主催: バイオメディカル・ファジィ・システム学会
会議名: 第30回バイオメディカル・ファジィ・システム学会
回次: 30
開催地: 大阪
開催日: 2017/11/25 - 2017/11/26
p. 20-23
In the sentiment analysis, we usually classify costumer reviews into some classes based on their topics. Technical term dictionary is very important to classify them into correct classes by some classifier such as support vector machine (SVM) learned by the customer review data. In other previous researches, the technical term dictionaries have been constructed by high frequency words in the same topic data. In this paper, we propose a method to construct a high quality technical term dictionary by extracting relevant words and their distributed representation for each topic semi-automatically and we estimated an effectiveness of our proposed method using experimental results.