0
10.1108/00251740610715687,doi_________::0092cb1b1920d556719385a26363ecaa,05 social sciences,0502 economics and business,050203 business & management,05020311 International business/International trade,0.605047881603241,0.2156608108845153
10.1108/00251740610715687,doi_________::0092cb1b1920d556719385a26363ecaa,05 social sciences,0502 economics and business,050211 marketing,N/A,0.394952118396759,0.0
10.1080/03067310500248098,doi_________::00a76678d230e3f20b6356804448028f,04 agricultural and veterinary sciences,0404 agricultural biotechnology,040401 food science,04040102 Food science/Food industry,0.5,0.5
10.1080/03067310500248098,doi_________::00a76678d230e3f20b6356804448028f,04 agricultural and veterinary sciences,0405 other agricultural sciences,040502 food science,04050202 Food science/Food industry,0.5,0.5
10.3152/147154306781778533,doi_________::00acc520f3939e5a6675343881fed4f2,05 social sciences,0502 economics and business,050203 business & management,05020307 Innovation/Product management,0.5293408632278442,0.5326762795448303
10.3152/147154306781778533,doi_________::00acc520f3939e5a6675343881fed4f2,05 social sciences,0509 other social sciences,050905 science studies,05090502 Social philosophy/Capitalism,0.4706590473651886,0.4673237204551697
10.1785/0120050806,doi_________::00d5831d329e7ae4523d78bfc3042e98,02 engineering and technology,0211 other engineering and technologies,021101 geological & geomatics engineering,02110103 Concrete/Building materials,0.5343400835990906,0.3285667930180677
1 DOI OAID level1 level2 level3 level4 score_for_L3 score_for_L4
2 10.1016/j.anucene.2006.02.004 doi_________::00059d9963edf633bec756fb21b5bd72 02 engineering and technology 0202 electrical engineering, electronic engineering, information engineering 020209 energy 02020908 Climate change policy/Ethanol fuel 0.5 0.5
3 10.1016/j.anucene.2006.02.004 doi_________::00059d9963edf633bec756fb21b5bd72 02 engineering and technology 0211 other engineering and technologies 021108 energy 02110808 Climate change policy/Ethanol fuel 0.5 0.5
4 10.1016/j.revmed.2006.07.010 doi_________::0026476c1651a92c933d752ff12496c7 03 medical and health sciences 0302 clinical medicine 030220 oncology & carcinogenesis N/A 0.5036656856536865 0.0
5 10.1016/j.revmed.2006.07.010 doi_________::0026476c1651a92c933d752ff12496c7 03 medical and health sciences 0302 clinical medicine 030212 general & internal medicine N/A 0.4963343143463135 0.0
6 10.20965/jrm.2006.p0312 doi_________::0028336a2f3826cc83c47dbefac71543 02 engineering and technology 0209 industrial biotechnology 020901 industrial engineering & automation 02090104 Robotics/Robots 0.6111094951629639 0.5053805979936855
7 10.20965/jrm.2006.p0312 doi_________::0028336a2f3826cc83c47dbefac71543 01 natural sciences 0104 chemical sciences 010401 analytical chemistry N/A 0.3888905048370361 0.0
8 10.1111/j.1747-7379.2006.040_1.x doi_________::002c7077e7c114a8304eb90f59e45fa4 05 social sciences 0506 political science 050602 political science & public administration 05060202 Ethnic groups/Ethnicity 0.6159052848815918 0.7369035568037298
9 10.1111/j.1747-7379.2006.040_1.x doi_________::002c7077e7c114a8304eb90f59e45fa4 05 social sciences 0502 economics and business 050207 economics N/A 0.3840946555137634 0.0
10 10.1007/s10512-006-0049-9 doi_________::003f29f9254819cf4c78558b1bc25f10 02 engineering and technology 0202 electrical engineering, electronic engineering, information engineering 020209 energy 02020908 Climate change policy/Ethanol fuel 0.5 0.5
11 10.1007/s10512-006-0049-9 doi_________::003f29f9254819cf4c78558b1bc25f10 02 engineering and technology 0211 other engineering and technologies 021108 energy 02110808 Climate change policy/Ethanol fuel 0.5 0.5
12 10.1111/j.1365-2621.2005.01045.x doi_________::00419355b4c3e0646bd0e1b301164c8e 04 agricultural and veterinary sciences 0404 agricultural biotechnology 040401 food science 04040102 Food science/Food industry 0.5 0.5
13 10.1111/j.1365-2621.2005.01045.x doi_________::00419355b4c3e0646bd0e1b301164c8e 04 agricultural and veterinary sciences 0405 other agricultural sciences 040502 food science 04050202 Food science/Food industry 0.5 0.5
14 10.1002/chin.200617262 doi_________::004c8cef80668904961b9e62841793c8 01 natural sciences 0104 chemical sciences 010405 organic chemistry 01040508 Functional groups/Ethers 0.5566747188568115 0.5582916736602783
15 10.1002/chin.200617262 doi_________::004c8cef80668904961b9e62841793c8 01 natural sciences 0104 chemical sciences 010402 general chemistry 01040207 Chemical synthesis/Total synthesis 0.4433253407478332 0.4417082965373993
16 10.1016/j.revmed.2006.07.012 doi_________::005b1d0fb650b680abaf6cfe26a21604 03 medical and health sciences 0302 clinical medicine 030204 cardiovascular system & hematology 03020409 Hematology/Coagulopathies 0.5101401805877686 0.0546871414174914
17 10.1016/j.revmed.2006.07.012 doi_________::005b1d0fb650b680abaf6cfe26a21604 03 medical and health sciences 0301 basic medicine 030105 genetics & heredity N/A 0.4898599088191986 0.0
18 10.4109/jslab.17.132 doi_________::00889baa06de363e37930daaf8e800c0 03 medical and health sciences 0301 basic medicine 030104 developmental biology N/A 0.5 0.0
19 10.4109/jslab.17.132 doi_________::00889baa06de363e37930daaf8e800c0 03 medical and health sciences 0303 health sciences 030304 developmental biology N/A 0.5 0.0
20 10.1108/00251740610715687 doi_________::0092cb1b1920d556719385a26363ecaa 05 social sciences 0502 economics and business 050203 business & management 05020311 International business/International trade 0.605047881603241 0.2156608108845153
21 10.1108/00251740610715687 doi_________::0092cb1b1920d556719385a26363ecaa 05 social sciences 0502 economics and business 050211 marketing N/A 0.394952118396759 0.0
22 10.1080/03067310500248098 doi_________::00a76678d230e3f20b6356804448028f 04 agricultural and veterinary sciences 0404 agricultural biotechnology 040401 food science 04040102 Food science/Food industry 0.5 0.5
23 10.1080/03067310500248098 doi_________::00a76678d230e3f20b6356804448028f 04 agricultural and veterinary sciences 0405 other agricultural sciences 040502 food science 04050202 Food science/Food industry 0.5 0.5
24 10.3152/147154306781778533 doi_________::00acc520f3939e5a6675343881fed4f2 05 social sciences 0502 economics and business 050203 business & management 05020307 Innovation/Product management 0.5293408632278442 0.5326762795448303
25 10.3152/147154306781778533 doi_________::00acc520f3939e5a6675343881fed4f2 05 social sciences 0509 other social sciences 050905 science studies 05090502 Social philosophy/Capitalism 0.4706590473651886 0.4673237204551697
26 10.1785/0120050806 doi_________::00d5831d329e7ae4523d78bfc3042e98 02 engineering and technology 0211 other engineering and technologies 021101 geological & geomatics engineering 02110103 Concrete/Building materials 0.5343400835990906 0.3285667930180677

