Abstract

Web is a huge repository of information, and a massive amount of data is generated everyday on online platforms. Information, can be facts and opinions, facts are objective statements about an event, and opinions are subjective statements that reflect the sentiments of a person towards an event. Research on sentiment analysis has increased tremendously in recent years due to its wide variety of applications. To analyze sentiments, certain methods have been proposed, which can be broadly categorized as supervised machine learning and lexicon based approaches. Supervised machine learning methods are giving high accuracy but these methods need training data and are domain dependent, while lexicon-based methods are not domain dependent. Although, building of lexicon is costly, but once constructed, it can be applied for a wide variety of domains, but still lexicon based methods are restricted to their dictionaries and are full-dependent on the presence of terms that explicitly reflect the sentiment, while in many cases the sentiment of a term is implicitly reflected by the semantics of its context. Therefore, we’ve proposed context aware, semantically rich (conceptual & contextual semantics) lexicon-based method which is different from traditional lexicon-based methods that assigns sentiment score and strength to terms in a dynamic way, and outperforms baselines.