News Recommendation Using Term Frequency and Document Similarity

Excessive overloading of information has become a serious problem recently. Extensive use of technology has made life easier but it also lead to access of information creation. There are several news portals where lots of information gets uploaded daily. As it is an era of E-News where online news reading has become a common habit of people. People are more likely to read News on Web rather than on Newspaper or other media. It becomes harder for user to find relevant and popular news in small time. Now a day it has become a key challenge as everyone has different liking and reading habits. A solution to this problem is news recommendation system. A Content Based Recommendation is developed which recommends news on the basis of article similarity with query and document similarity. Measures like term frequency count & document similarity are used to find out the similarity of query in the complete corpus of News articles. Each document is compared with every document available in corpus and content matching is performed to find out the similarity score. Results are evaluated on two different datasets using measures are used to evaluate the relevancy of recommended News articles.