Release:2017, Vol. 3. №1
About the authors:Elena G. Brunova, Dr. Sci. (Philol.), Head of the Department of Foreign Languages and Cross-Cultural Professional Communication for Natural Sciences, Tyumen State University; firstname.lastname@example.org
The purpose of this research is to develop a rule-based algorithm for sentiment analysis. The dataset comprises the reviews in Russian on the bank service quality from clients’ bank rating, www.banki.ru. Sentiment analysis is considered as the classification task, i.e. matching a text with one of the two classes: with positive or negative polarity. The algorithm is based on the use of certain lexical and syntactic structures, along with the sentiment lexicon consisting of positive and negative lexicons, as well as three service classes. The efficiency of the proposed algorithm is estimated with Precision, Recall and F-measure in comparison with the results of another algorithm widely used for sentiment analysis — the Naive Bayes Classifier. To estimate the efficiency, the dataset of 200 reviews on the bank service quality is used. The values of Precision, Recall and F-measure for the proposed algorithm are 5-8% higher than for the Naïve Bayes Classifier.