Merge pull request 'New Model from affRo' (#487) from affRoModelModificationOnmain into main

Reviewed-on: D-Net/dnet-hadoop#487

dhp-workflows/dhp-aggregation/src/main/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelations.java

@@ -45,6 +45,8 @@ public class PrepareAffiliationRelations implements Serializable {
 	public static final String BIP_INFERENCE_PROVENANCE = "openaire:affiliation";
 	public static final String OPENAIRE_DATASOURCE_ID = "10|infrastruct_::f66f1bd369679b5b077dcdf006089556";
 	public static final String OPENAIRE_DATASOURCE_NAME = "OpenAIRE";
+	public static final String DOI_URL_PREFIX = "https://doi.org/";
+	public static final int DOI_URL_PREFIX_LENGTH = 16;
 
 	public static <I extends Result> void main(String[] args) throws Exception {
 
@@ -97,35 +99,26 @@ public class PrepareAffiliationRelations implements Serializable {
 	private static void createActionSet(SparkSession spark, String crossrefInputPath, String pubmedInputPath,
 		String openapcInputPath, String dataciteInputPath, String webcrawlInputPath, String publisherlInputPath,
 		String outputPath) {
-		List<KeyValue> collectedFromCrossref = OafMapperUtils
+		List<KeyValue> collectedfromOpenAIRE = OafMapperUtils
-			.listKeyValues(ModelConstants.CROSSREF_ID, "Crossref");
+				.listKeyValues(OPENAIRE_DATASOURCE_ID, OPENAIRE_DATASOURCE_NAME);
-		JavaPairRDD<Text, Text> crossrefRelations = prepareAffiliationRelations(
+
-			spark, crossrefInputPath, collectedFromCrossref);
+		JavaPairRDD<Text, Text> crossrefRelations = prepareAffiliationRelationsNewModel(
+			spark, crossrefInputPath, collectedfromOpenAIRE);
 
-		List<KeyValue> collectedFromPubmed = OafMapperUtils
-			.listKeyValues(ModelConstants.PUBMED_CENTRAL_ID, "Pubmed");
 		JavaPairRDD<Text, Text> pubmedRelations = prepareAffiliationRelations(
-			spark, pubmedInputPath, collectedFromPubmed);
+			spark, pubmedInputPath, collectedfromOpenAIRE);
 
-		List<KeyValue> collectedFromOpenAPC = OafMapperUtils
+		JavaPairRDD<Text, Text> openAPCRelations = prepareAffiliationRelationsNewModel(
-			.listKeyValues(ModelConstants.OPEN_APC_ID, "OpenAPC");
+			spark, openapcInputPath, collectedfromOpenAIRE);
-		JavaPairRDD<Text, Text> openAPCRelations = prepareAffiliationRelations(
-			spark, openapcInputPath, collectedFromOpenAPC);
 
-		List<KeyValue> collectedFromDatacite = OafMapperUtils
-			.listKeyValues(ModelConstants.DATACITE_ID, "Datacite");
 		JavaPairRDD<Text, Text> dataciteRelations = prepareAffiliationRelations(
-			spark, dataciteInputPath, collectedFromDatacite);
+			spark, dataciteInputPath, collectedfromOpenAIRE);
 
-		List<KeyValue> collectedFromWebCrawl = OafMapperUtils
-			.listKeyValues(OPENAIRE_DATASOURCE_ID, OPENAIRE_DATASOURCE_NAME);
 		JavaPairRDD<Text, Text> webCrawlRelations = prepareAffiliationRelations(
-			spark, webcrawlInputPath, collectedFromWebCrawl);
+			spark, webcrawlInputPath, collectedfromOpenAIRE);
 
-		List<KeyValue> collectedfromPublisher = OafMapperUtils
-			.listKeyValues(OPENAIRE_DATASOURCE_ID, OPENAIRE_DATASOURCE_NAME);
 		JavaPairRDD<Text, Text> publisherRelations = prepareAffiliationRelationFromPublisher(
-			spark, publisherlInputPath, collectedfromPublisher);
+			spark, publisherlInputPath, collectedfromOpenAIRE);
 
 		crossrefRelations
 			.union(pubmedRelations)
@@ -137,6 +130,20 @@ public class PrepareAffiliationRelations implements Serializable {
 				outputPath, Text.class, Text.class, SequenceFileOutputFormat.class, BZip2Codec.class);
 	}
 
