A Word Sense Disambiguation Method Based on Extended Conceptual Ontology Graph
Lei Li · 2004
This paper presents a word sense disambiguation algorithm based on extended conceptual ontology graph. The algorithm tries to find the branches of the paths converging to the same common ancestor nodes from two senses of the word to be disambiguated, by searching the crotch of the ancestor nodes of the two senses. The branches of the paths here are called the decisive paths for one sense against another. Then, with the help of the context words in the running text, it calculates each decisive path's support, compares each two senses and finally chooses the sense with top support of the decisive paths as the best suited sense. For evaluation and comparison, we test our algorithm on SemCor according with WordNet release 1.6. The result shows that our algorithm can disambiguate word sense with quite efficient speed and promising precision.