Merge pull request '[prod] Openaire Affiliation Inference' (#453) from affRoFromRawStringmain into main

Reviewed-on: D-Net/dnet-hadoop#453

dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/Constants.java

@@ -42,6 +42,9 @@ public class Constants {
 	public static final String NULL = "NULL";
 	public static final String NA = "N/A";
 
+	public static final String WEB_CRAWL_ID = "10|openaire____::fb98a192f6a055ba495ef414c330834b";
+	public static final String WEB_CRAWL_NAME = "Web Crawl";
+
 	public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
 
 	private Constants() {

dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelations.java

@@ -9,6 +9,7 @@ import java.util.List;
 
 import org.apache.commons.io.IOUtils;
 import org.apache.hadoop.io.Text;
+import org.apache.hadoop.io.compress.BZip2Codec;
 import org.apache.hadoop.io.compress.GzipCodec;
 import org.apache.hadoop.mapred.SequenceFileOutputFormat;
 import org.apache.spark.SparkConf;
@@ -40,9 +41,9 @@ public class PrepareAffiliationRelations implements Serializable {
 	private static final Logger log = LoggerFactory.getLogger(PrepareAffiliationRelations.class);
 	private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
 	private static final String ID_PREFIX = "50|doi_________::";
-	public static final String BIP_AFFILIATIONS_CLASSID = "result:organization:bipinference";
+	public static final String BIP_AFFILIATIONS_CLASSID = "result:organization:openaireinference";
-	public static final String BIP_AFFILIATIONS_CLASSNAME = "Affiliation relation inferred by BIP!";
+	public static final String BIP_AFFILIATIONS_CLASSNAME = "Affiliation relation inferred by OpenAIRE";
-	public static final String BIP_INFERENCE_PROVENANCE = "bip:affiliation:crossref";
+	public static final String BIP_INFERENCE_PROVENANCE = "openaire:affiliation";
 
 	public static <I extends Result> void main(String[] args) throws Exception {
 
@@ -70,6 +71,9 @@ public class PrepareAffiliationRelations implements Serializable {
 		final String dataciteInputPath = parser.get("dataciteInputPath");
 		log.info("dataciteInputPath: {}", dataciteInputPath);
 
+		final String webcrawlInputPath = parser.get("webCrawlInputPath");
+		log.info("webcrawlInputPath: {}", webcrawlInputPath);
+
 		final String outputPath = parser.get("outputPath");
 		log.info("outputPath: {}", outputPath);
 
@@ -101,12 +105,18 @@ public class PrepareAffiliationRelations implements Serializable {
 				JavaPairRDD<Text, Text> dataciteRelations = prepareAffiliationRelations(
 					spark, dataciteInputPath, collectedFromDatacite);
 
+				List<KeyValue> collectedFromWebCrawl = OafMapperUtils
+					.listKeyValues(Constants.WEB_CRAWL_ID, Constants.WEB_CRAWL_NAME);
+				JavaPairRDD<Text, Text> webCrawlRelations = prepareAffiliationRelations(
+					spark, webcrawlInputPath, collectedFromWebCrawl);
+
 				crossrefRelations
 					.union(pubmedRelations)
 					.union(openAPCRelations)
 					.union(dataciteRelations)
+					.union(webCrawlRelations)
 					.saveAsHadoopFile(
-						outputPath, Text.class, Text.class, SequenceFileOutputFormat.class, GzipCodec.class);
+						outputPath, Text.class, Text.class, SequenceFileOutputFormat.class, BZip2Codec.class);
 
 			});
 	}

dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/webcrawl/CreateActionSetFromWebEntries.java

@@ -21,6 +21,7 @@ import org.slf4j.LoggerFactory;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
 
+import eu.dnetlib.dhp.actionmanager.Constants;
 import eu.dnetlib.dhp.application.ArgumentApplicationParser;
 import eu.dnetlib.dhp.schema.action.AtomicAction;
 import eu.dnetlib.dhp.schema.common.ModelConstants;
@@ -29,6 +30,7 @@ import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
 import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
 import eu.dnetlib.dhp.schema.oaf.utils.PidCleaner;
 import eu.dnetlib.dhp.schema.oaf.utils.PidType;
+import io.netty.util.Constant;
 import scala.Tuple2;
 
 /**
@@ -44,8 +46,7 @@ public class CreateActionSetFromWebEntries implements Serializable {
 	private static final String PMID_PREFIX = "50|pmid________::";
 
 	private static final String PMCID_PREFIX = "50|pmc_________::";
-	private static final String WEB_CRAWL_ID = "10|openaire____::fb98a192f6a055ba495ef414c330834b";
+
-	private static final String WEB_CRAWL_NAME = "Web Crawl";
 	public static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
 
 	public static void main(String[] args) throws Exception {
@@ -220,7 +221,7 @@ public class CreateActionSetFromWebEntries implements Serializable {
 						ModelConstants.IS_AUTHOR_INSTITUTION_OF,
 						Arrays
 							.asList(
-								OafMapperUtils.keyValue(WEB_CRAWL_ID, WEB_CRAWL_NAME)),
+								OafMapperUtils.keyValue(Constants.WEB_CRAWL_ID, Constants.WEB_CRAWL_NAME)),
 						OafMapperUtils
 							.dataInfo(
 								false, null, false, false,
@@ -239,7 +240,7 @@ public class CreateActionSetFromWebEntries implements Serializable {
 						ModelConstants.HAS_AUTHOR_INSTITUTION,
 						Arrays
 							.asList(
-								OafMapperUtils.keyValue(WEB_CRAWL_ID, WEB_CRAWL_NAME)),
+								OafMapperUtils.keyValue(Constants.WEB_CRAWL_ID, Constants.WEB_CRAWL_NAME)),
 						OafMapperUtils
 							.dataInfo(
 								false, null, false, false,

dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/input_actionset_parameter.json

