2020-03-27 10:42:17 +01:00
|
|
|
package eu.dnetlib.dhp.oa.dedup;
|
2019-12-06 13:38:00 +01:00
|
|
|
|
2019-12-18 16:28:22 +01:00
|
|
|
import com.google.common.hash.Hashing;
|
2020-03-27 10:42:17 +01:00
|
|
|
import eu.dnetlib.dhp.oa.dedup.graph.ConnectedComponent;
|
2019-12-06 13:38:00 +01:00
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
2020-03-27 10:42:17 +01:00
|
|
|
import eu.dnetlib.dhp.oa.dedup.graph.GraphProcessor;
|
2020-04-18 12:06:23 +02:00
|
|
|
import eu.dnetlib.dhp.schema.common.ModelSupport;
|
2020-04-17 13:12:44 +02:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.DataInfo;
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.Qualifier;
|
2019-12-06 13:38:00 +01:00
|
|
|
import eu.dnetlib.dhp.schema.oaf.Relation;
|
2020-04-06 16:30:31 +02:00
|
|
|
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
|
2020-03-20 13:01:56 +01:00
|
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
|
2020-04-06 16:30:31 +02:00
|
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
2019-12-06 13:38:00 +01:00
|
|
|
import eu.dnetlib.pace.config.DedupConfig;
|
|
|
|
import eu.dnetlib.pace.util.MapDocumentUtil;
|
|
|
|
import org.apache.commons.io.IOUtils;
|
2020-04-18 12:06:23 +02:00
|
|
|
import org.apache.spark.SparkConf;
|
2019-12-06 13:38:00 +01:00
|
|
|
import org.apache.spark.api.java.JavaPairRDD;
|
|
|
|
import org.apache.spark.api.java.JavaSparkContext;
|
|
|
|
import org.apache.spark.api.java.function.PairFunction;
|
|
|
|
import org.apache.spark.graphx.Edge;
|
|
|
|
import org.apache.spark.rdd.RDD;
|
|
|
|
import org.apache.spark.sql.Dataset;
|
|
|
|
import org.apache.spark.sql.Encoders;
|
2020-04-17 13:12:44 +02:00
|
|
|
import org.apache.spark.sql.SaveMode;
|
2019-12-06 13:38:00 +01:00
|
|
|
import org.apache.spark.sql.SparkSession;
|
2020-03-20 13:01:56 +01:00
|
|
|
import org.dom4j.DocumentException;
|
2019-12-06 13:38:00 +01:00
|
|
|
import scala.Tuple2;
|
2020-04-16 12:36:37 +02:00
|
|
|
import org.slf4j.Logger;
|
|
|
|
import org.slf4j.LoggerFactory;
|
2019-12-06 13:38:00 +01:00
|
|
|
|
2020-04-06 16:30:31 +02:00
|
|
|
import java.io.IOException;
|
2019-12-06 13:38:00 +01:00
|
|
|
import java.util.ArrayList;
|
2020-04-15 16:34:06 +02:00
|
|
|
import java.util.Iterator;
|
2019-12-06 13:38:00 +01:00
|
|
|
import java.util.List;
|
|
|
|
|
2020-04-16 12:13:26 +02:00
|
|
|
public class SparkCreateMergeRels extends AbstractSparkAction {
|
2020-04-06 16:30:31 +02:00
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
public static final String PROVENANCE_ACTION_CLASS = "sysimport:dedup";
|
2020-04-16 12:36:37 +02:00
|
|
|
private static final Logger log = LoggerFactory.getLogger(SparkCreateMergeRels.class);
|
2020-04-17 13:12:44 +02:00
|
|
|
public static final String DNET_PROVENANCE_ACTIONS = "dnet:provenanceActions";
|
2020-04-06 16:30:31 +02:00
|
|
|
|
2020-04-16 12:36:37 +02:00
|
|
|
public SparkCreateMergeRels(ArgumentApplicationParser parser, SparkSession spark) {
|
2020-04-06 16:30:31 +02:00
|
|
|
super(parser, spark);
