Scientific Paper Recommendation System

Advancements in research and technology have resulted in a flood of information in scientific research areas. It is challenging to find research papers of interest in today’s scenario, where hundreds of articles are published daily. It makes it impossible for the researcher and other readers to keep abreast of the research in their domain. Previously keyword matching techniques and recommender systems were used for finding relevant research papers. But these approaches were time-consuming and failed in scaling for a large dataset containing millions of articles. In this research, we propose a scalable end-to-end content-based scientific paper recommendation system capable of recommending research papers from the abstract or the context of the research paper for which we want to find the recommendation. We also evaluate our system using the average co-citation metric from [1]. Our end-to-end system performs with a average co-citation score of 14.88 and has an average response time of 1.4 seconds.