View File

@ -0,0 +1,25 @@
{"doi":"10.1016/j.anucene.2006.02.004","level1":"02 engineering and technology","level2":"0202 electrical engineering, electronic engineering, information engineering","level3":"020209 energy","level4":"02020908 Climate change policy/Ethanol fuel","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1016/j.anucene.2006.02.004","level1":"02 engineering and technology","level2":"0211 other engineering and technologies","level3":"021108 energy","level4":"02110808 Climate change policy/Ethanol fuel","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1016/j.revmed.2006.07.010","level1":"03 medical and health sciences","level2":"0302 clinical medicine","level3":"030220 oncology & carcinogenesis","level4":"N/A","scoreL3":"0.5036656856536865","scoreL4":"0.0"}
{"doi":"10.1016/j.revmed.2006.07.010","level1":"03 medical and health sciences","level2":"0302 clinical medicine","level3":"030212 general & internal medicine","level4":"N/A","scoreL3":"0.4963343143463135","scoreL4":"0.0"}
{"doi":"10.20965/jrm.2006.p0312","level1":"02 engineering and technology","level2":"0209 industrial biotechnology","level3":"020901 industrial engineering & automation","level4":"02090104 Robotics/Robots","scoreL3":"0.6111094951629639","scoreL4":"0.5053805979936855"}
{"doi":"10.20965/jrm.2006.p0312","level1":"01 natural sciences","level2":"0104 chemical sciences","level3":"010401 analytical chemistry","level4":"N/A","scoreL3":"0.3888905048370361","scoreL4":"0.0"}
{"doi":"10.1111/j.1747-7379.2006.040_1.x","level1":"05 social sciences","level2":"0506 political science","level3":"050602 political science & public administration","level4":"05060202 Ethnic groups/Ethnicity","scoreL3":"0.6159052848815918","scoreL4":"0.7369035568037298"}
{"doi":"10.1111/j.1747-7379.2006.040_1.x","level1":"05 social sciences","level2":"0502 economics and business","level3":"050207 economics","level4":"N/A","scoreL3":"0.3840946555137634","scoreL4":"0.0"}
{"doi":"10.1007/s10512-006-0049-9","level1":"02 engineering and technology","level2":"0202 electrical engineering, electronic engineering, information engineering","level3":"020209 energy","level4":"02020908 Climate change policy/Ethanol fuel","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1007/s10512-006-0049-9","level1":"02 engineering and technology","level2":"0211 other engineering and technologies","level3":"021108 energy","level4":"02110808 Climate change policy/Ethanol fuel","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1111/j.1365-2621.2005.01045.x","level1":"04 agricultural and veterinary sciences","level2":"0404 agricultural biotechnology","level3":"040401 food science","level4":"04040102 Food science/Food industry","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1111/j.1365-2621.2005.01045.x","level1":"04 agricultural and veterinary sciences","level2":"0405 other agricultural sciences","level3":"040502 food science","level4":"04050202 Food science/Food industry","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1002/chin.200617262","level1":"01 natural sciences","level2":"0104 chemical sciences","level3":"010405 organic chemistry","level4":"01040508 Functional groups/Ethers","scoreL3":"0.5566747188568115","scoreL4":"0.5582916736602783"}
{"doi":"10.1002/chin.200617262","level1":"01 natural sciences","level2":"0104 chemical sciences","level3":"010402 general chemistry","level4":"01040207 Chemical synthesis/Total synthesis","scoreL3":"0.4433253407478332","scoreL4":"0.4417082965373993"}
{"doi":"10.1016/j.revmed.2006.07.012","level1":"03 medical and health sciences","level2":"0302 clinical medicine","level3":"030204 cardiovascular system & hematology","level4":"03020409 Hematology/Coagulopathies","scoreL3":"0.5101401805877686","scoreL4":"0.0546871414174914"}
{"doi":"10.1016/j.revmed.2006.07.012","level1":"03 medical and health sciences","level2":"0301 basic medicine","level3":"030105 genetics & heredity","level4":"N/A","scoreL3":"0.4898599088191986","scoreL4":"0.0"}
{"doi":"10.4109/jslab.17.132","level1":"03 medical and health sciences","level2":"0301 basic medicine","level3":"030104 developmental biology","level4":"N/A","scoreL3":"0.5","scoreL4":"0.0"}
{"doi":"10.4109/jslab.17.132","level1":"03 medical and health sciences","level2":"0303 health sciences","level3":"030304 developmental biology","level4":"N/A","scoreL3":"0.5","scoreL4":"0.0"}
{"doi":"10.1108/00251740610715687","level1":"05 social sciences","level2":"0502 economics and business","level3":"050203 business & management","level4":"05020311 International business/International trade","scoreL3":"0.605047881603241","scoreL4":"0.2156608108845153"}
{"doi":"10.1108/00251740610715687","level1":"05 social sciences","level2":"0502 economics and business","level3":"050211 marketing","level4":"N/A","scoreL3":"0.394952118396759","scoreL4":"0.0"}
{"doi":"10.1080/03067310500248098","level1":"04 agricultural and veterinary sciences","level2":"0404 agricultural biotechnology","level3":"040401 food science","level4":"04040102 Food science/Food industry","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.1080/03067310500248098","level1":"04 agricultural and veterinary sciences","level2":"0405 other agricultural sciences","level3":"040502 food science","level4":"04050202 Food science/Food industry","scoreL3":"0.5","scoreL4":"0.5"}
{"doi":"10.3152/147154306781778533","level1":"05 social sciences","level2":"0502 economics and business","level3":"050203 business & management","level4":"05020307 Innovation/Product management","scoreL3":"0.5293408632278442","scoreL4":"0.5326762795448303"}
{"doi":"10.3152/147154306781778533","level1":"05 social sciences","level2":"0509 other social sciences","level3":"050905 science studies","level4":"05090502 Social philosophy/Capitalism","scoreL3":"0.4706590473651886","scoreL4":"0.4673237204551697"}
{"doi":"10.1785/0120050806","level1":"02 engineering and technology","level2":"0211 other engineering and technologies","level3":"021101 geological & geomatics engineering","level4":"02110103 Concrete/Building materials","scoreL3":"0.5343400835990906","scoreL4":"0.3285667930180677"}

View File

@ -24,13 +24,13 @@ function copydb() {
# drop tables from db # drop tables from db
for i in `impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`; for i in `impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do do
`impala-shell -i impala-cluster-dn1.openaire.eu -d -d ${db} -q "drop table $i;"`; `impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "drop table $i;"`;
done done
# drop views from db # drop views from db
for i in `impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`; for i in `impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do do
`impala-shell -i impala-cluster-dn1.openaire.eu -d -d ${db} -q "drop view $i;"`; `impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "drop view $i;"`;
done done
# delete the database # delete the database
@ -82,12 +82,12 @@ copydb $USAGE_STATS_DB
copydb $PROD_USAGE_STATS_DB copydb $PROD_USAGE_STATS_DB
copydb $EXT_DB copydb $EXT_DB
copydb $STATS_DB copydb $STATS_DB
#copydb $MONITOR_DB copydb $MONITOR_DB
copydb $OBSERVATORY_DB copydb $OBSERVATORY_DB
copydb $MONITOR_DB'_funded' copydb $MONITOR_DB'_funded'
copydb $MONITOR_DB'_institutions' copydb $MONITOR_DB'_institutions'
copydb $MONITOR_DB'_RIs_tail' copydb $MONITOR_DB'_ris_tail'
contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other" contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other"
for i in ${contexts} for i in ${contexts}

