dimitris.pierrakos
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

implemented filtering step

This commit is contained in:
Sandro La Bruzzo 2020-05-23 08:46:49 +02:00
parent 244f6e50cf
commit 2408083566
6 changed files with 254 additions and 106 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
dhp-workflows/dhp-dedup-openaire/src/main/java/eu/dnetlib/dhp/oa/dedup/AuthorMerger.java
View File

@ -85,7 +85,8 @@ public class AuthorMerger {
if (simAuthor.isPresent()) {
double th = THRESHOLD;
// increase the threshold if the surname is too short
if (simAuthor.get()._2().getSurname() != null && simAuthor.get()._2().getSurname().length()<=3)
if (simAuthor.get()._2().getSurname() != null
&& simAuthor.get()._2().getSurname().length() <= 3)
th = 0.99;
if (simAuthor.get()._1() > th) {
@ -100,8 +101,9 @@ public class AuthorMerger {
}
public static String pidToComparableString(StructuredProperty pid) {
return (pid.getQualifier() != null ?
pid.getQualifier().getClassid() != null ? pid.getQualifier().getClassid().toLowerCase() : "" : "")
return (pid.getQualifier() != null
? pid.getQualifier().getClassid() != null ? pid.getQualifier().getClassid().toLowerCase() : ""
: "")
+ (pid.getValue() != null ? pid.getValue().toLowerCase() : "");
}
@ -125,12 +127,11 @@ public class AuthorMerger {
// if both are accurate (e.g. they have name and surname)
if (pa.isAccurate() & pb.isAccurate()) {
return
new JaroWinkler().score(normalize(pa.getSurnameString()), normalize(pb.getSurnameString()))*0.5
return new JaroWinkler().score(normalize(pa.getSurnameString()), normalize(pb.getSurnameString())) * 0.5
+ new JaroWinkler().score(normalize(pa.getNameString()), normalize(pb.getNameString())) * 0.5;
} else {
return
new JaroWinkler().score(normalize(pa.getNormalisedFullname()), normalize(pb.getNormalisedFullname()));
return new JaroWinkler()
.score(normalize(pa.getNormalisedFullname()), normalize(pb.getNormalisedFullname()));
}
}

147
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/DoiBoostMappingUtil.scala
View File

@ -1,7 +1,16 @@
package eu.dnetlib.doiboost
import eu.dnetlib.dhp.schema.oaf.{DataInfo, Dataset, Field, KeyValue, Qualifier, Result, StructuredProperty}
import eu.dnetlib.dhp.schema.oaf.{DataInfo, Dataset, Field, Instance, KeyValue, Publication, Qualifier, Result, StructuredProperty}
import eu.dnetlib.dhp.utils.DHPUtils
import org.json4s
import org.json4s.DefaultFormats
import org.json4s.jackson.JsonMethods.parse
import scala.collection.JavaConverters._
import scala.io.Source
case class HostedByItemType(id: String, officialName: String, issn: String, eissn: String, lissn: String, openAccess: Boolean) {}
object DoiBoostMappingUtil {
@ -19,12 +28,127 @@ object DoiBoostMappingUtil {
val DNET_LANGUAGES = "dnet:languages"
val PID_TYPES = "dnet:pid_types"
val invalidName = List(",", "none none", "none, none", "none &na;", "(:null)", "test test test", "test test", "test", "&na; &na;")
def retrieveHostedByMap(): Map[String, HostedByItemType] = {
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
val jsonMap = Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/doiboost/hbMap.json")).mkString
lazy val json: json4s.JValue = parse(jsonMap)
json.extract[Map[String, HostedByItemType]]
}
def retrieveHostedByItem(issn: String, eissn: String, lissn: String, hostedByMap: Map[String, HostedByItemType]): HostedByItemType = {
if (issn != null && issn.nonEmpty && hostedByMap.contains(issn))
return hostedByMap(issn)
if (eissn != null && eissn.nonEmpty && hostedByMap.contains(eissn))
return hostedByMap(eissn)
if (lissn != null && lissn.nonEmpty && hostedByMap.contains(lissn))
return hostedByMap(lissn)
null
}
def fixPublication(publication: Publication, hostedByMap: Map[String, HostedByItemType]): Publication = {
if (publication.getJournal == null)
return publication
val issn = publication.getJournal.getIssnPrinted
val eissn = publication.getJournal.getIssnOnline
val lissn = publication.getJournal.getIssnLinking
val item = retrieveHostedByItem(issn, eissn, lissn, hostedByMap)
if (item!= null) {
val l = publication.getInstance().asScala.map(i =>{
val hb = new KeyValue
hb.setValue (item.officialName)
hb.setKey (s"10|${item.id}" )
i.setHostedby(hb)
if(item.openAccess)
i.setAccessright(createQualifier("Open", "dnet:access_modes"))
i
}).asJava
publication.setInstance(l)
}
publication
}
def generateDataInfo (): DataInfo = {
generateDataInfo ("0.9")
}
def filterPublication (publication: Publication): Boolean = {
//Case empty publication
if (publication == null)
return false
//Case publication with no title
if (publication.getTitle == null || publication.getTitle.size == 0)
return false
val s = publication.getTitle.asScala.count (p => p.getValue != null
&& p.getValue.nonEmpty && ! p.getValue.equalsIgnoreCase ("[NO TITLE AVAILABLE]") )
if (s == 0)
return false
// fixes #4360 (test publisher)
val publisher = if (publication.getPublisher != null) publication.getPublisher.getValue else null
if (publisher != null && (publisher.equalsIgnoreCase ("Test accounts") || publisher.equalsIgnoreCase ("CrossRef Test Account") ) ) {
return false;
}
//Publication with no Author
if (publication.getAuthor == null || publication.getAuthor.size () == 0)
return false
//filter invalid author
val authors = publication.getAuthor.asScala.map (s => {
if (s.getFullname.nonEmpty) {
s.getFullname
}
else
s"${
s.getName
} ${
s.getSurname
}"
})
val c = authors.count (isValidAuthorName)
if (c == 0)
return false
// fixes #4368
if (authors.count (s => s.equalsIgnoreCase ("Addie Jackson") ) > 0 && "Elsevier BV".equalsIgnoreCase (publication.getPublisher.getValue) )
return false
true
}
def isValidAuthorName (fullName: String): Boolean = {
if (fullName == null || fullName.isEmpty)
return false
if (invalidName.contains (fullName.toLowerCase.trim) )
return false
true
}
def generateDataInfo (trust: String): DataInfo = {
val di = new DataInfo
di.setDeletedbyinference (false)
@ -82,18 +206,26 @@ object DoiBoostMappingUtil {
}
def generateIdentifier (oaf: Result, doi: String): String = {
val id = DHPUtils.md5 (doi.toLowerCase)
if (oaf.isInstanceOf[Dataset] )
return s"60|${doiBoostNSPREFIX}${SEPARATOR}${id}"
s"50|${doiBoostNSPREFIX}${SEPARATOR}${id}"
return s"60|${
doiBoostNSPREFIX
}${
SEPARATOR
}${
id
}"
s"50|${
doiBoostNSPREFIX
}${
SEPARATOR
}${
id
}"
}
def createMAGCollectedFrom (): KeyValue = {
val cf = new KeyValue
@ -126,5 +258,4 @@ object DoiBoostMappingUtil {
}
}

