1
0
mirror of https://github.com/mitb-archive/filebot synced 2024-08-13 17:03:45 -04:00

* faster and more accurate (in most cases) name matchings

This commit is contained in:
Reinhard Pointner 2009-07-05 19:17:03 +00:00
parent 955c83bfd9
commit 2a291d57c0

View File

@ -3,20 +3,19 @@ package net.sourceforge.filebot.similarity;
import static net.sourceforge.filebot.FileBotUtilities.*;
import uk.ac.shef.wit.simmetrics.similaritymetrics.AbstractStringMetric;
import uk.ac.shef.wit.simmetrics.similaritymetrics.MongeElkan;
import uk.ac.shef.wit.simmetrics.tokenisers.TokeniserQGram3Extended;
import uk.ac.shef.wit.simmetrics.similaritymetrics.QGramsDistance;
import uk.ac.shef.wit.simmetrics.tokenisers.TokeniserQGram3;
public class NameSimilarityMetric implements SimilarityMetric {
private final AbstractStringMetric metric;
public NameSimilarityMetric() {
// MongeElkan metric with a QGram3Extended tokenizer seems to work best for similarity of names
metric = new MongeElkan(new TokeniserQGram3Extended());
// QGramsDistance with a word tokenizer seems to work best for similarity of names
metric = new QGramsDistance(new TokeniserQGram3());
}