Aspect Based Sentiment Analysis in Social Media
Microblogging today has become a very popular communication tool among Internet users. Millions of users share opinions on different aspects of life everyday in popular websites such as Twitter, Tumblr and Facebook. With the rapid increase of user generated content on the social media, they serve as a rich repository. Spurred by this growth, companies and media organisation are increasingly seeking ways to mine these social media for information about what people think about their companies and products and how they make decisions. However, the more fine-grained the understanding is, the better. So detecting sentiment polarity towards a product/service as a whole no longer suffices. We want to know users' sentiment towards individual aspects of the product/service. The aspects can be predefined or might have to be dynamically discovered, depending on the problem setting.
Sentiment Lexicon Generation: Devised a method based on breadth first graph traversal to generate a subjective lexicon using WordNet as a resource. The method proposed in this project is language independent, and is tested for Hindi and English. In this method, we expand a hand curated seed list using WordNet relation to a subjective lexicon.
Aspect Detection and Categorisation: This sub area deals with identifying entity attribute pairsMultiple emotions about an entity can be expressed by the user. In order to understand this we first need to identify the aspec of the entity which the user is referring followed by categorisation. Categorisation refers to classification of these attributes into one of the predefined categories. An aspect of an entity can belong to different categories. The categories which are pre-defined refer to the areas in which we want to understand the emotions expressed by the users.
Sentiment Analysis: This sub area deals with identifying the polarity of the sentiment expressed by the user about a particular aspect of an entity.