Wordnet-based emotion words extraction considering sense distance

초록

Word emotion analysis is a basic process that recognizes emotions. Emotional words belonging to dialogs are classified into two classes-direct and potential emotion words. Direct emotion words can clearly express emotion, and potential emotion words may represent a certain emotion in a contextual manner. Thus, potential emotion words, unlike direct emotion words, are difficult to extract or precisely identify. In this paper, we propose a method for the extraction and identification of potential emotion words as well as direct emotion words based on WordNet. Potential emotion words can be extracted by measuring lexical affinity. We then consider the sense distance in order to minimize variations in meaning. Additionally, we suggest the maximum sense distance that limits searching space and can best extract the potential emotion words through experimentation. © 2013 International Information Institute.

키워드