|
1
|
package eu.dnetlib.pace.distance.algo;
|
|
2
|
|
|
3
|
import java.util.HashSet;
|
|
4
|
import java.util.Map;
|
|
5
|
import java.util.Set;
|
|
6
|
|
|
7
|
import eu.dnetlib.pace.distance.ConfigurableDistanceAlgo;
|
|
8
|
import eu.dnetlib.pace.distance.DistanceAlgo;
|
|
9
|
import org.apache.commons.logging.Log;
|
|
10
|
import org.apache.commons.logging.LogFactory;
|
|
11
|
|
|
12
|
import com.google.common.base.Function;
|
|
13
|
import com.google.common.base.Predicates;
|
|
14
|
import com.google.common.collect.Iterables;
|
|
15
|
import com.google.common.collect.Sets;
|
|
16
|
import com.google.common.collect.Sets.SetView;
|
|
17
|
|
|
18
|
import eu.dnetlib.pace.model.Field;
|
|
19
|
import eu.dnetlib.pace.model.FieldListImpl;
|
|
20
|
import eu.dnetlib.pace.model.gt.CoAuthor;
|
|
21
|
import eu.dnetlib.pace.model.gt.CoAuthors;
|
|
22
|
import eu.dnetlib.pace.model.gt.GTAuthor;
|
|
23
|
|
|
24
|
public class PersonCoAuthorSurnamesDistance extends ConfigurableDistanceAlgo implements DistanceAlgo {
|
|
25
|
|
|
26
|
/**
|
|
27
|
* logger.
|
|
28
|
*/
|
|
29
|
private static final Log log = LogFactory.getLog(PersonCoAuthorSurnamesDistance.class); // NOPMD by marko on 11/24/08 5:02 PM
|
|
30
|
|
|
31
|
public PersonCoAuthorSurnamesDistance(final Map<String, String> params, final double weight) {
|
|
32
|
super(params, weight);
|
|
33
|
}
|
|
34
|
|
|
35
|
@Override
|
|
36
|
public double distance(final Field a, final Field b) {
|
|
37
|
|
|
38
|
final GTAuthor ga = Iterables.getOnlyElement(GTAuthor.fromOafJson(((FieldListImpl) a).stringList()));
|
|
39
|
final GTAuthor gb = Iterables.getOnlyElement(GTAuthor.fromOafJson(((FieldListImpl) b).stringList()));
|
|
40
|
|
|
41
|
final Integer commonSurnames = Integer.parseInt(getParams().get("common.surnames"));
|
|
42
|
// log.info("min common surnames: " + commonSurnames);
|
|
43
|
if (isEnabled(commonSurnames) && (surnamesInCommon(ga, gb).size() >= commonSurnames)) return 1.0;
|
|
44
|
|
|
45
|
return 0;
|
|
46
|
}
|
|
47
|
|
|
48
|
private boolean isEnabled(final Integer property) {
|
|
49
|
return (property != null) && (property >= 0);
|
|
50
|
}
|
|
51
|
|
|
52
|
private Set<String> surnamesInCommon(final GTAuthor a, final GTAuthor b) {
|
|
53
|
final SetView<String> intersection = Sets.intersection(getSurnames(a.getCoAuthors()), getSurnames(b.getCoAuthors()));
|
|
54
|
// log.info(String.format("surnames intersection '%s' - '%s': %s", a.getAuthor().getFullname(), b.getAuthor().getFullname(),
|
|
55
|
// intersection.size()));
|
|
56
|
return intersection;
|
|
57
|
}
|
|
58
|
|
|
59
|
private Set<String> getSurnames(final CoAuthors ca) {
|
|
60
|
if ((ca == null) || ca.isEmpty()) return new HashSet<String>();
|
|
61
|
return Sets.newHashSet(Iterables.filter(Iterables.transform(ca, new Function<CoAuthor, String>() {
|
|
62
|
|
|
63
|
@Override
|
|
64
|
public String apply(final CoAuthor c) {
|
|
65
|
return c.getSecondnames();
|
|
66
|
}
|
|
67
|
}), Predicates.notNull()));
|
|
68
|
}
|
|
69
|
|
|
70
|
@Override
|
|
71
|
public double getWeight() {
|
|
72
|
return getWeigth();
|
|
73
|
}
|
|
74
|
|
|
75
|
}
|