sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

bug fixing in the keywordsclustering class

This commit is contained in:
miconis 2019-07-08 11:01:49 +02:00
parent f5de20a508
commit 3c6f8d1e44
4 changed files with 25 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/KeywordsClustering.java
View File

@ -19,16 +19,17 @@ public class KeywordsClustering extends AbstractClusteringFunction {
@Override
protected Collection<String> doApply(String s) {
List<String> keywords = getCodes(s, translationMap, params.getOrDefault("windowSize", 4));
List<String> cities = getCodes(s, cityMap, params.getOrDefault("windowSize", 4));
//takes city codes and keywords codes without duplicates
Set<String> keywords = getCodes(s, translationMap, params.getOrDefault("windowSize", 4));
Set<String> cities = getCodes(s, cityMap, params.getOrDefault("windowSize", 4));
//list of combination to return as result
final Collection<String> combinations = new LinkedHashSet<String>();
int size = 0;
for (String keyword: keywords){
for (String city: cities) {
combinations.add(keyword+"-"+city);
if (++size>params.getOrDefault("max", 2)) {
if (combinations.size()>=params.getOrDefault("max", 2)) {
return combinations;
}
}

4
dnet-pace-core/src/main/java/eu/dnetlib/pace/common/AbstractPaceFunctions.java
View File

@ -320,7 +320,7 @@ public abstract class AbstractPaceFunctions {
}
//get the list of codes into the input string
public List<String> getCodes(String s1, Map<String, String> translationMap, int windowSize){
public Set<String> getCodes(String s1, Map<String, String> translationMap, int windowSize){
String s = cleanup(s1);
@ -328,7 +328,7 @@ public abstract class AbstractPaceFunctions {
List<String> tokens = Arrays.asList(s.toLowerCase().split(" "));
List<String> codes = new ArrayList<>();
Set<String> codes = new HashSet<>();
if (tokens.size()<windowSize)
windowSize = tokens.size();

7
dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
View File

@ -129,6 +129,13 @@ public class ClusteringFunctionTest extends AbstractPaceTest {
System.out.println(s1);
System.out.println(cf.apply(Lists.newArrayList(title(s1))));
final String s2 = "Universita farmaceutica culturale di milano bergamo";
System.out.println("s2 = " + s2);
System.out.println(cf.apply(Lists.newArrayList(title(s2))));
final String s3 = "universita universita milano milano";
System.out.println("s3 = " + s3);
System.out.println(cf.apply(Lists.newArrayList(title(s3))));
}

11
release.properties Normal file
View File

@ -0,0 +1,11 @@
#release configuration
#Mon Jul 08 10:03:15 CEST 2019
scm.tagNameFormat=@{project.artifactId}-@{project.version}
pushChanges=true
scm.url=scm\:git\:https\://github.com/dnet-team/dnet-dedup.git
preparationGoals=clean verify
projectVersionPolicyId=default
remoteTagging=true
scm.commentPrefix=[maven-release-plugin]
exec.snapshotReleasePluginAllowed=false
completedPhase=create-backup-poms