+	private static JavaPairRDD<Text, Text> prepareAffiliationRelationFromPublisherNewModel(SparkSession spark, String inputPath,
+		List<KeyValue> collectedfrom) {
+
+		Dataset<Row> df = spark
+			.read()
+			.schema(
+				"`DOI` STRING, `Organizations` ARRAY<STRUCT<`PID`:STRING, `Value`:STRING,`Confidence`:DOUBLE, `Status`:STRING>>")
+			.json(inputPath)
+			.where("DOI is not null");
+
+		return getTextTextJavaPairRDD(collectedfrom, df.selectExpr("DOI", "Organizations as Matchings"));
+
+	}
+
 	private static JavaPairRDD<Text, Text> prepareAffiliationRelationFromPublisher(SparkSession spark, String inputPath,
 																				   List<KeyValue> collectedfrom) {
 
@@ -146,6 +153,7 @@ public class PrepareAffiliationRelations implements Serializable {
 				.json(inputPath)
 				.where("DOI is not null");
 
+
 		return getTextTextJavaPairRDD(collectedfrom, df.selectExpr("DOI", "Organizations as Matchings"));
 
 	}
@@ -164,6 +172,20 @@ public class PrepareAffiliationRelations implements Serializable {
 		return getTextTextJavaPairRDD(collectedfrom, df);
 	}
 
+	private static <I extends Result> JavaPairRDD<Text, Text> prepareAffiliationRelationsNewModel(SparkSession spark,
+		String inputPath,
+		List<KeyValue> collectedfrom) {
+		// load and parse affiliation relations from HDFS
+		Dataset<Row> df = spark
+			.read()
+			.schema(
+				"`DOI` STRING, `Matchings` ARRAY<STRUCT<`PID`:STRING, `Value`:STRING,`Confidence`:DOUBLE, `Status`:STRING>>")
+			.json(inputPath)
+			.where("DOI is not null");
+
+		return getTextTextJavaPairRDDNew(collectedfrom, df);
+	}
+
 	private static JavaPairRDD<Text, Text> getTextTextJavaPairRDD(List<KeyValue> collectedfrom, Dataset<Row> df) {
 		// unroll nested arrays
 		df = df
@@ -180,7 +202,7 @@ public class PrepareAffiliationRelations implements Serializable {
 
 					// DOI to OpenAIRE id
 					final String paperId = ID_PREFIX
-					+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", row.getAs("doi")));
+							+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", removePrefix(row.getAs("doi"))));
 
 					// ROR id to OpenAIRE id
 					final String affId = GenerateRorActionSetJob.calculateOpenaireId(row.getAs("rorid"));
@@ -212,6 +234,69 @@ public class PrepareAffiliationRelations implements Serializable {
 								new Text(OBJECT_MAPPER.writeValueAsString(aa))));
 	}
 
