sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

removed serialization points

This commit is contained in:
Miriam Baglioni 2020-05-15 12:49:58 +02:00
parent 1d35836a58
commit 5ec8c49ad5
1 changed files with 0 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
dhp-workflows/dhp-blacklist/src/main/java/eu/dnetlib/dhp/blacklist/SparkRemoveBlacklistedRelationJob.java
View File

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