1 |
41642
|
claudio.at
|
package eu.dnetlib.pace.distance.algo;
|
2 |
37195
|
claudio.at
|
|
3 |
|
|
import com.wcohen.ss.AbstractStringDistance;
|
4 |
41642
|
claudio.at
|
import eu.dnetlib.pace.distance.SecondStringDistanceAlgo;
|
5 |
37195
|
claudio.at
|
|
6 |
|
|
public class LevensteinTitle extends SecondStringDistanceAlgo {
|
7 |
|
|
|
8 |
|
|
public LevensteinTitle(final double w) {
|
9 |
|
|
super(w, new com.wcohen.ss.Levenstein());
|
10 |
|
|
}
|
11 |
|
|
|
12 |
|
|
protected LevensteinTitle(final double w, final AbstractStringDistance ssalgo) {
|
13 |
|
|
super(w, ssalgo);
|
14 |
|
|
}
|
15 |
|
|
|
16 |
|
|
@Override
|
17 |
|
|
public double distance(final String a, final String b) {
|
18 |
|
|
final String ca = cleanup(a);
|
19 |
|
|
final String cb = cleanup(b);
|
20 |
|
|
|
21 |
|
|
final boolean check = checkNumbers(ca, cb);
|
22 |
|
|
|
23 |
|
|
if (check) return 0.5;
|
24 |
|
|
|
25 |
|
|
final String cca = finalCleanup(ca);
|
26 |
|
|
final String ccb = finalCleanup(cb);
|
27 |
|
|
|
28 |
|
|
return normalize(ssalgo.score(cca, ccb), cca.length(), ccb.length());
|
29 |
|
|
}
|
30 |
|
|
|
31 |
|
|
private double normalize(final double score, final int la, final int lb) {
|
32 |
|
|
return 1 - (Math.abs(score) / Math.max(la, lb));
|
33 |
|
|
}
|
34 |
|
|
|
35 |
|
|
@Override
|
36 |
|
|
public double getWeight() {
|
37 |
|
|
return super.weight;
|
38 |
|
|
}
|
39 |
|
|
|
40 |
|
|
@Override
|
41 |
|
|
protected double normalize(final double d) {
|
42 |
|
|
return 1 / Math.pow(Math.abs(d) + 1, 0.1);
|
43 |
|
|
}
|
44 |
|
|
|
45 |
|
|
}
|