+	private static JavaPairRDD<Text, Text> getTextTextJavaPairRDDNew(List<KeyValue> collectedfrom, Dataset<Row> df) {
+		// unroll nested arrays
+		df = df
+			.withColumn("matching", functions.explode(new Column("Matchings")))
+			.select(
+				new Column("DOI").as("doi"),
+				new Column("matching.PID").as("pidtype"),
+				new Column("matching.Value").as("pidvalue"),
+				new Column("matching.Confidence").as("confidence"),
+				new Column("matching.Status").as("status"))
+			.where("status = 'active'");
+
+		// prepare action sets for affiliation relations
+		return df
+			.toJavaRDD()
+			.flatMap((FlatMapFunction<Row, Relation>) row -> {
+
+				// DOI to OpenAIRE id
+				final String paperId = ID_PREFIX
+					+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", removePrefix(row.getAs("doi"))));
+
+				// Organization to OpenAIRE identifier
+				String affId = null;
+				if (row.getAs("pidtype").equals("ROR"))
+					// ROR id to OpenIARE id
+					affId = GenerateRorActionSetJob.calculateOpenaireId(row.getAs("pidvalue"));
+				else
+					// getting the OpenOrgs identifier for the organization
+					affId = row.getAs("pidvalue");
+
+				Qualifier qualifier = OafMapperUtils
+					.qualifier(
+						BIP_AFFILIATIONS_CLASSID,
+						BIP_AFFILIATIONS_CLASSNAME,
+						ModelConstants.DNET_PROVENANCE_ACTIONS,
+						ModelConstants.DNET_PROVENANCE_ACTIONS);
+
+				// format data info; setting `confidence` into relation's `trust`
+				DataInfo dataInfo = OafMapperUtils
+					.dataInfo(
+						false,
+						BIP_INFERENCE_PROVENANCE,
+						true,
+						false,
+						qualifier,
+						Double.toString(row.getAs("confidence")));
+
+				// return bi-directional relations
+				return getAffiliationRelationPair(paperId, affId, collectedfrom, dataInfo).iterator();
+
+			})
+			.map(p -> new AtomicAction(Relation.class, p))
+			.mapToPair(
+				aa -> new Tuple2<>(new Text(aa.getClazz().getCanonicalName()),
+					new Text(OBJECT_MAPPER.writeValueAsString(aa))));
+	}
+
+	private static String removePrefix(String doi) {
+		if(doi.startsWith(DOI_URL_PREFIX))
+			return doi.substring(DOI_URL_PREFIX_LENGTH);
+		return doi;
+	}
+
 	private static List<Relation> getAffiliationRelationPair(String paperId, String affId, List<KeyValue> collectedfrom,
 		DataInfo dataInfo) {
 		return Arrays

dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/actionmanager/bipaffiliations/PrepareAffiliationRelationsTest.java

@@ -74,26 +74,34 @@ public class PrepareAffiliationRelationsTest {
 	@Test
 	void testMatch() throws Exception {
 
-		String crossrefAffiliationRelationPath = getClass()
+		String crossrefAffiliationRelationPathNew = getClass()
 			.getResource("/eu/dnetlib/dhp/actionmanager/bipaffiliations/doi_to_ror.json")
 			.getPath();
 
+		String crossrefAffiliationRelationPath = getClass()
+				.getResource("/eu/dnetlib/dhp/actionmanager/bipaffiliations/doi_to_ror_old.json")
+				.getPath();
+
 		String publisherAffiliationRelationPath = getClass()
 			.getResource("/eu/dnetlib/dhp/actionmanager/bipaffiliations/publishers")
 			.getPath();
 
+		String publisherAffiliationRelationOldPath = getClass()
+				.getResource("/eu/dnetlib/dhp/actionmanager/bipaffiliations/publichers_old")
+				.getPath();
+
 		String outputPath = workingDir.toString() + "/actionSet";
 
 		PrepareAffiliationRelations
 			.main(
 				new String[] {
 					"-isSparkSessionManaged", Boolean.FALSE.toString(),
-					"-crossrefInputPath", crossrefAffiliationRelationPath,
+					"-crossrefInputPath", crossrefAffiliationRelationPathNew,
 					"-pubmedInputPath", crossrefAffiliationRelationPath,
-					"-openapcInputPath", crossrefAffiliationRelationPath,
+					"-openapcInputPath", crossrefAffiliationRelationPathNew,
 					"-dataciteInputPath", crossrefAffiliationRelationPath,
 					"-webCrawlInputPath", crossrefAffiliationRelationPath,
-					"-publisherInputPath", publisherAffiliationRelationPath,
+					"-publisherInputPath", publisherAffiliationRelationOldPath,
 					"-outputPath", outputPath
 				});
 
@@ -104,13 +112,9 @@ public class PrepareAffiliationRelationsTest {
 			.map(value -> OBJECT_MAPPER.readValue(value._2().toString(), AtomicAction.class))
 			.map(aa -> ((Relation) aa.getPayload()));
 
-//        for (Relation r : tmp.collect()) {
+
-//            System.out.println(
-//                    r.getSource() + "\t" + r.getTarget() + "\t" + r.getRelType() + "\t" + r.getRelClass() + "\t" + r.getSubRelType() + "\t" + r.getValidationDate() + "\t" + r.getDataInfo().getTrust() + "\t" + r.getDataInfo().getInferred()
-//            );
-//        }
 		// count the number of relations
-		assertEquals(138, tmp.count());
+		assertEquals(150, tmp.count());//  18 + 24 *3 + 30 * 2 =
 