View File

@ -13,7 +13,7 @@ function createShadowDB() {
# drop views from db # drop views from db
for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} --delimited -q "show tables"`; for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} --delimited -q "show tables"`;
do do
`impala-shell -i impala-cluster-dn1.openaire.eu -d -d ${SHADOW} -q "drop view $i;"`; `impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} -q "drop view $i;"`;
done done
impala-shell -i impala-cluster-dn1.openaire.eu -q "drop database ${SHADOW} CASCADE"; impala-shell -i impala-cluster-dn1.openaire.eu -q "drop database ${SHADOW} CASCADE";
@ -36,13 +36,13 @@ createShadowDB $MONITOR_DB $MONITOR_DB_SHADOW
createShadowDB $OBSERVATORY_DB $OBSERVATORY_DB_SHADOW createShadowDB $OBSERVATORY_DB $OBSERVATORY_DB_SHADOW
createShadowDB USAGE_STATS_DB USAGE_STATS_DB_SHADOW createShadowDB USAGE_STATS_DB USAGE_STATS_DB_SHADOW
createShadowDB $MONITOR_DB'_funded' $MONITOR_DB'_funded_shadow' createShadowDB $MONITOR_DB'_funded' $MONITOR_DB_SHADOW'_shadow_funded'
createShadowDB $MONITOR_DB'_institutions' $MONITOR_DB'_institutions_shadow' createShadowDB $MONITOR_DB'_institutions' $MONITOR_DB_SHADOW'_shadow_institutions'
createShadowDB $MONITOR_DB'_RIs_tail' $MONITOR_DB'_RIs_tail_shadow' createShadowDB $MONITOR_DB'_ris_tail' $MONITOR_DB_SHADOW'_shadow_ris_tail'
contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other" contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other"
for i in ${contexts} for i in ${contexts}
do do
tmp=`echo "$i" | sed 's/'-'/'_'/g' | sed 's/'::'/'_'/g'` tmp=`echo "$i" | sed 's/'-'/'_'/g' | sed 's/'::'/'_'/g'`
createShadowDB ${MONITOR_DB}'_'${tmp} ${MONITOR_DB}'_'${tmp}'_shadow' createShadowDB ${MONITOR_DB}'_'${tmp} ${MONITOR_DB_SHADOW}'_shadow_'${tmp}
done done

View File

@ -14,6 +14,7 @@ export SCRIPT_PATH2=$5
export SCRIPT_PATH3=$6 export SCRIPT_PATH3=$6
export SCRIPT_PATH4=$7 export SCRIPT_PATH4=$7
export SCRIPT_PATH5=$8 export SCRIPT_PATH5=$8
export SCRIPT_PATH6=$9
export HIVE_OPTS="-hiveconf mapred.job.queue.name=analytics -hiveconf hive.spark.client.connect.timeout=120000ms -hiveconf hive.spark.client.server.connect.timeout=300000ms -hiveconf spark.executor.memory=19166291558 -hiveconf spark.yarn.executor.memoryOverhead=3225 -hiveconf spark.driver.memory=11596411699 -hiveconf spark.yarn.driver.memoryOverhead=1228" export HIVE_OPTS="-hiveconf mapred.job.queue.name=analytics -hiveconf hive.spark.client.connect.timeout=120000ms -hiveconf hive.spark.client.server.connect.timeout=300000ms -hiveconf spark.executor.memory=19166291558 -hiveconf spark.yarn.executor.memoryOverhead=3225 -hiveconf spark.driver.memory=11596411699 -hiveconf spark.yarn.driver.memoryOverhead=1228"
export HADOOP_USER_NAME="oozie" export HADOOP_USER_NAME="oozie"
@ -33,12 +34,19 @@ hdfs dfs -copyToLocal $7
echo "Getting file from " $8 echo "Getting file from " $8
hdfs dfs -copyToLocal $8 hdfs dfs -copyToLocal $8
echo "Getting file from " $9
hdfs dfs -copyToLocal $9
echo "Creating monitor database" echo "Creating monitor database"
cat step20-createMonitorDBAll.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2/g1" > foo
hive $HIVE_OPTS -f foo
cat step20-createMonitorDB_funded.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_funded/g1" > foo cat step20-createMonitorDB_funded.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_funded/g1" > foo
hive $HIVE_OPTS -f foo hive $HIVE_OPTS -f foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_funded/g1" > foo cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_funded/g1" > foo
hive $HIVE_OPTS -f foo hive $HIVE_OPTS -f foo
#
cat step20-createMonitorDB_institutions.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo cat step20-createMonitorDB_institutions.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo
hive $HIVE_OPTS -f foo hive $HIVE_OPTS -f foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo
@ -56,14 +64,20 @@ do
hive $HIVE_OPTS -f foo hive $HIVE_OPTS -f foo
done done
cat step20-createMonitorDB_RIs_tail.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_ris_tail/g1" | sed "s/CONTEXTS/\"'knowmad::other','dh-ch::other', 'enermaps::other', 'gotriple::other', 'neanias-atmospheric::other', 'rural-digital-europe::other', 'covid-19::other', 'aurora::other', 'neanias-space::other', 'north-america-studies::other', 'north-american-studies::other', 'eutopia::other'\"/g" > foo
cat step20-createMonitorDB_RIs_tail.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_RIs_tail/g1" | sed "s/CONTEXTS/\"'knowmad::other','dh-ch::other', 'enermaps::other', 'gotriple::other', 'neanias-atmospheric::other', 'rural-digital-europe::other', 'covid-19::other', 'aurora::other', 'neanias-space::other', 'north-america-studies::other', 'north-american-studies::other', 'eutopia::other'\"/g" > foo
hive $HIVE_OPTS -f foo hive $HIVE_OPTS -f foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_RIs_tail/g1" > foo cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_ris_tail/g1" > foo
hive $HIVE_OPTS -f foo hive $HIVE_OPTS -f foo
echo "Hive shell finished" echo "Hive shell finished"
echo "Updating shadow monitor all database"
hive -e "drop database if exists ${SHADOW} cascade"
hive -e "create database if not exists ${SHADOW}"
hive $HIVE_OPTS --database ${2} -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${2}.\1;/" > foo
hive -f foo
echo "Updated shadow monitor all database"
echo "Updating shadow monitor funded database" echo "Updating shadow monitor funded database"
hive -e "drop database if exists ${SHADOW}_funded cascade" hive -e "drop database if exists ${SHADOW}_funded cascade"
hive -e "create database if not exists ${SHADOW}_funded" hive -e "create database if not exists ${SHADOW}_funded"

View File

@ -37,8 +37,15 @@ select * from ${stats_db_name}.otherresearchproduct_refereed;
create table if not exists ${stats_db_name}.indi_impact_measures STORED AS PARQUET as create table if not exists ${stats_db_name}.indi_impact_measures STORED AS PARQUET as
select substr(id, 4) as id, measures_ids.id impactmetric, cast(measures_ids.unit.value[0] as double) score, select substr(id, 4) as id, measures_ids.id impactmetric, cast(measures_ids.unit.value[0] as double) score,
cast(measures_ids.unit.value[0] as decimal(6,3)) score_dec, measures_ids.unit.value[1] class cast(measures_ids.unit.value[0] as decimal(6,3)) score_dec, measures_ids.unit.value[1] impact_class
from ${openaire_db_name}.result lateral view explode(measures) measures as measures_ids from ${openaire_db_name}.result lateral view explode(measures) measures as measures_ids
where measures_ids.id!='views' and measures_ids.id!='downloads'; where measures_ids.id!='views' and measures_ids.id!='downloads';
ANALYZE TABLE indi_impact_measures COMPUTE STATISTICS; create table if not exists ${stats_db_name}.result_apc_affiliations STORED AS PARQUET as
select distinct substr(rel.target,4) id, substr(rel.source,4) organization, o.legalname.value name,
cast(rel.properties[0].value as double) apc_amount,
rel.properties[1].value apc_currency
from ${openaire_db_name}.relation rel
join ${openaire_db_name}.organization o on o.id=rel.source
join ${openaire_db_name}.result r on r.id=rel.target
where rel.subreltype = 'affiliation' and rel.datainfo.deletedbyinference = false and size(rel.properties)>0;

