CONCEPT BASED TEXT CLASSIFICATION

Abstract

The shift from Web 2.0 to Web 3.0 has significantly changed the perception of users for the internet and the Web. Web 2.0 has improved information sharing among the users, the contribution and collaboration of users, and Web 3.0 has improved the structure and representation of data. Web 3.0 (Semantic Web) is all about the concepts which relates more to real-world entities, which proves to be more realistic and practical. One of the most popular applications of Web 2.0 is blogging and its services. For example, Twitter has evolved as a great platform to share opinions and views on anything and everything related to daily life. As a result, these blogging websites have emerged as rich database for sentiment analysis and opinion mining. However, due to the nature of tweets (syntactically inconsistent), text-based sentiment classifiers fail to prove efficient. Concept based text classifiers are now-a-days gaining popularity and proving to be more efficient in such situations. Sentiment Analysis is inherently a prominent Text Classification problem. In order to improve efficiency of text classifiers, concept based techniques can be used. So, the inclusion of conceptual features of Semantic Web into the Text Classification problem can benefit the process of analysing the opinions of users. Therefore, in this research work, ontology-based techniques are used as concept based text classifier, to classify the text (tweets) more efficiently. In this approach, ontology is used to extract features or attributes on which tweets are to be analysed and accordingly score is given. The domain chosen is Ministries of Indian Government.