tailieunhanh - A sentiment analyzer for informal text in social media

This paper introduces an approach to Twitter sentiment analysis, with the task of classifying tweets as positive, negative or neutral. In the preprocessing task, we propose a method to deal with two problems: (i) repeated characters in informal expression of words; and (ii) the affect of contrast word in determining sentence polarity. | Journal of Science and Technology 131 (2018) 006-012 A Sentiment Analyzer for Informal Text in Social Media Huong Thanh Le *, Nhan Trong Tran Hanoi University of Science and Technology - No. 1, Dai Co Viet, Hai Ba Trung, Hanoi, Viet Nam Received: May 05, 2018; Accepted: November 26, 2018 Abstract This paper introduces an approach to Twitter sentiment analysis, with the task of classifying tweets as positive, negative or neutral. In the preprocessing task, we propose a method to deal with two problems: (i) repeated characters in informal expression of words; and (ii) the affect of contrast word in determining sentence polarity. We propose features used in this task, investigate and select an optimal classifying algorithm among Decision Tree, K Nearest Neighbor, Support Vector Machine, and a Voting Classifier for solving Twitter sentiment analysis problem. Experiment results with Twitter 2016 test dataset shown that our system achieved good results ( F1-score) compared to related research in this field. Keywords: sentiment analysis, word embedding, decision tree, kNN, SVM, Voting Classifier 1. Introduction* For example, "4" can be understood as the number "four" or the preposition "for". Nowadays, social networking sites such as Facebook and Twitter become more and more popular with millions of users sharing either information or opinions about personalities, politicians, products, and events every day. They are valuable resources for business analysis, marketing, social analysis, etc. Because of that, Twitter sentiment analysis has received a lot of interest from research community. Examples below illustrate these difficulties: Example 1: Ha-ha. I want to see. E macdonalds here cheaper. Yum. Example 2: Ya. She wans. But now so late dunno still can arrange 4 tmr anot. The sentiment of Example 1 can be recognized as positive basing on words "want", "cheaper", "yum". Example 2 is harder to automatically analyze since it contains many informal words, .