View File

@ -35,6 +35,7 @@ create or replace view ${stats_db_name}.doctoratestudents as select * from stats
create or replace view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers; create or replace view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers;
create or replace view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft; create or replace view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft;
create or replace view ${stats_db_name}.hrrst as select * from stats_ext.hrrst; create or replace view ${stats_db_name}.hrrst as select * from stats_ext.hrrst;
create or replace view ${stats_db_name}.graduatedoctorates as select * from stats_ext.graduatedoctorates;
create table if not exists ${stats_db_name}.result_instance stored as parquet as create table if not exists ${stats_db_name}.result_instance stored as parquet as
select distinct r.* select distinct r.*

View File

@ -88,85 +88,88 @@ create view if not exists TARGET.doctoratestudents as select * from SOURCE.docto
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers; create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft; create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst; create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
create view if not exists TARGET.graduatedoctorates as select * from SOURCE.graduatedoctorates;
create table TARGET.result_citations stored as parquet as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_citations stored as parquet as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
create table TARGET.result_references_oc stored as parquet as select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_references_oc stored as parquet as select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
create table TARGET.result_citations_oc stored as parquet as select * from SOURCE.result_citations_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_citations_oc stored as parquet as select * from SOURCE.result_citations_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_citations_oc COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_citations_oc COMPUTE STATISTICS;
create table TARGET.result_classifications stored as parquet as select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_classifications stored as parquet as select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
create table TARGET.result_apc stored as parquet as select * from SOURCE.result_apc orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_apc stored as parquet as select * from SOURCE.result_apc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
create table TARGET.result_concepts stored as parquet as select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_concepts stored as parquet as select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
create table TARGET.result_datasources stored as parquet as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_datasources stored as parquet as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
create table TARGET.result_fundercount stored as parquet as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_fundercount stored as parquet as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
create table TARGET.result_gold stored as parquet as select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_gold stored as parquet as select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
create table TARGET.result_greenoa stored as parquet as select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_greenoa stored as parquet as select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
create table TARGET.result_languages stored as parquet as select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_languages stored as parquet as select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
create table TARGET.result_licenses stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_licenses stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
create table TARGET.licenses_normalized STORED AS PARQUET as select * from SOURCE.licenses_normalized; create table TARGET.licenses_normalized STORED AS PARQUET as select * from SOURCE.licenses_normalized;
ANALYZE TABLE TARGET.licenses_normalized COMPUTE STATISTICS; --ANALYZE TABLE TARGET.licenses_normalized COMPUTE STATISTICS;
create table TARGET.result_oids stored as parquet as select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_oids stored as parquet as select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
create table TARGET.result_organization stored as parquet as select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_organization stored as parquet as select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
create table TARGET.result_peerreviewed stored as parquet as select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_peerreviewed stored as parquet as select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
create table TARGET.result_pids stored as parquet as select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_pids stored as parquet as select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
create table TARGET.result_projectcount stored as parquet as select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_projectcount stored as parquet as select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
create table TARGET.result_projects stored as parquet as select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_projects stored as parquet as select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
create table TARGET.result_refereed stored as parquet as select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_refereed stored as parquet as select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
create table TARGET.result_sources stored as parquet as select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_sources stored as parquet as select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
create table TARGET.result_topics stored as parquet as select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_topics stored as parquet as select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
create table TARGET.result_fos stored as parquet as select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_fos stored as parquet as select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
create table TARGET.result_accessroute stored as parquet as select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_accessroute stored as parquet as select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
create table TARGET.result_orcid stored as parquet as select * from SOURCE.result_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create view TARGET.foo1 as select * from SOURCE.result_result rr where rr.source in (select id from TARGET.result); create view TARGET.foo1 as select * from SOURCE.result_result rr where rr.source in (select id from TARGET.result);
create view TARGET.foo2 as select * from SOURCE.result_result rr where rr.target in (select id from TARGET.result); create view TARGET.foo2 as select * from SOURCE.result_result rr where rr.target in (select id from TARGET.result);
create table TARGET.result_result STORED AS PARQUET as select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou; create table TARGET.result_result STORED AS PARQUET as select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou;
drop view TARGET.foo1; drop view TARGET.foo1;
drop view TARGET.foo2; drop view TARGET.foo2;
ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
-- datasources -- datasources
create view if not exists TARGET.datasource as select * from SOURCE.datasource; create view if not exists TARGET.datasource as select * from SOURCE.datasource;
@ -175,7 +178,7 @@ create view if not exists TARGET.datasource_organizations as select * from SOURC
create view if not exists TARGET.datasource_sources as select * from SOURCE.datasource_sources; create view if not exists TARGET.datasource_sources as select * from SOURCE.datasource_sources;
create table TARGET.datasource_results stored as parquet as select id as result, datasource as id from TARGET.result_datasources; create table TARGET.datasource_results stored as parquet as select id as result, datasource as id from TARGET.result_datasources;
ANALYZE TABLE TARGET.datasource_results COMPUTE STATISTICS; --ANALYZE TABLE TARGET.datasource_results COMPUTE STATISTICS;
-- organizations -- organizations
create view if not exists TARGET.organization as select * from SOURCE.organization; create view if not exists TARGET.organization as select * from SOURCE.organization;
@ -193,28 +196,28 @@ create view if not exists TARGET.project_classification as select * from SOURCE.
