|
The documents, ontologies, semantic metrics. The most interesting is comparison of all of these regarding their importance to NLP.
This question is marked "community wiki".
|
|
Wikipedia analysis can lead to a number of benefits. For example, semantic-relatedness of words by exploiting Wikipedia is better captured as compared to approaches such as WordNet (see this paper). Wikipedia has also been used to give documents a representation in terms of "concepts", a feature used by the Explicit Semantic Analysis. In fact, using the hyperlink structure alone (without actual text) of Wikipedia has been shown to give pretty good results for capturing semantic relatedness of documents. Other benefits include word-sense disambiguation (this and this), coreference resolution, deriving taxonomies, ontologies, improving webpage clustering, cross-lingual mining, etc. to name a few. thanks very much, do you know some works where Wikipedia is used for text categorization? (Categorization to wikipedia categories)
(Aug 03 '10 at 09:21)
yura
Look at this paper (and also maybe the references therein): "Exploiting Wikipedia as external knowledge for document clustering" http://www.cis.drexel.edu/faculty/thu/research-papers/fp294-hu.pdf
(Aug 03 '10 at 23:38)
spinxl39
|
