Paper Key : IRJ************350
Author: Mayur Markad
Date Published: 13 Nov 2024
Abstract
AbstractThe paper addresses the challenges of citation recommendation in scholarly big data, such as the cold start problemand lack of paper ratings. The authors propose CNCRec, ahybrid model combining collaborative filtering and networkrepresentation learning to recommend citations based on bothpaper content and citation network topology. CNCRec createsa paper rating matrix using attributed citation networks andimproves neighbor selection by utilizing coauthor ship networks.Extensive experiments on the DBLP and APS datasets demonstrate that CNCRec significantly outperforms state-of-the-artmethods in terms of precision, recall, and Mean Reciprocal Rank(MRR), effectively mitigating data sparsity issues in citationrecommendationIndex TermsCollaborative Filtering (CF) and Network Representation Learning (NRL). These methods are applied toaddress the Cold Start Problem (CSP) and Data Sparsity (DS)in Scholarly Big Data (SBD). The research integrates GraphNeural Networks (GNNs) and Topic Models like Latent DirichletAllocation (LDA) to enhance recommendations
DOI Requested