Destek Vektör Makinesi Yöntemi ile Bir Duygu Çözümlemesi Evrim Kasaba1, Engin Yıldıztepe1

Anahtar Sözcükler: Duygu çözümleme, destek vektör makinesi, Twitter
A Sentiment Analysis Study Using Support Vector Machine
Abstract: Sentiment analysis is a popular area which uses natural language processing, statistical learning and text mining. It can also be named as “Thought Analysis” and “Opinion Mining”. Sentiment analysis is a problem of text classification as well as an attractive study field due to its popularity and commercial gains. This field has found more attention since the need for automatic classification of the comments given in social networks; web blogs and shopping websites, has greatly increased recently. It is always a significant information for many persons and commercial companies, producers and politicians what sentiment their names and events may arouse in people. Before the internet and the social media turned out to be widespread, it was extremely hard and expensive to obtain such data. However, today various platforms where people can share their feelings and ideas have become important sources of information for the researches on this field. There are two main approaches in sentiment analysis: dictionary based and statistical based approaches. Dictionary based approaches are semi-supervised approaches which use a semantic dictionary database in sentiment analysis procedures. On the other hand, statistical and machine learning approaches are supervised methods using labeled training data for learning. Support vector machine is one of these methods. In this study, a brief information about sentiment analysis is given and an application using support vector machine is presented. Twitter data are used in the application. This study is conducted with R statistical programming language.