A Quadratic 0-1 Programming Approach for Word Sense Disambiguation
This work addresses the problem of disambiguating word senses in natural language processing for applications like machine translation, but it is incremental as it builds on existing optimization methods without claiming broad SOTA.
The paper tackled Word Sense Disambiguation by modeling it as a combinatorial optimization problem, using a Quadratic 0-1 Integer Programming model to maximize sense-word similarity and sense-sense relatedness, achieving results that address inter-sense interactions but without concrete performance numbers provided.
Word Sense Disambiguation (WSD) is the task to determine the sense of an ambiguous word in a given context. Previous approaches for WSD have focused on supervised and knowledge-based methods, but inter-sense interactions patterns or regularities for disambiguation remain to be found. We argue the following cause as one of the major difficulties behind finding the right patterns: for a particular context, the intended senses of a sequence of ambiguous words are dependent on each other, i.e. the choice of one word's sense is associated with the choice of another word's sense, making WSD a combinatorial optimization problem.In this work, we approach the interactions between senses of different target words by a Quadratic 0-1 Integer Programming model (QIP) that maximizes the objective function consisting of (1) the similarity between candidate senses of a target word and the word in a context (the sense-word similarity), and (2) the semantic interactions (relatedness) between senses of all words in the context (the sense-sense relatedness).