create view if not exists TARGET.project_organization_contribution as select * from SOURCE.project_organization_contribution; create view if not exists TARGET.project_organization_contribution as select * from SOURCE.project_organization_contribution;
create table TARGET.project_results stored as parquet as select id as result, project as id from TARGET.result_projects; create table TARGET.project_results stored as parquet as select id as result, project as id from TARGET.result_projects;
ANALYZE TABLE TARGET.project_results COMPUTE STATISTICS; --ANALYZE TABLE TARGET.project_results COMPUTE STATISTICS;
-- indicators -- indicators
-- Sprint 1 ---- -- Sprint 1 ----
create table TARGET.indi_pub_green_oa stored as parquet as select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_green_oa stored as parquet as select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_grey_lit stored as parquet as select * from SOURCE.indi_pub_grey_lit orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_grey_lit stored as parquet as select * from SOURCE.indi_pub_grey_lit orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS;
create table TARGET.indi_pub_doi_from_crossref stored as parquet as select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_doi_from_crossref stored as parquet as select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
-- Sprint 2 ---- -- Sprint 2 ----
create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
create table TARGET.indi_result_has_cc_licence_url stored as parquet as select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_result_has_cc_licence_url stored as parquet as select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
create table TARGET.indi_result_with_orcid stored as parquet as select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_result_with_orcid stored as parquet as select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
---- Sprint 3 ---- ---- Sprint 3 ----
create table TARGET.indi_funded_result_with_fundref stored as parquet as select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_funded_result_with_fundref stored as parquet as select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
create view TARGET.indi_result_org_collab as select * from SOURCE.indi_result_org_collab; create view TARGET.indi_result_org_collab as select * from SOURCE.indi_result_org_collab;
create view TARGET.indi_result_org_country_collab as select * from SOURCE.indi_result_org_country_collab; create view TARGET.indi_result_org_country_collab as select * from SOURCE.indi_result_org_country_collab;
create view TARGET.indi_project_collab_org as select * from SOURCE.indi_project_collab_org; create view TARGET.indi_project_collab_org as select * from SOURCE.indi_project_collab_org;
@ -223,32 +226,32 @@ create view TARGET.indi_funder_country_collab as select * from SOURCE.indi_funde
create view TARGET.indi_result_country_collab as select * from SOURCE.indi_result_country_collab; create view TARGET.indi_result_country_collab as select * from SOURCE.indi_result_country_collab;
---- Sprint 4 ---- ---- Sprint 4 ----
create table TARGET.indi_pub_diamond stored as parquet as select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_diamond stored as parquet as select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
create table TARGET.indi_pub_in_transformative stored as parquet as select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_in_transformative stored as parquet as select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
create table TARGET.indi_pub_closed_other_open stored as parquet as select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_closed_other_open stored as parquet as select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
---- Sprint 5 ---- ---- Sprint 5 ----
create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
---- Sprint 6 ---- ---- Sprint 6 ----
create table TARGET.indi_pub_hybrid_oa_with_cc stored as parquet as select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_hybrid_oa_with_cc stored as parquet as select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
create table TARGET.indi_pub_bronze_oa stored as parquet as select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_bronze_oa stored as parquet as select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads stored as parquet as select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result r where r.id=orig.result_id); create table TARGET.indi_pub_downloads stored as parquet as select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource stored as parquet as select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result r where r.id=orig.result_id); create table TARGET.indi_pub_downloads_datasource stored as parquet as select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_year stored as parquet as select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id); create table TARGET.indi_pub_downloads_year stored as parquet as select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource_year stored as parquet as select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id); create table TARGET.indi_pub_downloads_datasource_year stored as parquet as select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
---- Sprint 7 ---- ---- Sprint 7 ----
create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_hybrid stored as parquet as select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_hybrid stored as parquet as select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
create view TARGET.indi_org_fairness as select * from SOURCE.indi_org_fairness; create view TARGET.indi_org_fairness as select * from SOURCE.indi_org_fairness;
create view TARGET.indi_org_fairness_pub_pr as select * from SOURCE.indi_org_fairness_pub_pr; create view TARGET.indi_org_fairness_pub_pr as select * from SOURCE.indi_org_fairness_pub_pr;
create view TARGET.indi_org_fairness_pub_year as select * from SOURCE.indi_org_fairness_pub_year; create view TARGET.indi_org_fairness_pub_year as select * from SOURCE.indi_org_fairness_pub_year;
@ -259,12 +262,22 @@ create view TARGET.indi_org_findable as select * from SOURCE.indi_org_findable;
create view TARGET.indi_org_openess as select * from SOURCE.indi_org_openess; create view TARGET.indi_org_openess as select * from SOURCE.indi_org_openess;
create view TARGET.indi_org_openess_year as select * from SOURCE.indi_org_openess_year; create view TARGET.indi_org_openess_year as select * from SOURCE.indi_org_openess_year;
create table TARGET.indi_pub_has_preprint stored as parquet as select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_has_preprint stored as parquet as select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
create table TARGET.indi_pub_in_subscribed stored as parquet as select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_in_subscribed stored as parquet as select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
create table TARGET.indi_result_with_pid stored as parquet as select * from SOURCE.indi_result_with_pid orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_result_with_pid stored as parquet as select * from SOURCE.indi_result_with_pid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS;
create table TARGET.indi_impact_measures stored as parquet as select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_impact_measures stored as parquet as select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
create table TARGET.indi_pub_interdisciplinarity stored as parquet as select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.indi_pub_interdisciplinarity stored as parquet as select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS; --ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
create table TARGET.result_apc_affiliations stored as parquet as select * from SOURCE.result_apc_affiliations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_apc_affiliations COMPUTE STATISTICS;
create table TARGET.indi_is_project_result_after stored as parquet as select * from SOURCE.indi_is_project_result_after orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
create table TARGET.indi_is_funder_plan_s stored as parquet as select * from SOURCE.indi_is_funder_plan_s orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create view TARGET.indi_funder_fairness as select * from SOURCE.indi_funder_fairness;
create view TARGET.indi_funder_openess as select * from SOURCE.indi_funder_openess;
create view TARGET.indi_funder_findable as select * from SOURCE.indi_funder_findable;
create view TARGET.indi_ris_fairness as select * from SOURCE.indi_ris_fairness;
create view TARGET.indi_ris_openess as select * from SOURCE.indi_ris_openess;
create view TARGET.indi_ris_findable as select * from SOURCE.indi_ris_findable;

