2023-07-05 23:51:01 +02:00
|
|
|
|
|
|
|
package eu.dnetlib.dhp.actionmanager.bipaffiliations;
|
|
|
|
|
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
|
|
|
|
import java.io.Serializable;
|
2023-07-06 17:29:15 +02:00
|
|
|
import java.util.Arrays;
|
|
|
|
import java.util.List;
|
2023-07-05 23:51:01 +02:00
|
|
|
import java.util.stream.Collectors;
|
|
|
|
|
|
|
|
import eu.dnetlib.dhp.actionmanager.Constants;
|
|
|
|
import eu.dnetlib.dhp.actionmanager.bipaffiliations.model.*;
|
2023-07-06 17:08:33 +02:00
|
|
|
import eu.dnetlib.dhp.actionmanager.ror.GenerateRorActionSetJob;
|
2023-07-06 14:56:12 +02:00
|
|
|
import eu.dnetlib.dhp.schema.action.AtomicAction;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.utils.CleaningFunctions;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory;
|
2023-07-05 23:51:01 +02:00
|
|
|
import org.apache.commons.io.IOUtils;
|
2023-07-06 14:56:12 +02:00
|
|
|
import org.apache.hadoop.io.Text;
|
|
|
|
import org.apache.hadoop.mapred.SequenceFileOutputFormat;
|
2023-07-05 23:51:01 +02:00
|
|
|
import org.apache.spark.SparkConf;
|
|
|
|
import org.apache.spark.api.java.JavaRDD;
|
|
|
|
import org.apache.spark.api.java.JavaSparkContext;
|
2023-07-06 17:29:15 +02:00
|
|
|
import org.apache.spark.api.java.function.FlatMapFunction;
|
2023-07-06 14:56:12 +02:00
|
|
|
import org.apache.spark.api.java.function.MapFunction;
|
2023-07-05 23:51:01 +02:00
|
|
|
import org.apache.spark.sql.Encoders;
|
|
|
|
import org.apache.spark.sql.SparkSession;
|
|
|
|
import org.slf4j.Logger;
|
|
|
|
import org.slf4j.LoggerFactory;
|
|
|
|
import org.apache.spark.sql.Dataset;
|
|
|
|
|
|
|
|
import com.fasterxml.jackson.databind.ObjectMapper;
|
|
|
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
|
|
|
import eu.dnetlib.dhp.schema.common.ModelConstants;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.*;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils;
|
2023-07-06 14:56:12 +02:00
|
|
|
import scala.Tuple2;
|
2023-07-05 23:51:01 +02:00
|
|
|
|
|
|
|
/**
|
2023-07-06 17:08:33 +02:00
|
|
|
* Creates action sets for Crossref affiliation relations inferred by BIP!
|
2023-07-05 23:51:01 +02:00
|
|
|
*/
|
|
|
|
public class PrepareAffiliationRelations implements Serializable {
|
|
|
|
|
|
|
|
private static final Logger log = LoggerFactory.getLogger(PrepareAffiliationRelations.class);
|
|
|
|
private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
|
2023-07-06 14:56:12 +02:00
|
|
|
private static final String ID_PREFIX = "50|doi_________::";
|
2023-07-06 17:08:33 +02:00
|
|
|
public static final String BIP_AFFILIATIONS_CLASSID = "result:organization:bipinference";
|
|
|
|
public static final String BIP_AFFILIATIONS_CLASSNAME = "Affiliation relation inferred by BIP!";
|
|
|
|
public static final String BIP_INFERENCE_PROVENANCE = "bip_affiliation";
|
2023-07-05 23:51:01 +02:00
|
|
|
|
|
|
|
public static <I extends Result> void main(String[] args) throws Exception {
|
|
|
|
|
|
|
|
String jsonConfiguration = IOUtils
|
2023-07-06 17:08:33 +02:00
|
|
|
.toString(
|
|
|
|
PrepareAffiliationRelations.class
|
|
|
|
.getResourceAsStream(
|
|
|
|
"/eu/dnetlib/dhp/actionmanager/bipaffiliations/input_actionset_parameter.json"));
|
2023-07-05 23:51:01 +02:00
|
|
|
|
|
|
|
final ArgumentApplicationParser parser = new ArgumentApplicationParser(jsonConfiguration);
|
|
|
|
parser.parseArgument(args);
|
|
|
|
|
|
|
|
Boolean isSparkSessionManaged = Constants.isSparkSessionManaged(parser);
|
|
|
|
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
|
|
|
|
|
|
|
|
final String inputPath = parser.get("inputPath");
|
|
|
|
log.info("inputPath {}: ", inputPath);
|
|
|
|
|
|
|
|
final String outputPath = parser.get("outputPath");
|
|
|
|
log.info("outputPath {}: ", outputPath);
|
|
|
|
|
|
|
|
SparkConf conf = new SparkConf();
|
|
|
|
|
|
|
|
