Require help on accessing WordNet hierarchy for implementing algorithm

Question

I am trying to implement Resnik's similarity measure using WordNet hierarchy in java.

Example:- deposit money in bank. In above example suppose I want to disambiguate DEPOSIT & BANK. So, I will have to 1st access nodes of senses of respective words & then apply algorithm.

My question is:- How do I programatically access the WordNet hierarchy/node positions?

See related question/answers: http://stackoverflow.com/questions/5976537/wordnet-similarity-in-java-jaws-jwnl-or-java-wnsimilarity — Kenston Choi, Oct 28 '12 at 04:22
also see http://stackoverflow.com/questions/2827671/wordnet-java-interface?rq=1 — Qnan, Oct 28 '12 at 17:35
I understand you already know about JAWS. What's the problem then? — Qnan, Oct 28 '12 at 17:36
@Qnan, The reason is:-Say bank1=financial institution;bank2=slope of land near water body, etc. also deposit1=xxx, and deposit2=yyy(gloss). According to Resnik I have to first get the location of the nodes in order to compute the LCS and then only I can apply the algorithm. So LCS for bank1&deposit1 will be different compared to bank1&deposit2.. are you getting me? — variable, Oct 30 '12 at 04:49
You can use the Java WN Similarity project : http://www.sussex.ac.uk/Users/drh21/ — damned, Jun 09 '13 at 08:16

score 1 · Answer 1 · answered Mar 14 '13 at 11:50

dirty, not efficient, but obvious?: why don't you calculate all the possible pairs of glosses/senses for the given terms and keep the max similarity score (or before that the corresponding LCS min/max, do not know the details of the algorithm) given that the words are in the same context, e.g. phrase, sentence or paragraph level?

score 1 · Accepted Answer · answered Jan 18 '14 at 07:29

1

I have used the Java WN Similarity project as suggested by @damned

answered Jan 18 '14 at 07:29

variable

8,262
9
95
215

Require help on accessing WordNet hierarchy for implementing algorithm

2 Answers2