Using Adjusted Laplace Smoothing to Extract Implicit Aspects from Turkish Hotel Reviews

Omurca S., Ekinci E.

IEEE (SMC) International Conference on Innovations in Intelligent Systems and Applications (INISTA), Thessaloniki, Yunanistan, 3 - 05 Temmuz 2018 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/inista.2018.8466288
  • Basıldığı Şehir: Thessaloniki
  • Basıldığı Ülke: Yunanistan
  • Anahtar Kelimeler: aspect extraction, implicit aspect, sentiment analysis, laplace smoothing, graph, FEATURE IDENTIFICATION, CHINESE REVIEWS
  • Kocaeli Üniversitesi Adresli: Evet

Özet

With the rapid advances in the Web technologies, people have changed their way to obtain others' opinion via the online review websites. Due to increase in user generated content on these mediums an important source for sentiment analysis is provided. For more detailed analysis, aspect based sentiment analysis is getting increasingly important. And to complete this analysis implicit aspect extraction should be carried out. In this paper we suggested a graph based Laplace smoothing method for extracting implicit aspects from hotel reviews in Turkish. The proposed method is evaluated according to F-measure. Experimental results show that proposed method is robust and efficient in extracting implicit aspects.