| Resumen |
Most frequent sense (MFS) is a very powerful heuristic in word sense disambiguation, extremely difficult to outperform with sophisticated methods. We show that counting the number of words, characters, or relationships of a word’s sense definitions allows guessing the most frequent sense of the word: the MFS usually has a longer gloss, more examples of usage, and more relationships with other words (synonyms, hyponyms, etc.). In addition, we show that this effect is resource-dependent, making some algorithms to perform differently with different dictionaries |