|
|
|
|
}
|
2019-12-06 13:38:00 +01:00
|
|
|
|
|
|
|
public static void main(String[] args) throws Exception {
|
2020-04-06 16:30:31 +02:00
|
|
|
ArgumentApplicationParser parser = new ArgumentApplicationParser(
|
2020-03-27 10:42:17 +01:00
|
|
|
IOUtils.toString(
|
2020-04-06 16:30:31 +02:00
|
|
|
SparkCreateSimRels.class.getResourceAsStream("/eu/dnetlib/dhp/oa/dedup/createCC_parameters.json")));
|
2019-12-06 13:38:00 +01:00
|
|
|
parser.parseArgument(args);
|
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
final String isLookUpUrl = parser.get("isLookUpUrl");
|
|
|
|
log.info("isLookupUrl {}", isLookUpUrl);
|
|
|
|
|
2020-04-18 12:06:23 +02:00
|
|
|
SparkConf conf = new SparkConf();
|
|
|
|
conf.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer");
|
|
|
|
conf.registerKryoClasses(ModelSupport.getOafModelClasses());
|
|
|
|
|
|
|
|
new SparkCreateMergeRels(parser, getSparkSession(conf)).run(ISLookupClientFactory.getLookUpService(isLookUpUrl));
|
2020-03-20 13:01:56 +01:00
|
|
|
}
|
|
|
|
|
2020-04-06 16:30:31 +02:00
|
|
|
@Override
|
|
|
|
public void run(ISLookUpService isLookUpService) throws ISLookUpException, DocumentException, IOException {
|
2020-03-20 13:01:56 +01:00
|
|
|
|
|
|
|
final String graphBasePath = parser.get("graphBasePath");
|
|
|
|
final String workingPath = parser.get("workingPath");
|
|
|
|
final String isLookUpUrl = parser.get("isLookUpUrl");
|
|
|
|
final String actionSetId = parser.get("actionSetId");
|
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
log.info("graphBasePath: '{}'", graphBasePath);
|
|
|
|
log.info("isLookUpUrl: '{}'", isLookUpUrl);
|
|
|
|
log.info("actionSetId: '{}'", actionSetId);
|
|
|
|
log.info("workingPath: '{}'", workingPath);
|
2020-04-15 16:34:06 +02:00
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
final JavaSparkContext sc = JavaSparkContext.fromSparkContext(spark.sparkContext());
|
2020-04-06 16:30:31 +02:00
|
|
|
|
|
|
|
for (DedupConfig dedupConf: getConfigurations(isLookUpService, actionSetId)) {
|
|
|
|
final String subEntity = dedupConf.getWf().getSubEntityValue();
|
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
log.info("Creating mergerels for: '{}'", subEntity);
|
|
|
|
|
|
|
|
final int maxIterations = dedupConf.getWf().getMaxIterations();
|
|
|
|
log.info("Max iterations {}", maxIterations);
|
|
|
|
|
|
|
|
final String mergeRelPath = DedupUtility.createMergeRelPath(workingPath, actionSetId, subEntity);
|
2020-04-18 12:06:23 +02:00
|
|
|
|
2020-04-06 16:30:31 +02:00
|
|
|
final JavaPairRDD<Object, String> vertexes = sc.textFile(graphBasePath + "/" + subEntity)
|
|
|
|
.map(s -> MapDocumentUtil.getJPathString(dedupConf.getWf().getIdPath(), s))
|
2020-04-18 12:06:23 +02:00
|
|
|
.mapToPair((PairFunction<String, Object, String>) s -> new Tuple2<>(hash(s), s));
|
2020-04-17 13:12:44 +02:00
|
|
|
|
2020-04-18 12:06:23 +02:00
|
|
|
final RDD<Edge<String>> edgeRdd = spark
|
2020-04-17 13:12:44 +02:00
|
|
|
.read()
|
|
|
|
