Project

General

Profile

1 41642 claudio.at
package eu.dnetlib.pace.distance.algo;
2 38600 claudio.at
3
import java.util.HashSet;
4
import java.util.Map;
5
import java.util.Set;
6
7
import com.google.common.base.Function;
8
import com.google.common.base.Predicates;
9
import com.google.common.collect.Iterables;
10
import com.google.common.collect.Sets;
11
import com.google.common.collect.Sets.SetView;
12 41642 claudio.at
import eu.dnetlib.pace.distance.ConfigurableDistanceAlgo;
13
import eu.dnetlib.pace.distance.DistanceAlgo;
14 38600 claudio.at
import eu.dnetlib.pace.model.Field;
15
import eu.dnetlib.pace.model.FieldListImpl;
16
import eu.dnetlib.pace.model.gt.CoAuthor;
17
import eu.dnetlib.pace.model.gt.CoAuthors;
18
import eu.dnetlib.pace.model.gt.GTAuthor;
19 40910 claudio.at
import org.apache.commons.logging.Log;
20
import org.apache.commons.logging.LogFactory;
21 38600 claudio.at
22
public class PersonCoAnchorsDistance extends ConfigurableDistanceAlgo implements DistanceAlgo {
23
24
	/**
25
	 * logger.
26
	 */
27
	private static final Log log = LogFactory.getLog(PersonCoAnchorsDistance.class); // NOPMD by marko on 11/24/08 5:02 PM
28
29
	public PersonCoAnchorsDistance(final Map<String, String> params, final double weight) {
30
		super(params, weight);
31
	}
32
33
	@Override
34
	public double distance(final Field a, final Field b) {
35
36 41504 claudio.at
		final GTAuthor ga = Iterables.getLast(GTAuthor.fromOafJson(((FieldListImpl) a).stringList()));
37
		final GTAuthor gb = Iterables.getLast(GTAuthor.fromOafJson(((FieldListImpl) b).stringList()));
38 38600 claudio.at
39
		final Integer commonAnchors = Integer.parseInt(getParams().get("common.anchors"));
40
		// log.info("min common anchors: " + commonAnchors);
41
		if (isEnabled(commonAnchors) && (anchorsInCommon(ga, gb).size() >= commonAnchors)) // log.info("matched coauthor intersection");
42
			return 1.0;
43
44
		return 0;
45
	}
46
47
	private boolean isEnabled(final Integer property) {
48
		return (property != null) && (property >= 0);
49
	}
50
51
	private Set<String> anchorsInCommon(final GTAuthor a, final GTAuthor b) {
52
		final SetView<String> intersection = Sets.intersection(getAnchorIds(a.getCoAuthors()), getAnchorIds(b.getCoAuthors()));
53 40910 claudio.at
		//log.info(String.format("anchors intersection '%s' - '%s': %s", a.getAuthor().getFullname(), b.getAuthor().getFullname(), intersection.size()));
54 38600 claudio.at
		return intersection;
55
	}
56
57
	private Set<String> getAnchorIds(final CoAuthors ca) {
58
		if ((ca == null) || ca.isEmpty()) return new HashSet<String>();
59
		final Iterable<String> anchorIds = Iterables.transform(ca, new Function<CoAuthor, String>() {
60
61
			@Override
62
			public String apply(final CoAuthor c) {
63
				return c.getAnchorId();
64
			}
65
		});
66
		final Iterable<String> filtered = Iterables.filter(anchorIds, Predicates.notNull());
67
		return Sets.newHashSet(filtered);
68
	}
69
70
	@Override
71
	public double getWeight() {
72
		return getWeigth();
73
	}
74
75
}