Improving research paper recommendation system using conceptual clustering algorithm

Abstract

The advent of web lead to the information overloads from the information scarcity. In particular, the scientific literature publication grew greatly, and the high development of digital library enabled researchers to search them conveniently. However the information overload made it difficult for researchers to keep current on important information. We suggest research paper recommendation system to overcome these problems. Our suggested system aims to deliver personalized research papers to researchers to help them understand their re-search areas or to help expand their ideas. Analyzing the current recommendation systems, we observed that little emphasis was placed on the quality of user profiling methods and conceptual clustering algorithm is typically used by IR area. This was the major intuition to propose and implement a new method of making concise and meaningful user profiles auto-matically and grouping users for recommendation. The suggestion is largely divided into two. First is to suggest a new approach of user profiling, and second is in the area of collaborative recommendation, also it follows clus-tering collaborative recommendation technique but it has its difference by clustering users conceptually, then recommend top-N item among groups.