Research Paper Recommendation System
With the ever-increasing amount of data being accessible over the web, the need to make recommendation systems more and more accurate is pressing. The approaches involving the semantic comparison of documents tend to become infeasible when querying a very large amount of data. This problem is not just restricted to a few domains but is slowly and gradually becoming a part of almost all the domains involving look-up for bulk data. The same goes for the research community as well. Research paper recommendation aims to recommend new articles that match researchers’ interests. It has become an attractive area of study since the number of scholarly papers increases exponentially. There are already approaches that make use of personalized suggestions based on user information but these approaches deal with the issues of lack of data for a newly registered user. This problem is referred to as cold-start. Cold-start problem occurs when trying to suggest a newly registered user regarding whom we don’t have much data and hence the recommendations systems are not able to figure out what to suggest.
The aim of this project is to devise a graph-based network involving the use of Natural Language Processing based techniques for filtering to reduce the search space for documents. This will be followed by a personalized look-up in the search space and ultimately a semantic comparison in the resultant search space. The goal is to reduce the search space for semantic comparisons from as large as 1,00,000 documents to a few thousand documents. This will lead to really fast searches. Moreover, the addition of personalized search approaches will also contribute to increased accuracy in search results.