Opinion Mining and Fuzzy Quantification in Hotel Reviews


Dundar B., Ozdemir S., AKAY D.

International Symposium on Networks, Computers and Communications (ISNCC), Hammamet, Tunus, 11 - 13 Mayıs 2016 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/isncc.2016.7746066
  • Basıldığı Şehir: Hammamet
  • Basıldığı Ülke: Tunus

Özet

In this study, fuzzy quantification sentences are used to create short summaries from hotel reviews. Also, opinion mining is used to extract opinion expressions from reviews. Fuzzy quantified sentences offer a brief information about the hotel attributes from customers feedback. These sentences are generated considering one type of fuzzy sets (triangular) using 3 types of quantifiers ("most", "about half" or "a few"). The degree of truth of generated summaries is calculated according to occurrence of attributes and opinions. Opinion mining extracts positive, negative and neutral emoticons from reviews. The extracted opinions and short fuzzy quantified sentences have the characteristics of presenting novel ideas for hotel recommendation.