Accuracy enhancement of collaborative filtering recommender system for blogs using latent semantic indexing

The selection of an interesting blog is becoming ever increasingly tough work for most users as the number of blogs is exponentially increasing every day. A Recommender System (RS) can help the user to find out the blog of his choice or based on parameters specified by him. The accuracy of RS to find out the right blogs for a particular user gets enhanced, if the text of blog is analyzed using semantic analysis. The RS uses the classification of blogs to find out the right blogs for a user. The information with respect to blogs like blog's title, tags given by blogger, blog image etc., provides good information to classify the blogs. A classification of text can be done using Part of Speech (POS) text analyzer. POS tagger of text provides information about the blogs that can be used to find the similarity between two blogs. The similar types of blogs can be kept together to provide better indexing in the set of large blogs. The paper finds the better accuracy of Collaborative Filtering Recommender System (CF RS) for bloggers when the blogs are processed with different text analysis using semantic information of blogs. The observation is addressed by computing various measures like Precision, Recall and F1.