Information Retrieval Using Subjective Analysis on Social Blogs

Abstract—Number of blogs is increasing at a rapid pace and many potential applications for opinion detection and monitoring are arising as a result. Perspectives vary from one individual to another. In subjective analysis of various domain related reviews, one important problem is to produce a summary of opinions based on attributes. People leave on the web their opinions on domains and services they have known, it has become important to develop methods of (semi-)automatically classifying and gauging them. Subjective analysis helps in classifying adjectives which are commented in blogs into good, bad and neutral sets. Basing on these words one can decide whether the specific domain which is being searched has a positive or a negative review. We represent the blog posts in the form of bags of sentiments and use SVM(Support Vector Machine).When we use SVM, the objects are represented as vectors. A hyper plane will be chosen so that the margin or separation between two classes is distinct and clear. This separates objects of one class to the other.In this way,categorizing the opinions turns to be easy than the rest of approaches. Moving one step ahead of this procedure, we intend to develop an application which elucidates the customer to know whether the domain they are tending to know about is better than the rest of domains. We retrieve the information about the domains which are given as comments in social blogs, based on these comments we give the results in a comparative form with rest of the domains which makes it much easier for the user to choose the best. Not only this, but it helps analytics to identify the domains which are unsuccessful and try in improvising their attributes so that there would be exciting improvement and achievements. I