15
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/SparkGenerateDoiBoost.scala
View File

@ -32,11 +32,11 @@ object SparkGenerateDoiBoost {
logger.info("Phase 1) repartition and move all the dataset in a same working folder")
spark.read.load(crossrefPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/crossrefPublication")
spark.read.load(crossrefDatasetPath).as(Encoders.bean(classOf[OafDataset])).map(s => s)(Encoders.kryo[OafDataset]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/crossrefDataset")
spark.read.load(uwPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/uwPublication")
spark.read.load(orcidPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/orcidPublication")
spark.read.load(magPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/magPublication")
// spark.read.load(crossrefPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/crossrefPublication")
// spark.read.load(crossrefDatasetPath).as(Encoders.bean(classOf[OafDataset])).map(s => s)(Encoders.kryo[OafDataset]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/crossrefDataset")
// spark.read.load(uwPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/uwPublication")
// spark.read.load(orcidPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/orcidPublication")
// spark.read.load(magPublicationPath).as(Encoders.bean(classOf[Publication])).map(s => s)(Encoders.kryo[Publication]).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/magPublication")
implicit val mapEncoderPub: Encoder[Publication] = Encoders.kryo[Publication]
implicit val mapEncoderDataset: Encoder[OafDataset] = Encoders.kryo[OafDataset]
@ -67,14 +67,13 @@ object SparkGenerateDoiBoost {
logger.info("Phase 3) Join Result with MAG")
val sj: Dataset[(String, Publication)] = spark.read.load(s"$workingDirPath/secondJoin").as[Publication].map(p => (p.getId, p))
sj.where(sj("_1").like())
val magPublication: Dataset[(String, Publication)] = spark.read.load(s"$workingDirPath/magPublication").as[Publication].map(p => (p.getId, p))
sj.joinWith(magPublication, sj("_1").equalTo(magPublication("_1")), "left").map(applyMerge).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/doiBoostPublication")
val doiBoostPublication: Dataset[Publication] = spark.read.load(s"$workingDirPath/doiBoostPublication").as[Publication]
doiBoostPublication.filter(p=>DoiBoostMappingUtil.filterPublication(p)).write.mode(SaveMode.Overwrite).save(s"$workingDirPath/doiBoostPublicationFiltered")

1
dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/hbMap.json Normal file
View File

File diff suppressed because one or more lines are too long

16
dhp-workflows/dhp-doiboost/src/test/java/eu/dnetlib/dhp/doiboost/DoiBoostHostedByMapTest.scala Normal file
View File

@ -0,0 +1,16 @@
package eu.dnetlib.dhp.doiboost
import eu.dnetlib.doiboost.DoiBoostMappingUtil
import org.junit.jupiter.api.Test
class DoiBoostHostedByMapTest {
@Test
def testLoadMap(): Unit = {
println(DoiBoostMappingUtil.retrieveHostedByMap().keys.size)
}
}