View File

@ -0,0 +1,292 @@
drop database if exists TARGET cascade;
create database if not exists TARGET;
create view if not exists TARGET.category as select * from SOURCE.category;
create view if not exists TARGET.concept as select * from SOURCE.concept;
create view if not exists TARGET.context as select * from SOURCE.context;
create view if not exists TARGET.country as select * from SOURCE.country;
create view if not exists TARGET.countrygdp as select * from SOURCE.countrygdp;
create view if not exists TARGET.creation_date as select * from SOURCE.creation_date;
create view if not exists TARGET.funder as select * from SOURCE.funder;
create view if not exists TARGET.fundref as select * from SOURCE.fundref;
create view if not exists TARGET.rndexpenditure as select * from SOURCE.rndexpediture;
create view if not exists TARGET.rndgdpexpenditure as select * from SOURCE.rndgdpexpenditure;
create view if not exists TARGET.doctoratestudents as select * from SOURCE.doctoratestudents;
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
create view if not exists TARGET.graduatedoctorates as select * from SOURCE.graduatedoctorates;
create table TARGET.result stored as parquet as
select distinct * from (
select * from SOURCE.result r where exists (select 1 from SOURCE.result_projects rp join SOURCE.project p on rp.project=p.id where rp.id=r.id)
union all
select * from SOURCE.result r where exists (select 1 from SOURCE.result_concepts rc where rc.id=r.id)
union all
select * from SOURCE.result r where exists (select 1 from SOURCE.result_organization ro where ro.id=r.id and ro.organization in (
'openorgs____::b84450f9864182c67b8611b5593f4250', --"Athena Research and Innovation Center In Information Communication & Knowledge Technologies', --ARC"
'openorgs____::d41cf6bd4ab1b1362a44397e0b95c975', --National Research Council
'openorgs____::d2a09b9d5eabb10c95f9470e172d05d2', --??? Not exists ??
'openorgs____::d169c7407dd417152596908d48c11460', --Masaryk University
'openorgs____::1ec924b1759bb16d0a02f2dad8689b21', --University of Belgrade
'openorgs____::0ae431b820e4c33db8967fbb2b919150', --University of Helsinki
'openorgs____::759d59f05d77188faee99b7493b46805', --University of Minho
'openorgs____::cad284878801b9465fa51a95b1d779db', --Universidad Politécnica de Madrid
'openorgs____::eadc8da90a546e98c03f896661a2e4d4', --University of Göttingen
'openorgs____::c0286313e36479eff8676dba9b724b40', --National and Kapodistrian University of Athens
-- 'openorgs____::c80a8243a5e5c620d7931c88d93bf17a', --Université Paris Diderot
'openorgs____::c08634f0a6b0081c3dc6e6c93a4314f3', --Bielefeld University
'openorgs____::6fc85e4a8f7ecaf4b0c738d010e967ea', --University of Southern Denmark
'openorgs____::3d6122f87f9a97a99d8f6e3d73313720', --Humboldt-Universität zu Berlin
'openorgs____::16720ada63d0fa8ca41601feae7d1aa5', --TU Darmstadt
'openorgs____::ccc0a066b56d2cfaf90c2ae369df16f5', --KU Leuven
'openorgs____::4c6f119632adf789746f0a057ed73e90', --University of the Western Cape
'openorgs____::ec3665affa01aeafa28b7852c4176dbd', --Rudjer Boskovic Institute
'openorgs____::5f31346d444a7f06a28c880fb170b0f6', --Ghent University
'openorgs____::2dbe47117fd5409f9c61620813456632', --University of Luxembourg
'openorgs____::6445d7758d3a40c4d997953b6632a368', --National Institute of Informatics (NII)
'openorgs____::b77c01aa15de3675da34277d48de2ec1', -- Valencia Catholic University Saint Vincent Martyr
'openorgs____::7fe2f66cdc43983c6b24816bfe9cf6a0', -- Unviersity of Warsaw
'openorgs____::15e7921fc50d9aa1229a82a84429419e', -- University Of Thessaly
'openorgs____::11f7919dadc8f8a7251af54bba60c956', -- Technical University of Crete
'openorgs____::84f0c5f5dbb6daf42748485924efde4b', -- University of Piraeus
'openorgs____::4ac562f0376fce3539504567649cb373', -- University of Patras
'openorgs____::3e8d1f8c3f6cd7f418b09f1f58b4873b', -- Aristotle University of Thessaloniki
'openorgs____::3fcef6e1c469c10f2a84b281372c9814', -- World Bank
'openorgs____::1698a2eb1885ef8adb5a4a969e745ad3', -- École des Ponts ParisTech
'openorgs____::e15adb13c4dadd49de4d35c39b5da93a', -- Nanyang Technological University
'openorgs____::4b34103bde246228fcd837f5f1bf4212', -- Autonomous University of Barcelona
'openorgs____::72ec75fcfc4e0df1a76dc4c49007fceb', -- McMaster University
'openorgs____::51c7fc556e46381734a25a6fbc3fd398', -- University of Modena and Reggio Emilia
'openorgs____::235d7f9ad18ecd7e6dc62ea4990cb9db', -- Bilkent University
'openorgs____::31f2fa9e05b49d4cf40a19c3fed8eb06', -- Saints Cyril and Methodius University of Skopje
'openorgs____::db7686f30f22cbe73a4fde872ce812a6', -- University of Milan
'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork
'openorgs____::38d7097854736583dde879d12dacafca', -- Brown University
'openorgs____::57784c9e047e826fefdb1ef816120d92', --Arts et Métiers ParisTech
'openorgs____::2530baca8a15936ba2e3297f2bce2e7e', -- University of Cape Town
'openorgs____::d11f981828c485cd23d93f7f24f24db1', -- Technological University Dublin
'openorgs____::5e6bf8962665cdd040341171e5c631d8', -- Delft University of Technology
'openorgs____::846cb428d3f52a445f7275561a7beb5d', -- University of Manitoba
'openorgs____::eb391317ed0dc684aa81ac16265de041', -- Universitat Rovira i Virgili
'openorgs____::66aa9fc2fceb271423dfabcc38752dc0', -- Lund University
'openorgs____::3cff625a4370d51e08624cc586138b2f', -- IMT Atlantique
'openorgs____::c0b262bd6eab819e4c994914f9c010e2', -- National Institute of Geophysics and Volcanology
'openorgs____::1624ff7c01bb641b91f4518539a0c28a', -- Vrije Universiteit Amsterdam
'openorgs____::4d4051b56708688235252f1d8fddb8c1', --Iscte - Instituto Universitário de Lisboa
'openorgs____::ab4ac74c35fa5dada770cf08e5110fab', -- Universidade Católica Portuguesa
'openorgs____::4d4051b56708688235252f1d8fddb8c1', -- Iscte - Instituto Universitário de Lisboa
'openorgs____::5d55fb216b14691cf68218daf5d78cd9', -- Munster Technological University
'openorgs____::0fccc7640f0cb44d5cd1b06b312a06b9', -- Cardiff University
'openorgs____::8839b55dae0c84d56fd533f52d5d483a' -- Leibniz Institute of Ecological Urban and Regional Development
) )) foo;
--ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
create view if not exists TARGET.category as select * from SOURCE.category;
create view if not exists TARGET.concept as select * from SOURCE.concept;
create view if not exists TARGET.context as select * from SOURCE.context;
create view if not exists TARGET.country as select * from SOURCE.country;
create view if not exists TARGET.countrygdp as select * from SOURCE.countrygdp;
create view if not exists TARGET.creation_date as select * from SOURCE.creation_date;
create view if not exists TARGET.funder as select * from SOURCE.funder;
create view if not exists TARGET.fundref as select * from SOURCE.fundref;
create view if not exists TARGET.rndexpenditure as select * from SOURCE.rndexpediture;
create view if not exists TARGET.rndgdpexpenditure as select * from SOURCE.rndgdpexpenditure;
create view if not exists TARGET.doctoratestudents as select * from SOURCE.doctoratestudents;
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
--create view if not exists TARGET.graduatedoctorates as select * from SOURCE.graduatedoctorates;
create table TARGET.result_citations stored as parquet as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
create table TARGET.result_references_oc stored as parquet as select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
create table TARGET.result_citations_oc stored as parquet as select * from SOURCE.result_citations_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_citations_oc COMPUTE STATISTICS;
create table TARGET.result_classifications stored as parquet as select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
create table TARGET.result_apc stored as parquet as select * from SOURCE.result_apc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
create table TARGET.result_concepts stored as parquet as select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
create table TARGET.result_datasources stored as parquet as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
create table TARGET.result_fundercount stored as parquet as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
create table TARGET.result_gold stored as parquet as select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
create table TARGET.result_greenoa stored as parquet as select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
create table TARGET.result_languages stored as parquet as select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
create table TARGET.result_licenses stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
create table TARGET.licenses_normalized STORED AS PARQUET as select * from SOURCE.licenses_normalized;
--ANALYZE TABLE TARGET.licenses_normalized COMPUTE STATISTICS;
