Predict user mood according to facebook postings

Abstract

We live in an era where communication is growing fast in the cyberspace. As a part of this people tend to be online 24 hours of a day and they write postings in social media. The interesting point is people may put a social mask to hide their feelings in the real world, but they reveal it on their post unknowingly. With this context discussion regarding opinion, mining has dominated research in recent years. Because of the ambiguity of human language, it is difficult to extract the sentiment precisely. Using appropriate machine learning approaches this paper explores to extract the polarity of the postings and predict the mood of a user accordingly. It could use to manage, communicate and collaborate with people more effectively and to manage own personal well-being and happiness. This study applies sentiment analysis for analyzing the hidden information present in the text on social media postings. It is an application of Natural Language Processing. In order to perform Sentiment analysis, need to identify the subjective and objective in the text. Because only the subjective text describes the sentimental information. Then the subjective text is preprocessing using various text preprocessing methods to extract the features. Text preprocessing may include stop word removal, stemming, tokenization, conjunction handling, and negation handling. After performing sentiment classification sentiment polarity can be extracted. To achieve this study uses the lexicon sentiment analysis process. Next, for sentiment classification, machine learning approaches can be used. It is an automatic classification technique and classification is performed using text features. The study uses supervised learning techniques. Using predefined emotion classes and sentiment polarity classifier is built accordingly.