reimplementation of the sim between two authors. now it takes into account both name and surname. threshold incremented to 1.0 if the name is too short

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

@@ -20,12 +20,7 @@ public class AuthorMerger {
 
 	public static List<Author> merge(List<List<Author>> authors) {
 
-		authors.sort(new Comparator<List<Author>>() {
-			@Override
-			public int compare(List<Author> o1, List<Author> o2) {
-				return -Integer.compare(countAuthorsPids(o1), countAuthorsPids(o2));
-			}
-		});
+		authors.sort((o1, o2) -> -Integer.compare(countAuthorsPids(o1), countAuthorsPids(o2)));
 
 		List<Author> author = new ArrayList<>();
 
@@ -86,20 +81,28 @@ public class AuthorMerger {
 						.stream()
 						.map(ba -> new Tuple2<>(sim(ba, a._2()), ba))
 						.max(Comparator.comparing(Tuple2::_1));
-					if (simAuthor.isPresent() && simAuthor.get()._1() > THRESHOLD) {
+
+					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)
+							th = 0.99;
+
+						if (simAuthor.get()._1() > th) {
 							Author r = simAuthor.get()._2();
 							if (r.getPid() == null) {
 								r.setPid(new ArrayList<>());
 							}
 							r.getPid().add(a._1());
 						}
+					}
 				});
 	}
 
 	public static String pidToComparableString(StructuredProperty pid) {
-		return (pid.getQualifier() != null
-			? pid.getQualifier().getClassid() != null ? pid.getQualifier().getClassid().toLowerCase() : ""
-			: "") + (pid.getValue() != null ? pid.getValue().toLowerCase() : "");
+		return (pid.getQualifier() != null ?
+						pid.getQualifier().getClassid() != null ? pid.getQualifier().getClassid().toLowerCase() : "" : "")
+				+ (pid.getValue() != null ? pid.getValue().toLowerCase() : "");
 	}
 
 	public static int countAuthorsPids(List<Author> authors) {
@@ -120,12 +123,14 @@ public class AuthorMerger {
 		final Person pa = parse(a);
 		final Person pb = parse(b);
 
+		//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()));
+			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()));
 		}
 	}
 

dhp-workflows/dhp-dedup-openaire/src/test/java/eu/dnetlib/dhp/oa/dedup/EntityMergerTest.java

@@ -21,6 +21,7 @@ import scala.Tuple2;
 public class EntityMergerTest implements Serializable {
 
 	List<Tuple2<String, Publication>> publications;
+	List<Tuple2<String, Publication>> publications2;
 
 	String testEntityBasePath;
 	DataInfo dataInfo;
@@ -36,6 +37,7 @@ public class EntityMergerTest implements Serializable {
 			.getAbsolutePath();
 
 		publications = readSample(testEntityBasePath + "/publication_merge.json", Publication.class);
+		publications2 = readSample(testEntityBasePath + "/publication_merge2.json", Publication.class);
 
 		pub_top = getTopPub(publications);
 
@@ -90,6 +92,17 @@ public class EntityMergerTest implements Serializable {
 		assertEquals(AuthorMerger.countAuthorsPids(pub_merged.getAuthor()), 4);
 	}
 
+	@Test
+	public void publicationMergerTest2() throws InstantiationException, IllegalAccessException, IOException {
+
+		Publication pub_merged = DedupRecordFactory
+				.entityMerger(dedupId, publications2.iterator(), 0, dataInfo, Publication.class);
+
+		assertEquals(pub_merged.getAuthor().size(), 27);
+		// insert assertions here
+
+	}
+
 	public DataInfo setDI() {
 		DataInfo dataInfo = new DataInfo();
 		dataInfo.setTrust("0.9");