 		Dataset<Relation> dataset = spark.createDataset(tmp.rdd(), Encoders.bean(Relation.class));
 		dataset.createOrReplaceTempView("result");
@@ -121,7 +125,7 @@ public class PrepareAffiliationRelationsTest {
 		// verify that we have equal number of bi-directional relations
 		Assertions
 			.assertEquals(
-				69, execVerification
+				75, execVerification
 					.filter(
 						"relClass='" + ModelConstants.HAS_AUTHOR_INSTITUTION + "'")
 					.collectAsList()
@@ -129,21 +133,21 @@ public class PrepareAffiliationRelationsTest {
 
 		Assertions
 			.assertEquals(
-				69, execVerification
+				75, execVerification
 					.filter(
 						"relClass='" + ModelConstants.IS_AUTHOR_INSTITUTION_OF + "'")
 					.collectAsList()
 					.size());
 
 		// check confidence value of a specific relation
-		String sourceDOI = "10.1061/(asce)0733-9399(2002)128:7(759)";
+		String sourceDOI = "10.1089/10872910260066679";
 
 		final String sourceOpenaireId = ID_PREFIX
 			+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", sourceDOI));
 
 		Assertions
 			.assertEquals(
-				"0.7071067812", execVerification
+				"1.0", execVerification
 					.filter(
 						"source='" + sourceOpenaireId + "'")
 					.collectAsList()
@@ -151,11 +155,36 @@ public class PrepareAffiliationRelationsTest {
 					.getString(4));
 
 		final String publisherid = ID_PREFIX
-			+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s00217-010-1268-9"));
+			+ IdentifierFactory.md5(CleaningFunctions.normalizePidValue("doi", "10.1089/10872910260066679"));
-		final String rorId = "20|ror_________::" + IdentifierFactory.md5("https://ror.org/03265fv13");
+		final String rorId = "20|ror_________::" + IdentifierFactory.md5("https://ror.org/05cf8a891");
 
 		Assertions
 			.assertEquals(
-				1, execVerification.filter("source = '" + publisherid + "' and target = '" + rorId + "'").count());
+				2, execVerification.filter("source = '" + publisherid + "' and target = '" + rorId + "'").count());
+
+		Assertions
+			.assertEquals(
+				1, execVerification
+					.filter(
+						"source = '" + ID_PREFIX
+							+ IdentifierFactory
+								.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/s00217-010-1268-9"))
+							+ "' and target = '" + "20|ror_________::"
+							+ IdentifierFactory.md5("https://ror.org/03265fv13") + "'")
+					.count());
+
+
+		Assertions
+				.assertEquals(
+						3, execVerification
+								.filter(
+										"source = '" + ID_PREFIX
+												+ IdentifierFactory
+												.md5(CleaningFunctions.normalizePidValue("doi", "10.1007/3-540-47984-8_14"))
+												+ "' and target = '" + "20|ror_________::"
+												+ IdentifierFactory.md5("https://ror.org/00a0n9e72") + "'")
+								.count());
+
+
 	}
 }

dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/doi_to_ror.json

@@ -1,9 +1,10 @@
-{"DOI":"10.1061\/(asce)0733-9399(2002)128:7(759)","Matchings":[{"RORid":"https:\/\/ror.org\/03yxnpp24","Confidence":0.7071067812},{"RORid":"https:\/\/ror.org\/01teme464","Confidence":0.89}]}
+{"DOI":"10.1021\/ac020069k","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/01f5ytq51","Status":"active","Confidence":1}]}
-{"DOI":"10.1105\/tpc.8.3.343","Matchings":[{"RORid":"https:\/\/ror.org\/02k40bc56","Confidence":0.7071067812}]}
+{"DOI":"10.1161\/01.cir.0000013846.72805.7e","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/02pttbw34","Status":"active","Confidence":1}]}
-{"DOI":"10.1161\/01.cir.0000013305.01850.37","Matchings":[{"RORid":"https:\/\/ror.org\/00qjgza05","Confidence":1}]}
+{"DOI":"10.1161\/hy02t2.102992","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/00qqv6244","Status":"active","Confidence":1},{"PID":"ROR","Value":"https:\/\/ror.org\/00p991c53","Status":"active","Confidence":1}]}
-{"DOI":"10.1142\/s021821650200186x","Matchings":[{"RORid":"https:\/\/ror.org\/035xkbk20","Confidence":1},{"RORid":"https:\/\/ror.org\/05apxxy63","Confidence":1}]}
+{"DOI":"10.1126\/science.1073633","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/03xez1567","Status":"active","Confidence":1},{"PID":"ROR","Value":"https:\/\/ror.org\/006w34k90","Status":"active","Confidence":1}]}
-{"DOI":"10.1061\/(asce)0733-9372(2002)128:7(575)","Matchings":[{"RORid":"https:\/\/ror.org\/04j198w64","Confidence":0.82}]}
+{"DOI":"10.1089\/10872910260066679","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/05cf8a891","Status":"active","Confidence":1}]}
-{"DOI":"10.1061\/(asce)0733-9372(2002)128:7(588)","Matchings":[{"RORid":"https:\/\/ror.org\/03m8km719","Confidence":0.8660254038},{"RORid":"https:\/\/ror.org\/02aze4h65","Confidence":0.87}]}
+{"DOI":"10.1108\/02656719610116117","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/03mnm0t94","Status":"active","Confidence":1},{"PID":"ROR","Value":"https:\/\/ror.org\/007tn5k56","Status":"active","Confidence":1}]}
-{"DOI":"10.1161\/hy0202.103001","Matchings":[{"RORid":"https:\/\/ror.org\/057xtrt18","Confidence":0.7071067812}]}
+{"DOI":"10.1080\/01443610050111986","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/001x4vz59","Status":"active","Confidence":1},{"PID":"ROR","Value":"https:\/\/ror.org\/01tmqtf75","Status":"active","Confidence":1}]}
-{"DOI": "10.1080/13669877.2015.1042504", "Matchings": [{"Confidence": 1.0, "RORid": "https://ror.org/03265fv13"}]}
+{"DOI":"10.1021\/cm020118+","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/02cf1je33","Confidence":1,"Status":"inactive"},{"PID":"ROR","Value":"https:\/\/ror.org\/01hvx5h04","Confidence":1,"Status":"active"}]}
-{"DOI": "10.1007/3-540-47984-8_14", "Matchings": [{"Confidence": 1.0, "RORid": "https://ror.org/00a0n9e72"}]}
+{"DOI":"10.1161\/hc1202.104524","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/040r8fr65","Status":"active","Confidence":1},{"PID":"ROR","Value":"https:\/\/ror.org\/04fctr677","Status":"active","Confidence":1}]}
+{"DOI":"10.1021\/ma011134f","Matchings":[{"PID":"ROR","Value":"https:\/\/ror.org\/04tj63d06","Status":"active","Confidence":1}]}

dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/doi_to_ror_old.json

@@ -0,0 +1,9 @@
+{"DOI":"10.1061\/(asce)0733-9399(2002)128:7(759)","Matchings":[{"RORid":"https:\/\/ror.org\/03yxnpp24","Confidence":0.7071067812},{"RORid":"https:\/\/ror.org\/01teme464","Confidence":0.89}]}
+{"DOI":"10.1105\/tpc.8.3.343","Matchings":[{"RORid":"https:\/\/ror.org\/02k40bc56","Confidence":0.7071067812}]}
+{"DOI":"10.1161\/01.cir.0000013305.01850.37","Matchings":[{"RORid":"https:\/\/ror.org\/00qjgza05","Confidence":1}]}
+{"DOI":"10.1142\/s021821650200186x","Matchings":[{"RORid":"https:\/\/ror.org\/035xkbk20","Confidence":1},{"RORid":"https:\/\/ror.org\/05apxxy63","Confidence":1}]}
+{"DOI":"10.1061\/(asce)0733-9372(2002)128:7(575)","Matchings":[{"RORid":"https:\/\/ror.org\/04j198w64","Confidence":0.82}]}
+{"DOI":"10.1061\/(asce)0733-9372(2002)128:7(588)","Matchings":[{"RORid":"https:\/\/ror.org\/03m8km719","Confidence":0.8660254038},{"RORid":"https:\/\/ror.org\/02aze4h65","Confidence":0.87}]}
+{"DOI":"10.1161\/hy0202.103001","Matchings":[{"RORid":"https:\/\/ror.org\/057xtrt18","Confidence":0.7071067812}]}
+{"DOI": "10.1080/13669877.2015.1042504", "Matchings": [{"Confidence": 1.0, "RORid": "https://ror.org/03265fv13"}]}
+{"DOI": "https://doi.org/10.1007/3-540-47984-8_14", "Matchings": [{"Confidence": 1.0, "RORid": "https://ror.org/00a0n9e72"}]}

dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/publichers_old/publisher

@@ -0,0 +1,6 @@
+{"DOI": "10.1007/s00217-010-1268-9", "Authors": [{"Name": {"Full": "Martin Zarnkow", "First": null, "Last": null}, "Raw_affiliations": ["TU M\u00fcnchen, Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Andrea Faltermaier", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Werner Back", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Technologie der Brauerei I"], "Organization_PIDs": []}, {"Name": {"Full": "Martina Gastl", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Elkek K. Arendt", "First": null, "Last": null}, "Raw_affiliations": ["University College Cork"], "Organization_PIDs": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}
+{"DOI": "10.1007/BF01154707", "Authors": [{"Name": {"Full": "Buggy, M.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Materials Science and Technology, University of Limerick, Limerick, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/00a0n9e72", "Confidence": 1}]}, {"Name": {"Full": "Carew, A.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Materials Science and Technology, University of Limerick, Limerick, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/00a0n9e72", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/00a0n9e72", "Confidence": 1}]}
+{"DOI": "10.1007/s10237-017-0974-7", "Authors": [{"Name": {"Full": "Donnacha J. McGrath", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}]}, {"Name": {"Full": "Anja Lena Thiebes", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Christian G. Cornelissen", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Barry O\u2019Brien", "First": null, "Last": null}, "Raw_affiliations": ["Department for Internal Medicine \u2013 Section for Pneumology, Medical Faculty, RWTH Aachen University, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/04xfq0f34", "Confidence": 1}]}, {"Name": {"Full": "Stefan Jockenhoevel", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}]}, {"Name": {"Full": "Mark Bruzzi", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Peter E. McHugh", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}, {"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 1}]}
+{"DOI": "10.1007/BF03168973", "Authors": [{"Name": {"Full": "Sheehan, G.", "First": null, "Last": null}, "Raw_affiliations": ["Dept of Infectious Diseases, Mater Misercordiae Hospital, Dublin 7"], "Organization_PIDs": []}, {"Name": {"Full": "Chew, N.", "First": null, "Last": null}, "Raw_affiliations": ["Dept of Infectious Diseases, Mater Misercordiae Hospital, Dublin 7"], "Organization_PIDs": []}], "Organizations": []}
+{"DOI": "10.1007/s00338-009-0480-1", "Authors": [{"Name": {"Full": "Gleason, D. F.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biology, Georgia Southern University, Statesboro, USA"], "Organization_PIDs": [{"RORid": "https://ror.org/04agmb972", "Confidence": 1}]}, {"Name": {"Full": "Danilowicz, B. S.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biology, Georgia Southern University, Statesboro, USA"], "Organization_PIDs": [{"RORid": "https://ror.org/04agmb972", "Confidence": 1}]}, {"Name": {"Full": "Nolan, C. J.", "First": null, "Last": null}, "Raw_affiliations": ["School of Biology and Environmental Science, University College Dublin, Dublin 4, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/05m7pjf47", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/04agmb972", "Confidence": 1}, {"RORid": "https://ror.org/05m7pjf47", "Confidence": 1}]}
+{"DOI": "10.1007/s10993-010-9187-y", "Authors": [{"Name": {"Full": "Martin Howard", "First": null, "Last": null}, "Raw_affiliations": ["University College Cork"], "Organization_PIDs": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}

dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/actionmanager/bipaffiliations/publishers/publisher

