D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

Update to include a blackList that filters out the results we know are wrongly associated to IE - refactoring

This commit is contained in:
Miriam Baglioni 2024-05-24 15:23:42 +02:00 committed by Claudio Atzori
parent b55fed09f8
commit 87c9c61b41
2 changed files with 33 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/webcrawl/CreateActionSetFromWebEntries.java
View File

@ -95,7 +95,8 @@ public class CreateActionSetFromWebEntries implements Serializable {
final Dataset<Row> blackList = readBlackList(spark, blackListInputPath);
dataset.join(blackList, dataset.col("id").equalTo(blackList.col("OpenAlexId")), "left")
dataset
.join(blackList, dataset.col("id").equalTo(blackList.col("OpenAlexId")), "left")
.filter((FilterFunction<Row>) r -> r.getAs("OpenAlexId") == null)
.drop("OpenAlexId")
.flatMap((FlatMapFunction<Row, Relation>) row -> {
@ -114,7 +115,7 @@ public class CreateActionSetFromWebEntries implements Serializable {
.mapToPair(
aa -> new Tuple2<>(new Text(aa.getClazz().getCanonicalName()),
new Text(OBJECT_MAPPER.writeValueAsString(aa))))
.saveAsHadoopFile(outputPath, Text.class, Text.class, SequenceFileOutputFormat.class);//, GzipCodec.class);
.saveAsHadoopFile(outputPath, Text.class, Text.class, SequenceFileOutputFormat.class, GzipCodec.class);
}
@ -145,7 +146,7 @@ public class CreateActionSetFromWebEntries implements Serializable {
}
private static Dataset<Row> readBlackList(SparkSession spark, String inputPath){
private static Dataset<Row> readBlackList(SparkSession spark, String inputPath) {
return spark
.read()