SENTIMENT CLASSIFICATION OF THE LOCAL VISITORS’ SOCIAL MEDIA REVIEWS

Author:

Year-Number: 2018-26
Yayımlanma Tarihi: null
Language : null
Konu :
Number of pages: 5534-5538
Mendeley EndNote Alıntı Yap

Abstract

Bu çalışmanın amacı, tarihi ve turistik bölgelere yaptığı ziyaret sonrası sosyal medya üzerinden değerlendirmelerini paylaşan yerli ziyaretçilerin yorumlarının sınıflandırılmasıdır. Bu çerçevede, makine öğrenmesi yöntemlerinden denetimli öğrenme yaklaşımı kullanılarak sosyal medyada duygu analizi çalışması yapılmıştır. Analiz için Türkiye’nin turizm için önemli görülen iki turizm destinasyonu İstanbul ve Antalya’ya yönelik yerli ziyaretçilerin yorumları TripAdvisor.com’dan alınmıştır. Yorumlar olumlu, olumsuz ve nötr olarak işaretlenerek sınıflandırılmıştır. Çalışmada olumlu, olumsuz ve nötr sınıftaki veri dağılımı KNIME Programı’nda işlenerek denetimli öğrenme modellerinden karar ağaçları öğrenme yöntemiyle sınanmıştır. Denetimli öğrenme modellerinden Karar Ağaçları Öğrenme testi sonucunda başarı oranı %75,7 olarak bulunmuştur.

Keywords

Abstract

The aim of this study is to classify the local visitors’ reviews shared on social media after their visits to historical and touristic regions. In this context, sentiment analysis was conducted for social media reviews by using supervised learning approach from machine learning methods. For the analysis, the local visitors’ reviews, on Istanbul and Antalya, which are Turkey’s two important tourism cities, were imported from TripAdvisor website. The reviews were classified as positive, negative, and neutral. In the study, the data distribution was processed in positive, negative, and neutral classifications with KNIME software, and then it was tested with the learning method of decision trees from the supervised learning models. As a result of the decision trees learning test that is one of the supervised learning models, 75.7% was found as the success rate.

Keywords