The web interface uses a configuration file that looks like this: WordNet::Similarity::lesk stem::1 stop::stoplist.txt
The stoplist is the default stoplist that comes with WordNet::Similarity. Using a stoplist makes a big impact on the scores produced by lesk because it eliminates a lot of the (arguably) irrelevant matches on words like 'of', 'it', 'the', etc. Stemming will also have an impact on the scores; although, it think the use of a stoplist will have a much more noticeable impact. The relations used in the web interface is the default for lesk. See the end of this email for the default relation file. On a related note, the vector measure in the web interface also uses a configuration file, which is required for the vector measure. The file looks like this: WordNet::Similarity::vector stop::stoplist.txt stem::1 compounds::wn20compounds.txt vectordb::$wordvectors The stoplist is the same as the one used for lesk. The compounds file is the one that comes with WordNet::Similarity in the samples directory. The vectordb file was generated using wordVectors.pl, which comes with WordNet::Similarity (a vectordb file is required for the vector measure). Thanks, Jason Here is the Lesk relation file: LeskRelationFile also-also also-attr also-glos also-holo also-hype also-hypo also-mero also-pert also-sim attr-also attr-attr attr-glos attr-holo attr-hype attr-hypo attr-mero attr-pert attr-sim example-example example-glos example-syns glos-also glos-attr glos-example glos-glos glos-holo glos-hype glos-hypo glos-mero glos-pert glos-sim glos-syns holo-also holo-attr holo-glos holo-holo holo-hype holo-hypo holo-mero holo-pert holo-sim hype-also hype-attr hype-glos hype-holo hype-hype hype-hypo hype-mero hype-pert hype-sim hypo-also hypo-attr hypo-glos hypo-holo hypo-hype hypo-hypo hypo-mero hypo-pert hypo-sim mero-also mero-attr mero-glos mero-holo mero-hype mero-hypo mero-mero mero-pert mero-sim pert-also pert-attr pert-glos pert-holo pert-hype pert-hypo pert-mero pert-pert pert-sim sim-also sim-attr sim-glos sim-holo sim-hype sim-hypo sim-mero sim-pert sim-sim syns-example syns-glos -- http://www.d.umn.edu/~mich0212/ Yahoo! Groups Links <*> To visit your group on the web, go to: http://groups.yahoo.com/group/wn-similarity/ <*> To unsubscribe from this group, send an email to: [EMAIL PROTECTED] <*> Your use of Yahoo! Groups is subject to: http://docs.yahoo.com/info/terms/
