przemyslaw.jacewicz
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

WIP monodirectional citations, Datacite

This commit is contained in:
Claudio Atzori 2023-02-28 13:30:51 +01:00
parent 0559d8b412
commit 2f7346e9cf
2 changed files with 41 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/datacite/DataciteModelConstants.scala
View File

@ -78,16 +78,6 @@ object DataciteModelConstants {
OafMapperUtils.keyValue(ModelConstants.DATACITE_ID, DATACITE_NAME)
val subRelTypeMapping: Map[String, OAFRelations] = Map(
ModelConstants.REFERENCES -> OAFRelations(
ModelConstants.REFERENCES,
ModelConstants.IS_REFERENCED_BY,
ModelConstants.RELATIONSHIP
),
ModelConstants.IS_REFERENCED_BY -> OAFRelations(
ModelConstants.IS_REFERENCED_BY,
ModelConstants.REFERENCES,
ModelConstants.RELATIONSHIP
),
ModelConstants.IS_SUPPLEMENTED_BY -> OAFRelations(
ModelConstants.IS_SUPPLEMENTED_BY,
ModelConstants.IS_SUPPLEMENT_TO,
@ -163,16 +153,6 @@ object DataciteModelConstants {
ModelConstants.IS_SOURCE_OF,
ModelConstants.VERSION
),
ModelConstants.CITES -> OAFRelations(
ModelConstants.CITES,
ModelConstants.IS_CITED_BY,
ModelConstants.CITATION
),
ModelConstants.IS_CITED_BY -> OAFRelations(
ModelConstants.IS_CITED_BY,
ModelConstants.CITES,
ModelConstants.CITATION
),
ModelConstants.IS_VARIANT_FORM_OF -> OAFRelations(
ModelConstants.IS_VARIANT_FORM_OF,
ModelConstants.IS_DERIVED_FROM,

61
dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/datacite/DataciteToOAFTransformation.scala
View File

@ -645,7 +645,7 @@ object DataciteToOAFTransformation {
id: String,
date: String
): List[Relation] = {
rels
val bidirectionalRels: List[Relation] = rels
.filter(r =>
subRelTypeMapping
.contains(r.relationType) && (r.relatedIdentifierType.equalsIgnoreCase("doi") ||
@ -653,27 +653,48 @@ object DataciteToOAFTransformation {
r.relatedIdentifierType.equalsIgnoreCase("arxiv"))
)
.map(r => {
val rel = new Relation
rel.setCollectedfrom(List(DATACITE_COLLECTED_FROM).asJava)
rel.setDataInfo(dataInfo)
val subRelType = subRelTypeMapping(r.relationType).relType
rel.setRelType(REL_TYPE_VALUE)
rel.setSubRelType(subRelType)
rel.setRelClass(r.relationType)
val dateProps: KeyValue = OafMapperUtils.keyValue(DATE_RELATION_KEY, date)
rel.setProperties(List(dateProps).asJava)
rel.setSource(id)
rel.setTarget(
DHPUtils.generateUnresolvedIdentifier(r.relatedIdentifier, r.relatedIdentifierType)
)
rel.setCollectedfrom(List(DATACITE_COLLECTED_FROM).asJava)
rel.getCollectedfrom.asScala.map(c => c.getValue).toList
rel
val target = DHPUtils.generateUnresolvedIdentifier(r.relatedIdentifier, r.relatedIdentifierType)
relation(id, target, subRelType, r.relationType, date)
})
val citationRels:List[Relation] = rels
.filter(r =>
(r.relatedIdentifierType.equalsIgnoreCase("doi") ||
r.relatedIdentifierType.equalsIgnoreCase("pmid") ||
r.relatedIdentifierType.equalsIgnoreCase("arxiv")) &&
(r.relationType.toLowerCase.contains("cite") || r.relationType.toLowerCase.contains("reference")))
.map(r => {
r.relationType match {
case ModelConstants.CITES | ModelConstants.REFERENCES =>
val target = DHPUtils.generateUnresolvedIdentifier(r.relatedIdentifier, r.relatedIdentifierType)
relation(id, target, ModelConstants.CITATION, ModelConstants.CITES, date)
case ModelConstants.IS_CITED_BY | ModelConstants.IS_REFERENCED_BY =>
val source = DHPUtils.generateUnresolvedIdentifier(r.relatedIdentifier, r.relatedIdentifierType)
relation(source, id, ModelConstants.CITATION, ModelConstants.CITES, date)
}
})
citationRels ::: bidirectionalRels
}
def relation(source:String, target:String, subRelType:String, relClass:String, date:String): Relation = {
val rel = new Relation
rel.setCollectedfrom(List(DATACITE_COLLECTED_FROM).asJava)
rel.setDataInfo(dataInfo)
rel.setRelType(REL_TYPE_VALUE)
rel.setSubRelType(subRelType)
rel.setRelClass(relClass)
val dateProps: KeyValue = OafMapperUtils.keyValue(DATE_RELATION_KEY, date)
rel.setProperties(List(dateProps).asJava)
rel.setSource(source)
rel.setTarget(target)
rel.setCollectedfrom(List(DATACITE_COLLECTED_FROM).asJava)
rel.getCollectedfrom.asScala.map(c => c.getValue).toList
rel
}
def generateDSId(input: String): String = {