Opinion Classification using Pattern Mining and Fuzzy Logic
Leela Subhashini, Yuefeng Li, Jinglang Zhang, Ajantha S. Athukorale · 2018
With the increasing amount of customer reviews on the web, there is a growing need for effective ways to retrieve valuable information hidden in customer reviews. Therefore, reviews are available online and reading, understanding and analysing the reviews has become a main activity of online customers. Real methods of mining and summarizing customer reviews are required for users to have in depth understanding of customer reviews. With the development of opinion mining or sentiment analysis research, automatically mining and summarizing customer reviews without any human assessment has become possible. However, most existing sentiment classification methods affected uncertain opinions. Implementing a classifier model which can cope with uncertainty will increase the accuracy of the classifier. In this paper, we propose a novel classifier model based on frequent pattern mining and fuzzy logic to cope with uncertain opinions using one feature selection method. The proposed model is evaluated on a benchmark data collection and compared with three baseline models. The experiment results indicate significant performance improvement. The results of the proposed model expose ability to decrease the uncertainty of opinions in customer reviews.