Word Sense Disambiguation in Contextual Dynamic Network Using Associative Concept Dictionary

Many of the Japanese ideographs (Chi- nese characters) have a few meanings. They should be disambiguated by using their contextual information. Some of the ideographs have a few different pronun- ciations depending on their meanings. For example, we have an ideograph which has two pronunciations, /hitai/ and /gaku/, the former means a forehead of the human body and the latter means an amount of money or a picture frame. Conventional methods for such disam- biguation have been using statistical me- thods with co-occurrence of words in their context. In this research, Contextual Dynamic Network Model is developed using the Associative Concept Dictionary, which includes semantic relations among concepts/words and the relations are rep- resented with qualitative distances. In this model, an interactive activation model is used in the contextual semantic network where the activation in the net- work is calculated using the distances. The proposed method can disambiguate word senses by constructing dynamically the contextual semantic network accord- ing to the input text including an am- biguous word. The interactive activation model based on the contextual semantic network is evaluated by using a Japanese newspaper, Mainichi shinbun. Its effec- tiveness is shown by comparison with a Naive Bayes method.