Abstract
Studies have been conducted in regard to personalized news recommendation using collaborative filtering mechanisms based on users' click behaviors. However, few existing studies have focused on news recommendations depending on the rates of news-category interests. In this paper, we present a personalized news-recommendation system that builds profiles of users' news-category interests, determines the number of news articles to be recommended for each news category in proportion to news-category interests and ranks news articles according to the user's news interests. In order to find the news categories that are interesting to read, the smart device collects the web pages viewed by the user and classifies the contents. We use machine learning techniques in order to classify web pages into different categories, and the results of our experimentation show that Naive Bayes achieved the highest F-measure. The news preference is automatically calculated by the news reading time and the length of the news text. News articles with high preference generated by the collaborative filtering technique are recommended based on the rate of each category. The recommendation system we have implemented is based on collaborative filtering using Mahout Recommender API on Hadoop. Through user testing, we also assess whether the proposed system is useful.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call