[TL;DR] We borrow the idea of co-operation from game theory to model each word as an agent and utilize their co-operation to do word sense disambiguation in a sentence.

Word Sense Disambiguation: A Co-operative Game Theoretic Approach

Word Sense Disambiguation (WSD) is computationally identifying the real sense of the words in a given sentence. Applications such as Machine Translation, Speech Processing, Information Retrieval, Text Mining, and Content Analysis employ WSD in resolving the ambiguity of words in a given sentence.

Supervised learning methods have been successfully applied to the sense classification problem. Today, methods that train on manually sense-tagged corpora have become the mainstream approach to WSD. A significant problem with supervised approaches is the need for a sizable sense-tagged training set. Despite the availability of large corpora, manually sense-tagging of a corpus is very difficult, and very few sense-tagged data are available now. Additionally, a supervised system has problems adapting to different contexts, because it depends on prior knowledge which makes the algorithm rigid, therefore cannot efficiently adapt to domain-specific areas.

In game theory, cooperative or coalitional game is a type of game that has a competition between a group of players trying to be a part of the coalition that benefits them the most.

In this project we model WSD, using a cooperative game theory approach in formulating preference relations among the corpus, type of coalitions formed, and finding whether the outcomes under given coalition structures are optimal in resolving the ambiguity.