Package org.nlpub.watset.graph

Graph processing and clustering.

Interface Summary

Interface	Description
Clustering<V>	An instance of Clustering returns clusters after running the underlying algorithm.
NodeWeighting<V,E>	Node weighting for Chinese Whispers.

Class Summary

Class	Description
ChineseWhispers<V,E>	Implementation of the Chinese Whispers algorithm.
ComponentsClustering<V,E>	A trivial clustering algorithm that treats every connected component as a cluster.
EmptyClustering<V>	A trivial clustering algorithm that returns no clusters.
MarkovClustering<V,E>	Naïve implementation of the Markov Clustering (MCL) algorithm.
MarkovClustering.NormalizeVisitor	Visitor that normalizes columns.
MarkovClusteringOfficial<V,E>	A wrapper for the official implementation of the Markov Clustering (MCL) algorithm in C.
MaxMax<V,E>	Implementation of the MaxMax soft clustering algorithm.
NodeWeighting.LabelNodeWeighting<V,E>	A trivial and not particularly useful node weighting approach that assigns the current node label as the weight.
NodeWeighting.LinearNodeWeighting<V,E>	The node weighting approach that chooses the label with the highest total edge weight in the neighborhood divided by the neighbor node degree.
NodeWeighting.LogNodeWeighting<V,E>	The node weighting approach that chooses the label with the highest total edge weight in the neighborhood divided by the logarithm of the neighbor node degree.
NodeWeighting.TopNodeWeighting<V,E>	The node weighting approach that chooses the label with the highest total edge weight in the neighborhood.
SenseInduction<V,E>	A simple graph-based word sense induction approach that clusters node neighborhoods.
SimplifiedWatset<V,E>	A faster and simplified version of Watset that does not need a context similarity measure.
SingletonClustering<V,E>	A trivial clustering algorithm that puts every node in a separate cluster.
TogetherClustering<V,E>	A trivial clustering algorithm that puts every node together in a single large cluster.
Watset<V,E>	Deprecated. use SimplifiedWatset instead.