@@ -1,6 +1,6 @@
-{"DOI": "10.1007/s00217-010-1268-9", "Authors": [{"Name": {"Full": "Martin Zarnkow", "First": null, "Last": null}, "Raw_affiliations": ["TU M\u00fcnchen, Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Andrea Faltermaier", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Werner Back", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Technologie der Brauerei I"], "Organization_PIDs": []}, {"Name": {"Full": "Martina Gastl", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Elkek K. Arendt", "First": null, "Last": null}, "Raw_affiliations": ["University College Cork"], "Organization_PIDs": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}
+{"DOI": "10.1007/s00217-010-1268-9", "Authors": [{"Name": {"Full": "Martin Zarnkow", "First": null, "Last": null}, "Raw_affiliations": ["TU M\u00fcnchen, Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Andrea Faltermaier", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Werner Back", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Technologie der Brauerei I"], "Organization_PIDs": []}, {"Name": {"Full": "Martina Gastl", "First": null, "Last": null}, "Raw_affiliations": ["Lehrstuhl f\u00fcr Brau- und Getr\u00e4nketechnologie"], "Organization_PIDs": []}, {"Name": {"Full": "Elkek K. Arendt", "First": null, "Last": null}, "Raw_affiliations": ["University College Cork"], "Organization_PIDs": [{"Value": "https://ror.org/03265fv13", "Confidence": 1}]}], "Organizations": [{"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/03265fv13", "Confidence": 1}]}
-{"DOI": "10.1007/BF01154707", "Authors": [{"Name": {"Full": "Buggy, M.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Materials Science and Technology, University of Limerick, Limerick, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/00a0n9e72", "Confidence": 1}]}, {"Name": {"Full": "Carew, A.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Materials Science and Technology, University of Limerick, Limerick, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/00a0n9e72", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/00a0n9e72", "Confidence": 1}]}
+{"DOI": "10.1007/BF01154707", "Authors": [{"Name": {"Full": "Buggy, M.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Materials Science and Technology, University of Limerick, Limerick, Ireland"], "Organization_PIDs": [{"Value": "https://ror.org/00a0n9e72", "Confidence": 1}]}, {"Name": {"Full": "Carew, A.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Materials Science and Technology, University of Limerick, Limerick, Ireland"], "Organization_PIDs": [{"Value": "https://ror.org/00a0n9e72", "Confidence": 1}]}], "Organizations": [{"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/00a0n9e72", "Confidence": 1}]}
-{"DOI": "10.1007/s10237-017-0974-7", "Authors": [{"Name": {"Full": "Donnacha J. McGrath", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}]}, {"Name": {"Full": "Anja Lena Thiebes", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Christian G. Cornelissen", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Barry O\u2019Brien", "First": null, "Last": null}, "Raw_affiliations": ["Department for Internal Medicine \u2013 Section for Pneumology, Medical Faculty, RWTH Aachen University, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/04xfq0f34", "Confidence": 1}]}, {"Name": {"Full": "Stefan Jockenhoevel", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}]}, {"Name": {"Full": "Mark Bruzzi", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Peter E. McHugh", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/03bea9k73", "Confidence": 1}, {"RORid": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 0.87}, {"RORid": "https://ror.org/04xfq0f34", "Confidence": 1}]}
+{"DOI": "10.1007/s10237-017-0974-7", "Authors": [{"Name": {"Full": "Donnacha J. McGrath", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"Value": "https://ror.org/03bea9k73", "Confidence": 1}]}, {"Name": {"Full": "Anja Lena Thiebes", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"Value": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"Value": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Christian G. Cornelissen", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"Value": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"Value": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Barry O\u2019Brien", "First": null, "Last": null}, "Raw_affiliations": ["Department