Package org.nlpub.watset.eval

Class Pairwise<V>

java.lang.Object
org.nlpub.watset.eval.Pairwise<V>
Type Parameters:
V - the type of cluster elements
public class Pairwise<V> extends Object
Pairwise precision, recall, and F-score for cluster evaluation.
See Also:

  • Constructor Details

    • Pairwise

      public Pairwise()

  • Method Details

    • transform

      public static <V> Set<org.jgrapht.alg.util.Pair<V,V>> transform(Collection<? extends Collection<V>> clusters)
      Transform a collection of clusters to a collection of pairs generated using 2-combinations of the cluster elements.
      Type Parameters:
      V - the type of cluster elements
      Parameters:
      clusters - the collection of clusters
      Returns:
      a collection of pairs

    • combination

      public static <V> Stream<org.jgrapht.alg.util.Pair<V,V>> combination(Collection<V> cluster)
      Return a stream of pairs generated as 2-combinations of the cluster elements.
      Type Parameters:
      V - the type of cluster elements
      Parameters:
      cluster - the cluster
      Returns:
      a stream of 2-combinations

    • pairOf

      public static <V> org.jgrapht.alg.util.Pair<V,V> pairOf(V first, V second)
      Create a pair of elements ordered by hashCode.
      Type Parameters:
      V - the type of objects
      Parameters:
      first - the first object
      second - the second object
      Returns:
      a pair

    • evaluate

      public PrecisionRecall evaluate(Set<org.jgrapht.alg.util.Pair<V,V>> clusterPairs, Set<org.jgrapht.alg.util.Pair<V,V>> classPairs)
      Compute a pairwise precision, recall, and F-score.
      Parameters:
      clusterPairs - the cluster pairs to evaluate
      classPairs - the gold standard pairs to evaluate
      Returns:
      precision and recalled wrapped in an instance of PrecisionRecall

    • evaluate

      public PrecisionRecall evaluate(Collection<? extends Collection<V>> clusters, Collection<? extends Collection<V>> classes)
      Transform the clusters to pairs and compute a pairwise precision, recall, and F-score.
      Parameters:
      clusters - the collection of the clusters to evaluate
      classes - the collection of the gold standard clusters
      Returns:
      precision and recalled wrapped in an instance of PrecisionRecall