@@ -28,7 +28,13 @@
     "paramLongName": "dataciteInputPath",
     "paramDescription": "the path to get the input data from Datacite",
     "paramRequired": true
-  },
+  },{
+  "paramName": "wip",
+  "paramLongName": "webCrawlInputPath",
+  "paramDescription": "the path to get the input data from Web Crawl",
+  "paramRequired": true
+}
+,
   {
     "paramName": "o",
     "paramLongName": "outputPath",

dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/job.properties

@@ -35,5 +35,6 @@ crossrefInputPath=/data/bip-affiliations/crossref-data.json
 pubmedInputPath=/data/bip-affiliations/pubmed-data.json
 openapcInputPath=/data/bip-affiliations/openapc-data.json
 dataciteInputPath=/data/bip-affiliations/datacite-data.json
+webCrawlInputPath=/data/bip-affiliations/webCrawl/
 
 outputPath=/tmp/crossref-affiliations-output-v5

dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/oozie_app/workflow.xml

@@ -17,6 +17,10 @@
             <name>dataciteInputPath</name>
             <description>the path where to find the inferred affiliation relations from Datacite</description>
         </property>
+        <property>
+            <name>webCrawlInputPath</name>
+            <description>the path where to find the inferred affiliation relations from webCrawl</description>
+        </property>
         <property>
             <name>outputPath</name>
             <description>the path where to store the actionset</description>
@@ -112,7 +116,7 @@
             <arg>--pubmedInputPath</arg><arg>${pubmedInputPath}</arg>
             <arg>--openapcInputPath</arg><arg>${openapcInputPath}</arg>
             <arg>--dataciteInputPath</arg><arg>${dataciteInputPath}</arg>
-
+            <arg>--webCrawlInputPath</arg><arg>${webCrawlInputPath}</arg>
             <arg>--outputPath</arg><arg>${outputPath}</arg>
         </spark>
         <ok to="End"/>

dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelationsTest.java

@@ -88,6 +88,7 @@ public class PrepareAffiliationRelationsTest {
 					"-pubmedInputPath", crossrefAffiliationRelationPath,
 					"-openapcInputPath", crossrefAffiliationRelationPath,
 					"-dataciteInputPath", crossrefAffiliationRelationPath,
+					"-webCrawlInputPath", crossrefAffiliationRelationPath,
 					"-outputPath", outputPath
 				});
 
@@ -104,7 +105,7 @@ public class PrepareAffiliationRelationsTest {
 //            );
 //        }
 		// count the number of relations
-		assertEquals(80, tmp.count());
+		assertEquals(120, tmp.count());
 
 		Dataset<Relation> dataset = spark.createDataset(tmp.rdd(), Encoders.bean(Relation.class));
 		dataset.createOrReplaceTempView("result");
@@ -115,7 +116,7 @@ public class PrepareAffiliationRelationsTest {
 		// verify that we have equal number of bi-directional relations
 		Assertions
 			.assertEquals(
-				40, execVerification
+				60, execVerification
 					.filter(
 						"relClass='" + ModelConstants.HAS_AUTHOR_INSTITUTION + "'")
 					.collectAsList()
@@ -123,7 +124,7 @@ public class PrepareAffiliationRelationsTest {
 
 		Assertions
 			.assertEquals(
-				40, execVerification
+				60, execVerification
 					.filter(
 						"relClass='" + ModelConstants.IS_AUTHOR_INSTITUTION_OF + "'")
 					.collectAsList()

dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/doi_to_ror.json

@@ -4,4 +4,6 @@
 {"DOI":"10.1142\/s021821650200186x","Matchings":[{"RORid":"https:\/\/ror.org\/035xkbk20","Confidence":1},{"RORid":"https:\/\/ror.org\/05apxxy63","Confidence":1}]}
 {"DOI":"10.1061\/(asce)0733-9372(2002)128:7(575)","Matchings":[{"RORid":"https:\/\/ror.org\/04j198w64","Confidence":0.82}]}
 {"DOI":"10.1061\/(asce)0733-9372(2002)128:7(588)","Matchings":[{"RORid":"https:\/\/ror.org\/03m8km719","Confidence":0.8660254038},{"RORid":"https:\/\/ror.org\/02aze4h65","Confidence":0.87}]}
-{"DOI":"10.1161\/hy0202.103001","Matchings":[{"RORid":"https:\/\/ror.org\/057xtrt18","Confidence":0.7071067812}]}
+{"DOI":"10.1161\/hy0202.103001","Matchings":[{"RORid":"https:\/\/ror.org\/057xtrt18","Confidence":0.7071067812}]}
+{"DOI": "10.1080/13669877.2015.1042504", "Matchings": [{"Confidence": 1.0, "RORid": "https://ror.org/03265fv13"}]}
+{"DOI": "10.1007/3-540-47984-8_14", "Matchings": [{"Confidence": 1.0, "RORid": "https://ror.org/00a0n9e72"}]}