[4Rin1] Interactive 2

[4Rin1-50] Novel recommendation based on sentence similarity using the individual characteristic

〇Masato Maruyama¹, Takashi Takekawa¹ (1.Kogakuin University)

Keywords:Recommendation system, Natural language processing, Novel, Distributed Representations of words

The similarities of novels vary from person to person. Since it is difficult to find novels, it is important to assist with a recommendation system. In this study, we aim to estimate the similarity of novels by learning the texts of novels that individuals like. It was found that the word vectors learned by Word2vec were distributed together as words appearing in one novel. We use it to find the similarity of novels using the inter-group variance of the word vectors that appear in the novel. As a result, it was found that similarities between novels used in learning and other similarities were small.