Name
Abstract | ||
|---|---|---|
Pointwise mutual information (PMI) is a widely used word similarity measure, but it lacks a clear explanation of how it works. We explore how PMI differs from distributional similarity, and we introduce a novel metric, $({\rm PMI}_{max})$, that augments PMI with information about a word's number of senses. The coefficients of $({\rm PMI}_{max})$ are determined empirically by maximizing a utility function based on the performance of automatic thesaurus generation. We show that it outperforms traditional PMI in the application of automatic thesaurus generation and in two word similarity benchmark tasks: human similarity ratings and TOEFL synonym questions. $({\rm PMI}_{max})$ achieves a correlation coefficient comparable to the best knowledge-based approaches on the Miller-Charles similarity rating data set. |
Year | DOI | Venue |
|---|---|---|
2013 | 10.1109/TKDE.2012.30 | IEEE Trans. Knowl. Data Eng. |
Keywords | Field | DocType |
distributional similarity,automatic thesaurus generation,pointwise mutual information,human similarity rating,improving word similarity,word polysemy,word similarity measure,augments pmi,rm pmi,word similarity benchmark task,miller-charles similarity rating data,traditional pmi,augmenting pmi,semantics,mathematical model,metrics,automatic,correlation,semantic similarity,knowledge based systems,natural language processing,pmi,vectors,measurement | Data mining,Similarity measure,Computer science,Natural language processing,Artificial intelligence,Polysemy,Semantic similarity,Information retrieval,Synonym,Knowledge-based systems,Correlation,Pointwise mutual information,Semantics | Journal |
Volume | Issue | ISSN |
25 | 6 | 1041-4347 |
Citations | PageRank | References |
22 | 1.07 | 25 |
Authors | ||
5 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Lushan Han | 1 | 246 | 15.41 |
| Timothy W. Finin | 2 | 7345 | 821.22 |
| Paul McNamee | 3 | 38 | 3.50 |
| Anupam Joshi | 4 | 5708 | 480.12 |
| Yelena Yesha | 5 | 1756 | 253.96 |