Consumer Behavior Prediction through Sentiment Analysis of Web-Data Source

Leveraging Sentiment Analysis of Web-Data Source to Predict Consumer Behavior

Abstract

Sentiment analysis is an emerging subject area, with the availability of wide variety of web data. This web data contains opinions, sentiments and emotions expressed by people and this can be quantified through use of machine learning techniques and data-mining approaches. In this paper, the fundamental steps involved in sentiment analysis and consumer behavior prediction are presented with emphasis on estimating or classifying the level of people’s opinions, sentiments towards a subject, product, service or individuals. Sentiment analysis involves data gathering, text pre-processing, feature extraction, sentiment classification and determining the polarity. Two major classification techniques namely, Baye’s Classifier and Support Vector Machine based classifier are presented along with their relative merits and demerits. This information extraction is of immense value to companies, marketing teams, sociologists and psychologists who are concerned with opinions, views, and public mood. Sentiment analysis has tremendous potential to change the way the business processes are carried out with an aim of enhanced customer satisfaction and business profitability.