We introduce two different approaches for clustering semantically similar
words. We accommodate ambiguity by allowing a word to belong to several
clusters.
Both methods use a graph-theoretic representation of words and their
paradigmatic relationships. The first approach is based on the concept of
curvature and divides the word graph into classes of similar words by removing
words of low curvature which connect several dispersed clusters.
The second method, instead of clustering the nodes, clusters the links in our
graph. These contain more specific contextual information than nodes
representing just words. In so doing, we naturally accommodate ambiguity by
allowing multiple class membership.
Both methods are evaluated on a lexical acquisition task, using clustering to
add nouns to the WordNet taxonomy. The most effective method is link
clustering.
Description
Using Curvature and Markov Clustering in Graphs for Lexical Acquisition
and Word Sense Discrimination
%0 Generic
%1 dorow2004using
%A Dorow, Beate
%A Widdows, Dominic
%A Ling, Katarina
%A Eckmann, Jean-Pierre
%A Sergi, Danilo
%A Moses, Elisha
%D 2004
%K clustering curvature
%T Using Curvature and Markov Clustering in Graphs for Lexical Acquisition
and Word Sense Discrimination
%U http://arxiv.org/abs/cond-mat/0403693
%X We introduce two different approaches for clustering semantically similar
words. We accommodate ambiguity by allowing a word to belong to several
clusters.
Both methods use a graph-theoretic representation of words and their
paradigmatic relationships. The first approach is based on the concept of
curvature and divides the word graph into classes of similar words by removing
words of low curvature which connect several dispersed clusters.
The second method, instead of clustering the nodes, clusters the links in our
graph. These contain more specific contextual information than nodes
representing just words. In so doing, we naturally accommodate ambiguity by
allowing multiple class membership.
Both methods are evaluated on a lexical acquisition task, using clustering to
add nouns to the WordNet taxonomy. The most effective method is link
clustering.
@misc{dorow2004using,
abstract = {We introduce two different approaches for clustering semantically similar
words. We accommodate ambiguity by allowing a word to belong to several
clusters.
Both methods use a graph-theoretic representation of words and their
paradigmatic relationships. The first approach is based on the concept of
curvature and divides the word graph into classes of similar words by removing
words of low curvature which connect several dispersed clusters.
The second method, instead of clustering the nodes, clusters the links in our
graph. These contain more specific contextual information than nodes
representing just words. In so doing, we naturally accommodate ambiguity by
allowing multiple class membership.
Both methods are evaluated on a lexical acquisition task, using clustering to
add nouns to the WordNet taxonomy. The most effective method is link
clustering.},
added-at = {2017-02-02T18:18:57.000+0100},
author = {Dorow, Beate and Widdows, Dominic and Ling, Katarina and Eckmann, Jean-Pierre and Sergi, Danilo and Moses, Elisha},
biburl = {https://www.bibsonomy.org/bibtex/2e5b7100767cca0cf7062a6c7824b1cf0/l3s},
description = {Using Curvature and Markov Clustering in Graphs for Lexical Acquisition
and Word Sense Discrimination},
interhash = {348a42627b0716fe5987ce7d520d0940},
intrahash = {e5b7100767cca0cf7062a6c7824b1cf0},
keywords = {clustering curvature},
note = {cite arxiv:cond-mat/0403693},
timestamp = {2017-02-02T18:18:57.000+0100},
title = {Using Curvature and Markov Clustering in Graphs for Lexical Acquisition
and Word Sense Discrimination},
url = {http://arxiv.org/abs/cond-mat/0403693},
year = 2004
}