54 lines
1.3 KiB
Java
54 lines
1.3 KiB
Java
|
|
package net.filebot.similarity;
|
|
|
|
|
|
import static net.filebot.similarity.Normalization.*;
|
|
import uk.ac.shef.wit.simmetrics.similaritymetrics.AbstractStringMetric;
|
|
import uk.ac.shef.wit.simmetrics.similaritymetrics.QGramsDistance;
|
|
import uk.ac.shef.wit.simmetrics.tokenisers.TokeniserQGram3;
|
|
|
|
import com.ibm.icu.text.Transliterator;
|
|
|
|
|
|
public class NameSimilarityMetric implements SimilarityMetric {
|
|
|
|
private final AbstractStringMetric metric;
|
|
private final Transliterator transliterator;
|
|
|
|
|
|
public NameSimilarityMetric() {
|
|
// QGramsDistance with a QGram tokenizer seems to work best for similarity of names
|
|
this(new QGramsDistance(new TokeniserQGram3()), Transliterator.getInstance("Any-Latin;Latin-ASCII;[:Diacritic:]remove"));
|
|
}
|
|
|
|
|
|
public NameSimilarityMetric(AbstractStringMetric metric, Transliterator transliterator) {
|
|
this.metric = metric;
|
|
this.transliterator = transliterator;
|
|
}
|
|
|
|
|
|
@Override
|
|
public float getSimilarity(Object o1, Object o2) {
|
|
return metric.getSimilarity(normalize(o1), normalize(o2));
|
|
}
|
|
|
|
|
|
protected String normalize(Object object) {
|
|
// use string representation
|
|
String name = object.toString();
|
|
|
|
// apply transliterator
|
|
if (transliterator != null) {
|
|
name = transliterator.transform(name);
|
|
}
|
|
|
|
// normalize separators
|
|
name = normalizePunctuation(name);
|
|
|
|
// normalize case and trim
|
|
return name.toLowerCase();
|
|
}
|
|
|
|
}
|