News Recommendation System for New User using Hybrid Approach with Demographic Data

Abstract

Recommender systems are fundamental solutions to information overload on the web due to the availability of multitude information sources. This system filters and presents relevant information to customer and/ or online users, a small subset of items that she/he is most likely to be interested in. The application of Recommendation schemes ranges from entertainment application (i.e., movies, music) to online newspapers information sources to recommend for the users based on their preferences. News recommendation scheme utilize features of the news itself and information about users to suggest and recommend relevant news items to the users towards the interest they have. However, the effectiveness of existing news recommendation scheme is limited during a scenario where information about a user or information about set of users in the system is unavailable. This leads to the occurrence of new user cold start problem. Therefore, the main objective of the study is: designing news recommender system using hybrid approaches to address new user cold start problem to ease and suggest more related news article for new users. In order to achieve the aforementioned objective, User Demographic Information (Data) with Hybrid Recommendation system is proposed. This hybrid recommender scheme combines content-based and collaborative filtering approach with demographic filtering. To evaluate the effectiveness of the proposed model, an extensive experiment is conducted using news articles dataset with user rating value and user demographic data. The performance of the proposed model evaluated using precision, Recall and F1-Score metrics which support the effectiveness of the proposed model. The proposed model performance is done by two ways of experiment. So, the performance of proposed model performs around 68.05% of Precision, 42.46% of Recall and 52.1% of average of F1_score for the experiment based on individual user similarity in the system. And also performs around 93.75% of precision, 40.25% of recall and 56.31% F1-score for the similarity of users based on the similarity of users within the same cluster which is better than the first experiment.