Using Sentiment Analysis to Detect Customer Attitudes in Social Media Comments

Abstract

ig data analytics is increasingly replacing or complementing classical data collection methods like surveys. Business intelligence tools and systems now play a key role in decision-making and find applications in the areas such as customer profiling, customer support, market research, segmentation, brand monitoring. Transforming unstructured text data for structured quantitative analysis is an important challenge for business analytics. Sentiment analysis is a promising method in terms of transforming words into numbers to detect the tone of subjective expressions. The aim of this paper is scrutinize the role of sentiment analysis for emulating classical attitude and market research tools. We used sentiment analysis on product reviews data occurring in the e-market place in Turkey. We have applied a variety of machine learning algorithms and some term selection algorithms. Unigram approach for the term selection and Naive Bayes approach for machine learning have performed better than others. Our results suggest that sentiment analysis can be applicable for Turkish language after a rigorous text preprocessing and term selection process.