create table TARGET.result_oids stored as parquet as select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
create table TARGET.result_organization stored as parquet as select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
create table TARGET.result_peerreviewed stored as parquet as select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
create table TARGET.result_pids stored as parquet as select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
create table TARGET.result_projectcount stored as parquet as select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
create table TARGET.result_projects stored as parquet as select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
create table TARGET.result_refereed stored as parquet as select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
create table TARGET.result_sources stored as parquet as select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
create table TARGET.result_topics stored as parquet as select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
create table TARGET.result_fos stored as parquet as select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
create table TARGET.result_accessroute stored as parquet as select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
create view TARGET.foo1 as select * from SOURCE.result_result rr where rr.source in (select id from TARGET.result);
create view TARGET.foo2 as select * from SOURCE.result_result rr where rr.target in (select id from TARGET.result);
create table TARGET.result_result STORED AS PARQUET as select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou;
drop view TARGET.foo1;
drop view TARGET.foo2;
--ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
-- datasources
create view if not exists TARGET.datasource as select * from SOURCE.datasource;
create view if not exists TARGET.datasource_oids as select * from SOURCE.datasource_oids;
create view if not exists TARGET.datasource_organizations as select * from SOURCE.datasource_organizations;
create view if not exists TARGET.datasource_sources as select * from SOURCE.datasource_sources;
create table TARGET.datasource_results stored as parquet as select id as result, datasource as id from TARGET.result_datasources;
--ANALYZE TABLE TARGET.datasource_results COMPUTE STATISTICS;
-- organizations
create view if not exists TARGET.organization as select * from SOURCE.organization;
create view if not exists TARGET.organization_datasources as select * from SOURCE.organization_datasources;
create view if not exists TARGET.organization_pids as select * from SOURCE.organization_pids;
create view if not exists TARGET.organization_projects as select * from SOURCE.organization_projects;
create view if not exists TARGET.organization_sources as select * from SOURCE.organization_sources;
-- projects
create view if not exists TARGET.project as select * from SOURCE.project;
create view if not exists TARGET.project_oids as select * from SOURCE.project_oids;
create view if not exists TARGET.project_organizations as select * from SOURCE.project_organizations;
create view if not exists TARGET.project_resultcount as select * from SOURCE.project_resultcount;
create view if not exists TARGET.project_classification as select * from SOURCE.project_classification;
create view if not exists TARGET.project_organization_contribution as select * from SOURCE.project_organization_contribution;
create table TARGET.project_results stored as parquet as select id as result, project as id from TARGET.result_projects;
--ANALYZE TABLE TARGET.project_results COMPUTE STATISTICS;
-- indicators
-- Sprint 1 ----
create table TARGET.indi_pub_green_oa stored as parquet as select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_grey_lit stored as parquet as select * from SOURCE.indi_pub_grey_lit orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS;
create table TARGET.indi_pub_doi_from_crossref stored as parquet as select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
-- Sprint 2 ----
create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
create table TARGET.indi_result_has_cc_licence_url stored as parquet as select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
create table TARGET.indi_result_with_orcid stored as parquet as select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
---- Sprint 3 ----
create table TARGET.indi_funded_result_with_fundref stored as parquet as select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
create view TARGET.indi_result_org_collab as select * from SOURCE.indi_result_org_collab;
create view TARGET.indi_result_org_country_collab as select * from SOURCE.indi_result_org_country_collab;
create view TARGET.indi_project_collab_org as select * from SOURCE.indi_project_collab_org;
create view TARGET.indi_project_collab_org_country as select * from SOURCE.indi_project_collab_org_country;
create view TARGET.indi_funder_country_collab as select * from SOURCE.indi_funder_country_collab;
create view TARGET.indi_result_country_collab as select * from SOURCE.indi_result_country_collab;
---- Sprint 4 ----
create table TARGET.indi_pub_diamond stored as parquet as select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
create table TARGET.indi_pub_in_transformative stored as parquet as select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
create table TARGET.indi_pub_closed_other_open stored as parquet as select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
---- Sprint 5 ----
create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
---- Sprint 6 ----
create table TARGET.indi_pub_hybrid_oa_with_cc stored as parquet as select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
create table TARGET.indi_pub_bronze_oa stored as parquet as select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads stored as parquet as select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource stored as parquet as select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_year stored as parquet as select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource_year stored as parquet as select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
---- Sprint 7 ----
create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_hybrid stored as parquet as select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
create view TARGET.indi_org_fairness as select * from SOURCE.indi_org_fairness;
create view TARGET.indi_org_fairness_pub_pr as select * from SOURCE.indi_org_fairness_pub_pr;
create view TARGET.indi_org_fairness_pub_year as select * from SOURCE.indi_org_fairness_pub_year;
create view TARGET.indi_org_fairness_pub as select * from SOURCE.indi_org_fairness_pub;
create view TARGET.indi_org_fairness_year as select * from SOURCE.indi_org_fairness_year;
create view TARGET.indi_org_findable_year as select * from SOURCE.indi_org_findable_year;
create view TARGET.indi_org_findable as select * from SOURCE.indi_org_findable;
create view TARGET.indi_org_openess as select * from SOURCE.indi_org_openess;
create view TARGET.indi_org_openess_year as select * from SOURCE.indi_org_openess_year;
create table TARGET.indi_pub_has_preprint stored as parquet as select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
create table TARGET.indi_pub_in_subscribed stored as parquet as select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
create table TARGET.indi_result_with_pid stored as parquet as select * from SOURCE.indi_result_with_pid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS;
create table TARGET.indi_impact_measures stored as parquet as select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
create table TARGET.indi_pub_interdisciplinarity stored as parquet as select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
create table TARGET.result_apc_affiliations stored as parquet as select * from SOURCE.result_apc_affiliations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_apc_affiliations COMPUTE STATISTICS;
create table TARGET.indi_is_project_result_after stored as parquet as select * from SOURCE.indi_is_project_result_after orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create table TARGET.indi_is_funder_plan_s stored as parquet as select * from SOURCE.indi_is_funder_plan_s orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create view TARGET.indi_funder_fairness as select * from SOURCE.indi_funder_fairness;
create view TARGET.indi_funder_openess as select * from SOURCE.indi_funder_openess;
create view TARGET.indi_funder_findable as select * from SOURCE.indi_funder_findable;
create view TARGET.indi_ris_fairness as select * from SOURCE.indi_ris_fairness;
create view TARGET.indi_ris_openess as select * from SOURCE.indi_ris_openess;
create view TARGET.indi_ris_findable as select * from SOURCE.indi_ris_findable;

