Review Summarization using\nFeature-based Sentence Categorization

カテゴリ: 国際会議

論文No: MS3-2

グループ名: ACIS2015

発行日: 2015/10/15

著者名(英語): Teh Xin Xi (Universiti Sains Malaysia),Tan Zhiyu(Universiti Sains Malaysia), Wong Kean Yi(Universiti Sains Malaysia), Gan Keng Hoon(Universiti Sains Malaysia)

キーワード: Review, Summarization, Categorization,\nSentiment Analysis, Rating

要約(英語): When reading reviews, situation like too many comments often prevents a user from digesting the information efficiently. Hence, this research is motivated by the importance of improving the readability of large amount of review data based on common features related to entities like hotel, hand phone etc. With respect to this motivation, a feature-based sentence categorization approach is proposed. This approach incorporates three methods in processing (summarizing) reviews, i.e. entity detection, sentiment extraction and prediction rating. Entity detection focuses on the ways of detecting feature’s keywords using domain-specific dictionary while sentiment extraction provides basic rating data for prediction rating identification of adjective terms and assignment of sentiment value based on the nature of the adjective terms. As for prediction rating, a review-dependent approach is used by aggregating the rating based on each feature. In conclusion, the readability of reviews is improved by allowing user to select specific feature of an entity and thus minimizes the time needed to read reviews.

原稿種別: 英語

PDFファイルサイズ: 842 Kバイト