.load(DedupUtility.createSimRelPath(workingPath, actionSetId, subEntity))
|
2020-04-18 12:06:23 +02:00
|
|
|
.as(Encoders.bean(Relation.class))
|
2020-04-17 13:12:44 +02:00
|
|
|
.javaRDD()
|
2020-04-18 12:06:23 +02:00
|
|
|
.map(it -> new Edge<>(hash(it.getSource()), hash(it.getTarget()), it.getRelClass()))
|
2020-04-17 13:12:44 +02:00
|
|
|
.rdd();
|
|
|
|
|
2020-04-18 12:06:23 +02:00
|
|
|
final Dataset<Relation> mergeRels = spark
|
|
|
|
.createDataset(GraphProcessor.findCCs(vertexes.rdd(), edgeRdd, maxIterations)
|
2020-04-17 13:12:44 +02:00
|
|
|
.toJavaRDD()
|
|
|
|
.filter(k -> k.getDocIds().size() > 1)
|
|
|
|
.flatMap(cc -> ccToMergeRel(cc, dedupConf))
|
2020-04-18 12:06:23 +02:00
|
|
|
.rdd(), Encoders.bean(Relation.class));
|
2020-04-17 13:12:44 +02:00
|
|
|
|
2020-04-18 12:06:23 +02:00
|
|
|
mergeRels
|
2020-04-17 13:12:44 +02:00
|
|
|
.write()
|
|
|
|
.mode(SaveMode.Append)
|
2020-04-18 12:06:23 +02:00
|
|
|
.parquet(mergeRelPath);
|
2020-03-20 13:01:56 +01:00
|
|
|
}
|
2020-04-06 16:30:31 +02:00
|
|
|
|
2019-12-18 16:28:22 +01:00
|
|
|
}
|
2019-12-06 13:38:00 +01:00
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
public Iterator<Relation> ccToMergeRel(ConnectedComponent cc, DedupConfig dedupConf){
|
2020-04-15 16:34:06 +02:00
|
|
|
return cc.getDocIds()
|
|
|
|
.stream()
|
|
|
|
.flatMap(id -> {
|
|
|
|
List<Relation> tmp = new ArrayList<>();
|
2020-04-17 13:12:44 +02:00
|
|
|
|
|
|
|
tmp.add(rel(cc.getCcId(), id, "merges", dedupConf));
|
|
|
|
tmp.add(rel(id, cc.getCcId(), "isMergedIn", dedupConf));
|
|
|
|
|
2020-04-15 16:34:06 +02:00
|
|
|
return tmp.stream();
|
|
|
|
}).iterator();
|
|
|
|
}
|
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
private Relation rel(String source, String target, String relClass, DedupConfig dedupConf) {
|
|
|
|
Relation r = new Relation();
|
|
|
|
r.setSource(source);
|
|
|
|
r.setTarget(target);
|
|
|
|
r.setRelClass(relClass);
|
|
|
|
r.setSubRelType("dedup");
|
|
|
|
|
|
|
|
DataInfo info = new DataInfo();
|
|
|
|
info.setDeletedbyinference(false);
|
|
|
|
info.setInferred(true);
|
|
|
|
info.setInvisible(false);
|
|
|
|
info.setInferenceprovenance(dedupConf.getWf().getConfigurationId());
|
|
|
|
Qualifier provenanceAction = new Qualifier();
|
|
|
|
provenanceAction.setClassid(PROVENANCE_ACTION_CLASS);
|
|
|
|
provenanceAction.setClassname(PROVENANCE_ACTION_CLASS);
|
|
|
|
provenanceAction.setSchemeid(DNET_PROVENANCE_ACTIONS);
|
|
|
|
provenanceAction.setSchemename(DNET_PROVENANCE_ACTIONS);
|
|
|
|
info.setProvenanceaction(provenanceAction);
|
|
|
|
|
|
|
|
//TODO calculate the trust value based on the similarity score of the elements in the CC
|
|
|
|
//info.setTrust();
|
|
|
|
|
|
|
|
r.setDataInfo(info);
|
|
|
|
return r;
|
|
|
|
}
|
|
|
|
|
2020-04-18 12:06:23 +02:00
|
|
|
public static long hash(final String id) {
|
2020-03-20 16:48:14 +01:00
|
|
|
return Hashing.murmur3_128().hashString(id).asLong();
|
2019-12-06 13:38:00 +01:00
|
|
|
}
|
2020-03-20 13:01:56 +01:00
|
|
|
|
2019-12-06 13:38:00 +01:00
|
|
|
}
|