master #13

Merged
claudio.atzori merged 90 commits from :master into enrichment_wfs 2020-05-26 09:34:24 +02:00
1 changed files with 0 additions and 7 deletions
Showing only changes of commit 5ec8c49ad5 - Show all commits

View File

@ -79,8 +79,6 @@ public class SparkRemoveBlacklistedRelationJob {
Dataset<Relation> inputRelation = readRelations(spark, inputPath);
Dataset<Relation> mergesRelation = readRelations(spark, mergesPath);
log.info("InputRelationCount: {}", inputRelation.count());
Dataset<Relation> dedupSource = blackListed
.joinWith(
mergesRelation, blackListed.col("source").equalTo(mergesRelation.col("target")),
@ -103,11 +101,6 @@ public class SparkRemoveBlacklistedRelationJob {
return c._1();
}, Encoders.bean(Relation.class));
dedupBL
.write()
.mode(SaveMode.Overwrite)
.json(blacklistPath + "/deduped");
inputRelation
.joinWith(
dedupBL, (inputRelation