View File

@ -12,4 +12,4 @@ create table TARGET.result stored as parquet as
-- join SOURCE.result -- join SOURCE.result
where rc.id=r.id and conc.category like CONTEXT) where rc.id=r.id and conc.category like CONTEXT)
) foo; ) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

View File

@ -12,4 +12,4 @@ create table TARGET.result stored as parquet as
-- join SOURCE.result -- join SOURCE.result
where rc.id=r.id and conc.category not in (CONTEXTS)) where rc.id=r.id and conc.category not in (CONTEXTS))
) foo; ) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

View File

@ -6,4 +6,4 @@ create table TARGET.result stored as parquet as
select * from SOURCE.result r where exists (select 1 from SOURCE.result_projects rp join SOURCE.project p on rp.project=p.id where rp.id=r.id) select * from SOURCE.result r where exists (select 1 from SOURCE.result_projects rp join SOURCE.project p on rp.project=p.id where rp.id=r.id)
) foo; ) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

View File

@ -34,16 +34,16 @@ create table TARGET.result stored as parquet as
'openorgs____::3e8d1f8c3f6cd7f418b09f1f58b4873b', -- Aristotle University of Thessaloniki 'openorgs____::3e8d1f8c3f6cd7f418b09f1f58b4873b', -- Aristotle University of Thessaloniki
'openorgs____::3fcef6e1c469c10f2a84b281372c9814', -- World Bank 'openorgs____::3fcef6e1c469c10f2a84b281372c9814', -- World Bank
'openorgs____::1698a2eb1885ef8adb5a4a969e745ad3', -- École des Ponts ParisTech 'openorgs____::1698a2eb1885ef8adb5a4a969e745ad3', -- École des Ponts ParisTech
'openorgs____::e15adb13c4dadd49de4d35c39b5da93a', -- Nanyang Technological University 'openorgs____::e15adb13c4dadd49de4d35c39b5da93a', -- Nanyang Technological University
'openorgs____::4b34103bde246228fcd837f5f1bf4212', -- Autonomous University of Barcelona 'openorgs____::4b34103bde246228fcd837f5f1bf4212', -- Autonomous University of Barcelona
'openorgs____::72ec75fcfc4e0df1a76dc4c49007fceb', -- McMaster University 'openorgs____::72ec75fcfc4e0df1a76dc4c49007fceb', -- McMaster University
'openorgs____::51c7fc556e46381734a25a6fbc3fd398', -- University of Modena and Reggio Emilia 'openorgs____::51c7fc556e46381734a25a6fbc3fd398', -- University of Modena and Reggio Emilia
'openorgs____::235d7f9ad18ecd7e6dc62ea4990cb9db', -- Bilkent University 'openorgs____::235d7f9ad18ecd7e6dc62ea4990cb9db', -- Bilkent University
'openorgs____::31f2fa9e05b49d4cf40a19c3fed8eb06', -- Saints Cyril and Methodius University of Skopje 'openorgs____::31f2fa9e05b49d4cf40a19c3fed8eb06', -- Saints Cyril and Methodius University of Skopje
'openorgs____::db7686f30f22cbe73a4fde872ce812a6', -- University of Milan 'openorgs____::db7686f30f22cbe73a4fde872ce812a6', -- University of Milan
'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork 'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork
'openorgs____::38d7097854736583dde879d12dacafca' -- Brown University 'openorgs____::38d7097854736583dde879d12dacafca', -- Brown University
'openorgs____::57784c9e047e826fefdb1ef816120d92', --Arts et Métiers ParisTech 'openorgs____::57784c9e047e826fefdb1ef816120d92', --Arts et Métiers ParisTech
'openorgs____::2530baca8a15936ba2e3297f2bce2e7e', -- University of Cape Town 'openorgs____::2530baca8a15936ba2e3297f2bce2e7e', -- University of Cape Town
'openorgs____::d11f981828c485cd23d93f7f24f24db1', -- Technological University Dublin 'openorgs____::d11f981828c485cd23d93f7f24f24db1', -- Technological University Dublin
'openorgs____::5e6bf8962665cdd040341171e5c631d8', -- Delft University of Technology 'openorgs____::5e6bf8962665cdd040341171e5c631d8', -- Delft University of Technology
@ -52,7 +52,13 @@ create table TARGET.result stored as parquet as
'openorgs____::66aa9fc2fceb271423dfabcc38752dc0', -- Lund University 'openorgs____::66aa9fc2fceb271423dfabcc38752dc0', -- Lund University
'openorgs____::3cff625a4370d51e08624cc586138b2f', -- IMT Atlantique 'openorgs____::3cff625a4370d51e08624cc586138b2f', -- IMT Atlantique
'openorgs____::c0b262bd6eab819e4c994914f9c010e2', -- National Institute of Geophysics and Volcanology 'openorgs____::c0b262bd6eab819e4c994914f9c010e2', -- National Institute of Geophysics and Volcanology
'openorgs____::1624ff7c01bb641b91f4518539a0c28a' -- Vrije Universiteit Amsterdam 'openorgs____::1624ff7c01bb641b91f4518539a0c28a', -- Vrije Universiteit Amsterdam
'openorgs____::4d4051b56708688235252f1d8fddb8c1', --Iscte - Instituto Universitário de Lisboa
'openorgs____::ab4ac74c35fa5dada770cf08e5110fab', -- Universidade Católica Portuguesa
'openorgs____::4d4051b56708688235252f1d8fddb8c1', -- Iscte - Instituto Universitário de Lisboa
'openorgs____::5d55fb216b14691cf68218daf5d78cd9', -- Munster Technological University
'openorgs____::0fccc7640f0cb44d5cd1b06b312a06b9', -- Cardiff University
'openorgs____::8839b55dae0c84d56fd533f52d5d483a' -- Leibniz Institute of Ecological Urban and Regional Development
))) foo; ))) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS; --ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

View File

@ -8,7 +8,7 @@ from ${stats_db_name}.result r
group by rl.id group by rl.id
) rln on rln.id=r.id; ) rln on rln.id=r.id;
ANALYZE TABLE ${observatory_db_name}.result_cc_licence COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_cc_licence COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_country stored as parquet as create table ${observatory_db_name}.result_affiliated_country stored as parquet as
select select
@ -39,7 +39,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_year stored as parquet as create table ${observatory_db_name}.result_affiliated_year stored as parquet as
select select
@ -70,7 +70,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_year COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_year COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_year_country stored as parquet as create table ${observatory_db_name}.result_affiliated_year_country stored as parquet as
select select
@ -101,7 +101,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_year_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_year_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_datasource stored as parquet as create table ${observatory_db_name}.result_affiliated_datasource stored as parquet as
select select
@ -134,7 +134,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_datasource_country stored as parquet as create table ${observatory_db_name}.result_affiliated_datasource_country stored as parquet as
select select
@ -167,7 +167,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_organization stored as parquet as create table ${observatory_db_name}.result_affiliated_organization stored as parquet as
select select
@ -198,7 +198,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_organization_country stored as parquet as create table ${observatory_db_name}.result_affiliated_organization_country stored as parquet as
select select
@ -229,7 +229,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_funder stored as parquet as create table ${observatory_db_name}.result_affiliated_funder stored as parquet as
select select
@ -262,7 +262,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_funder_country stored as parquet as create table ${observatory_db_name}.result_affiliated_funder_country stored as parquet as
select select
@ -295,7 +295,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_country stored as parquet as create table ${observatory_db_name}.result_deposited_country stored as parquet as
select select
@ -328,7 +328,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_year stored as parquet as create table ${observatory_db_name}.result_deposited_year stored as parquet as
select select
@ -361,7 +361,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year;
ANALYZE TABLE ${observatory_db_name}.result_deposited_year COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_year COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_year_country stored as parquet as create table ${observatory_db_name}.result_deposited_year_country stored as parquet as
select select
@ -394,7 +394,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_year_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_year_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_datasource stored as parquet as create table ${observatory_db_name}.result_deposited_datasource stored as parquet as
select select
@ -427,7 +427,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_datasource_country stored as parquet as create table ${observatory_db_name}.result_deposited_datasource_country stored as parquet as
select select
@ -460,7 +460,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_organization stored as parquet as create table ${observatory_db_name}.result_deposited_organization stored as parquet as
select select
@ -493,7 +493,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_organization COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_organization COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_organization_country stored as parquet as create table ${observatory_db_name}.result_deposited_organization_country stored as parquet as
select select
@ -526,7 +526,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_organization_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_organization_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_funder stored as parquet as create table ${observatory_db_name}.result_deposited_funder stored as parquet as
select select
@ -561,7 +561,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder;
ANALYZE TABLE ${observatory_db_name}.result_deposited_funder COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_funder COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_funder_country stored as parquet as create table ${observatory_db_name}.result_deposited_funder_country stored as parquet as
select select
@ -596,4 +596,4 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract, case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder, c.code, c.name; cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_funder_country COMPUTE STATISTICS; --ANALYZE TABLE ${observatory_db_name}.result_deposited_funder_country COMPUTE STATISTICS;

View File

@ -317,15 +317,12 @@
</action> </action>
<action name="Step16-createIndicatorsTables"> <action name="Step16-createIndicatorsTables">
<shell xmlns="uri:oozie:shell-action:0.1"> <hive2 xmlns="uri:oozie:hive2-action:0.1">
<job-tracker>${jobTracker}</job-tracker> <jdbc-url>${hive_jdbc_url}</jdbc-url>
<name-node>${nameNode}</name-node> <script>scripts/step16-createIndicatorsTables.sql</script>
<exec>indicators.sh</exec> <param>stats_db_name=${stats_db_name}</param>
<argument>${stats_db_name}</argument> <param>external_stats_db_name=${external_stats_db_name}</param>
<argument>${external_stats_db_name}</argument> </hive2>
<argument>${wf:appPath()}/scripts/step16-createIndicatorsTables.sql</argument>
<file>indicators.sh</file>
</shell>
<ok to="Step16_1-definitions"/> <ok to="Step16_1-definitions"/>
<error to="Kill"/> <error to="Kill"/>
</action> </action>
@ -378,6 +375,7 @@
<argument>${wf:appPath()}/scripts/step20-createMonitorDB_institutions.sql</argument> <argument>${wf:appPath()}/scripts/step20-createMonitorDB_institutions.sql</argument>
<argument>${wf:appPath()}/scripts/step20-createMonitorDB_RIs.sql</argument> <argument>${wf:appPath()}/scripts/step20-createMonitorDB_RIs.sql</argument>
<argument>${wf:appPath()}/scripts/step20-createMonitorDB_RIs_tail.sql</argument> <argument>${wf:appPath()}/scripts/step20-createMonitorDB_RIs_tail.sql</argument>
<argument>${wf:appPath()}/scripts/step20-createMonitorDBAll.sql</argument>
<file>monitor.sh</file> <file>monitor.sh</file>
</shell> </shell>
<ok to="step21-createObservatoryDB-pre"/> <ok to="step21-createObservatoryDB-pre"/>