본문 바로가기
  • Home

A Study on Automatic Classification of Movie Review Based on Sensibility Evaluation Vocabulary

  • Korean Semantics
  • 2018, 59(), pp.31-54
  • Publisher : The Society Of Korean Semantics
  • Research Area : Humanities > Korean Language and Literature
  • Published : March 30, 2018

Park, Keon-Sook 1

1상명대학교

Accredited

ABSTRACT

The purpose of this study is to suggest that the frequency of use of emotion evaluation vocabulary and the evaluation attitude information can be applied to automatic classification of movie review. To this end, I analyzed the emotion evaluation vocabulary in the movie review and measured the word usage frequency according to the polarity of the evaluation attitude. In this process, the natural logarithm is used for the evaluation index score of each word. The results were as follows: First, as a result of automatic classification, the success rate was 55% and the failure rate was 30%. The success probability of automatic classification increased as the movie rating was higher or the length of the movie was longer. Second, it was found that weights should be given according to the semantic domain in the emotion evaluation vocabulary. If the evaluation index is increased for words that have the meaning of polarity, the accuracy of automatic classification will be improved. Based on this research, I would like to apply it to 'opinion classification' of news comment or product review.

Citation status

* References for papers published after 2022 are currently being built.