機械学習を活用したテキストマイニング : クチコミを用いた商品・サービスカテゴリーの横断分析

機関リポジトリ Web Site オープンアクセス

書誌事項

タイトル別名
  • Text Mining using Machine Learning : Cross-sectional Analysis of Products／Service Categories using Reviews
  • キカイ ガクシュウ オ カツヨウ シタ テキストマイニング : クチコミ オ モチイタ ショウヒン ・ サービスカテゴリー ノ オウダン ブンセキ

この論文をさがす

説明

In recent years, AlphaGo developed by google has become a topic not only in the world of Go but also in society, words related to AI such as machine learning and deep learning are becoming widely recognized. Advances in these technologies are quick, and discussion about medical assistance AI which applies image recognition technology is active, and it is expected to be applied in various fields even now. As an application possibility to management research and management practice, improvement of natural language processing technology in recent years, especially progressive technology concerning distributed representation, can be said to be a fully usable technology. And, it is not a direct competition relationship like a smartphone and a digital camera, www (world wide web) and a book, which is a threat of substitute in Porter’s 5force analysis, that is, competitive relations beyond product categories are topics. In this paper, we consider a recent social change and propose a method to analyze and visualize competing relationships beyond the categories of goods and services by machine mining using machine learning (AI) technology. The method of text mining adopted by this paper is not based on aggregate values such as mainstream current weighing text analysis but based on the distributed representation of words calculated by machine learning (fasttext). Therefore, the basis of the analysis is not the sum of the words in the text but the similarity using the distributed representation of the word. Text mining based on distributed representation is still an incomplete technology and it is not a fixed method, it is a field where further development is expected in the future. In this paper, we propose two analytical methods using this distributed representation. The method proposed by this paper is useful for business researchers, also for practitioners, when planning and model change of products/ services, comparison with other company’s product services becomes easier than ever, more detailed analysis.

収録刊行物

関連プロジェクト

もっと見る

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