for Internal Medicine \u2013 Section for Pneumology, Medical Faculty, RWTH Aachen University, Aachen, Germany"], "Organization_PIDs": [{"Value": "https://ror.org/04xfq0f34", "Confidence": 1}]}, {"Name": {"Full": "Stefan Jockenhoevel", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"Value": "https://ror.org/03bea9k73", "Confidence": 1}]}, {"Name": {"Full": "Mark Bruzzi", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biohybrid and Medical Textiles (BioTex), AME-Helmholtz Institute for Biomedical Engineering, ITA-Institut f\u00fcr Textiltechnik, RWTH Aachen University and at AMIBM Maastricht University, Maastricht, The Netherlands, Aachen, Germany"], "Organization_PIDs": [{"Value": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"Value": "https://ror.org/04xfq0f34", "Confidence": 0.87}]}, {"Name": {"Full": "Peter E. McHugh", "First": null, "Last": null}, "Raw_affiliations": ["Biomechanics Research Centre (BMEC), Biomedical Engineering, College of Engineering and Informatics, NUI Galway, Galway, Ireland"], "Organization_PIDs": [{"Value": "https://ror.org/03bea9k73", "Confidence": 1}]}], "Organizations": [{"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/03bea9k73", "Confidence": 1}, {"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/02jz4aj89", "Confidence": 0.82}, {"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/04xfq0f34", "Confidence": 0.87}, {"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/04xfq0f34", "Confidence": 1}]}
 {"DOI": "10.1007/BF03168973", "Authors": [{"Name": {"Full": "Sheehan, G.", "First": null, "Last": null}, "Raw_affiliations": ["Dept of Infectious Diseases, Mater Misercordiae Hospital, Dublin 7"], "Organization_PIDs": []}, {"Name": {"Full": "Chew, N.", "First": null, "Last": null}, "Raw_affiliations": ["Dept of Infectious Diseases, Mater Misercordiae Hospital, Dublin 7"], "Organization_PIDs": []}], "Organizations": []}
-{"DOI": "10.1007/s00338-009-0480-1", "Authors": [{"Name": {"Full": "Gleason, D. F.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biology, Georgia Southern University, Statesboro, USA"], "Organization_PIDs": [{"RORid": "https://ror.org/04agmb972", "Confidence": 1}]}, {"Name": {"Full": "Danilowicz, B. S.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biology, Georgia Southern University, Statesboro, USA"], "Organization_PIDs": [{"RORid": "https://ror.org/04agmb972", "Confidence": 1}]}, {"Name": {"Full": "Nolan, C. J.", "First": null, "Last": null}, "Raw_affiliations": ["School of Biology and Environmental Science, University College Dublin, Dublin 4, Ireland"], "Organization_PIDs": [{"RORid": "https://ror.org/05m7pjf47", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/04agmb972", "Confidence": 1}, {"RORid": "https://ror.org/05m7pjf47", "Confidence": 1}]}
+{"DOI": "10.1007/s00338-009-0480-1", "Authors": [{"Name": {"Full": "Gleason, D. F.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biology, Georgia Southern University, Statesboro, USA"], "Organization_PIDs": [{"Value": "https://ror.org/04agmb972", "Confidence": 1}]}, {"Name": {"Full": "Danilowicz, B. S.", "First": null, "Last": null}, "Raw_affiliations": ["Department of Biology, Georgia Southern University, Statesboro, USA"], "Organization_PIDs": [{"Value": "https://ror.org/04agmb972", "Confidence": 1}]}, {"Name": {"Full": "Nolan, C. J.", "First": null, "Last": null}, "Raw_affiliations": ["School of Biology and Environmental Science, University College Dublin, Dublin 4, Ireland"], "Organization_PIDs": [{"Value": "https://ror.org/05m7pjf47", "Confidence": 1}]}], "Organizations": [{"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/04agmb972", "Confidence": 1}, {"Provenance":"AffRo","PID":"ROR","Status":"active","Value": "https://ror.org/05m7pjf47", "Confidence": 1}]}
-{"DOI": "10.1007/s10993-010-9187-y", "Authors": [{"Name": {"Full": "Martin Howard", "First": null, "Last": null}, "Raw_affiliations": ["University College Cork"], "Organization_PIDs": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}], "Organizations": [{"RORid": "https://ror.org/03265fv13", "Confidence": 1}]}
+{"DOI": "10.1007/s10993-010-9187-y", "Authors": [{"Name": {"Full": "Martin Howard", "First": null, "Last": null}, "Raw_affiliations": ["University College Cork"], "Organization_PIDs": [{"Value": "https://ror.org/03265fv13", "Confidence": 1}]}], "Organizations": [{"PID":"ROR","Status":"active","Value": "https://ror.org/03265fv13", "Confidence": 1}]}