SEEUNRS: Semantically Enriched Entity-Based Urdu News Recommendation System

Abstract

The advancement in the production, distribution, and consumption of news has fostered easy access to the news with fair challenges. The main challenge is to present the right news to the right audience. The news recommendation system is one of the technological solutions to this problem. Much work has been done on news recommendation systems for the major languages of the world, but trivial work has been done for resource-poor languages like Urdu. Another significant hurdle in the development of an efficient news recommendation system is the scarcity of an accessible and suitable Urdu dataset. To this end, an Urdu news mobile application was used to collect the news data and user feedback for 1 month. After refinement, the first-ever Urdu dataset of 100 users and 23,250 news was curated for the Urdu news recommendation system. In addition, SEEUNRS , a semantically enriched entity-based Urdu news recommendation system, is proposed. The proposed scheme exploits the hidden features of a news article and entities to suggest the right article to the right audience. Results have shown that the presented model has an improvement of 6.9% in the F1 measure from traditional recommendation system techniques.