Sentiment classification with current intelligent optimization algorithms

In recent years, social networks have become an important part of daily life and are becoming increasingly important. Network analysis is an approach that examines social structure with the assumption that the best way to examine a social system is to examine the links between actors. Nowadays, it is used in many areas such as health, education and academic. The active use of social network has revealed requirement of social network analysis. Sentiment Analysis - Opinion Mining which is one of the most well-known social network analysis problems, is a field of study that analyzes people's opinions, sentiments, evaluations, ratings, attitudes, and feelings about assets such as products, services, organizations, individuals, subjects, activities, titles, and qualifications. In this thesis, sentiment analysis is considered as an optimization problem for the first time. Whale Optimization Algorithm and Social Impact Theory based Optimization Algorithm which are one of the current intelligent optimization algorithms, have been adapted for the sentiment analysis problem. Memory feature has been integrated into Social Impact Theory based Optimization Algorithm in order to obtain effective results. IMDB, Polarity, and Amazon datasets were used in this study. In order to evaluate the performance of the results, accuracy percentage, precision, recall, F-Measure, and MCC which are the five most commonly used measure in the literature were used. The results were compared with thirty-three supervised learning algorithms. When the results were examined, it was seen that metaheuristic optimization algorithms gave successful results in sentiment analysis.