Title: A HYBRID CREDIBILITY ANALYSIS METHOD APPLIED ON TURKISH TWEETS WITH TV NEWS AND DISCUSSION PROGRAMS RELATED CONTENT

Issue Number: Vol. 6, No. 4
Year of Publication: Oct - 2016
Page Numbers: 281-291
Authors: Ali Fatih Gunduz, Pınar Karagöz
Journal Name: International Journal of Digital Information and Wireless Communications (IJDIWC)
- Hong Kong
DOI:  http://dx.doi.org/10.17781/P002153

Abstract:


In this paper, credibility analysis of microblog messages is studied. We collected our data from one of the most important microblogging services, Twitter. Our data set is created from the Turkish tweets written for weekly television programs broadcast in Turkey with political, cultural and/or financial contents. We adapted a new credibility definition based on three dimensions: being free from offensive words, not being spam and being newsworthy. To analyse credibility of tweets, we proposed a method by hybridizing supervised learning based techniques with graph based techniques. More- over we used content based and collaborative filtering based techniques in this hybrid method. The proposed method consists of two phases: super- vised learning phase and graph based improvement phase. Supervised classification algorithms are applied in the first phase and the results are improved in the graph based phase of the method. We focused on tweet-tweet, tweet-writer and writer-writer relations in the second phase of our study. The performance of the proposed method is measured by comparing it with human volunteers’ overall evaluation at the end.