runWithSparkSession(
|
|
|
|
conf,
|
|
|
|
isSparkSessionManaged,
|
|
|
|
spark -> {
|
|
|
|
Constants.removeOutputDir(spark, outputPath);
|
|
|
|
prepareAffiliationRelations(spark, inputPath, outputPath);
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
private static <I extends Result> void prepareAffiliationRelations(SparkSession spark, String inputPath, String outputPath) {
|
|
|
|
|
|
|
|
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
|
|
|
|
|
2023-07-06 17:08:33 +02:00
|
|
|
// load and parse affiliation relations from HDFS
|
2023-07-06 14:56:12 +02:00
|
|
|
JavaRDD<AffiliationRelationDeserializer> affiliationRelationsDeserializeRDD = sc
|
2023-07-05 23:51:01 +02:00
|
|
|
.textFile(inputPath)
|
|
|
|
.map(item -> OBJECT_MAPPER.readValue(item, AffiliationRelationDeserializer.class));
|
|
|
|
|
2023-07-06 17:08:33 +02:00
|
|
|
// convert affiliation to an internal representation
|
2023-07-06 14:56:12 +02:00
|
|
|
Dataset<AffiliationRelationModel> affiliationRelations =
|
|
|
|
spark.createDataset(
|
|
|
|
affiliationRelationsDeserializeRDD.flatMap(entry ->
|
|
|
|
entry.getMatchings().stream().flatMap(matching ->
|
|
|
|
matching.getRorId().stream().map( rorId -> new AffiliationRelationModel(
|
|
|
|
entry.getDoi(),
|
|
|
|
rorId,
|
|
|
|
matching.getConfidence()
|
|
|
|
))).collect(Collectors.toList())
|
|
|
|
.iterator())
|
|
|
|
.rdd(),
|
|
|
|
Encoders.bean(AffiliationRelationModel.class));
|
|
|
|
|
2023-07-06 17:08:33 +02:00
|
|
|
// prepare action sets for affiliation relations
|
2023-07-06 14:56:12 +02:00
|
|
|
affiliationRelations
|
2023-07-06 17:29:15 +02:00
|
|
|
.flatMap((FlatMapFunction<AffiliationRelationModel, Relation>) affRel -> {
|
2023-07-06 14:56:12 +02:00
|
|
|
|
2023-07-06 17:08:33 +02:00
|
|
|
// DOI to OpenAIRE id
|
|
|
|
final String paperId = ID_PREFIX + IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", affRel.getDoi()));
|
|
|
|
|
|
|
|
// ROR id to OpenAIRE id
|
|
|
|
final String affId = GenerateRorActionSetJob.calculateOpenaireId(affRel.getRorId());
|
|
|
|
|
|
|
|
Qualifier qualifier = OafMapperUtils.qualifier(
|
|
|
|
BIP_AFFILIATIONS_CLASSID,
|
|
|
|
BIP_AFFILIATIONS_CLASSNAME,
|
|
|
|
ModelConstants.DNET_PROVENANCE_ACTIONS,
|
|
|
|
ModelConstants.DNET_PROVENANCE_ACTIONS);
|
|
|
|
|
|
|
|
// format data info; setting `confidence` into relation's `trust`
|
|
|
|
DataInfo dataInfo = OafMapperUtils.dataInfo(
|
|
|
|
false,
|
|
|
|
BIP_INFERENCE_PROVENANCE,
|
|
|
|
true,
|
|
|
|
false,
|
|
|
|
qualifier,
|
|
|
|
Double.toString(affRel.getConfidence()));
|
|
|
|
|
2023-07-06 17:29:15 +02:00
|
|
|
// return bi-directional relations
|
|
|
|
return getAffiliationRelationPair(paperId, affId, dataInfo).iterator();
|
2023-07-06 14:56:12 +02:00
|
|
|
|
|
|
|
}, Encoders.bean(Relation.class))
|
|
|
|
.toJavaRDD()
|
|
|
|
.map(p -> new AtomicAction(Relation.class, p))
|
|
|
|
.mapToPair(
|
|
|
|
aa -> new Tuple2<>(new Text(aa.getClazz().getCanonicalName()),
|
|
|
|
new Text(OBJECT_MAPPER.writeValueAsString(aa))))
|
|
|
|
.saveAsHadoopFile(outputPath, Text.class, Text.class, SequenceFileOutputFormat.class);
|
|
|
|
|
|
|
|
}
|
2023-07-06 17:29:15 +02:00
|
|
|
|
|
|
|
private static List<Relation> getAffiliationRelationPair(String paperId, String affId, DataInfo dataInfo) {
|
|
|
|
return Arrays.asList(
|
|
|
|
OafMapperUtils.getRelation(
|
|
|
|
paperId,
|
|
|
|
affId,
|
|
|
|
ModelConstants.RESULT_ORGANIZATION,
|
|
|
|
ModelConstants.AFFILIATION,
|
|
|
|
ModelConstants.HAS_AUTHOR_INSTITUTION,
|
|
|
|
null,
|
|
|
|
dataInfo,
|
|
|
|
null),
|
|
|
|
OafMapperUtils.getRelation(
|
|
|
|
affId,
|
|
|
|
paperId,
|
|
|
|
ModelConstants.RESULT_ORGANIZATION,
|
|
|
|
ModelConstants.AFFILIATION,
|
|
|
|
ModelConstants.IS_AUTHOR_INSTITUTION_OF,
|
|
|
|
null,
|
|
|
|
dataInfo,
|
|
|
|
null)
|
|
|
|
);
|
|
|
|
}
|
2023-07-05 23:51:01 +02:00
|
|
|
}
|