Enhancing Citation Recommendation Accuracy by Incorporating a Map Reduce‐Infused Framework for Optimized Scholarly Insights

ABSTRACT Researchers and scholars mostly prefer citation recommendation for identifying the papers relevant to their research during academic writing. Based on the given query text, it recommends the citation, but as it relies on keyword‐based searches and algorithms, it lags in capturing complex semantic links and multifactorial connections in scholarly research. Hence, a novel Semantic‐Aware Distributed Contextual Graph Attention Network (SADC‐GAN) is proposed in this research. As heterogeneous data sources and sparse connections create issues with integrating diverse bibliographic networks, the network sparsity constrains the relevance signal and faces issues with cold‐start problems in recommending less‐cited papers. Therefore, a novel Distributed Semantic Graph Attention Network (DSGAN) is presented for efficiently handling large‐scale bibliographic datasets, which brings better capturing and prioritizing of the relevant information while improving the contextual understanding. Besides, as conventional citation recommendation systems overlook relevant articles from innovative and diverse areas, in the absence of advanced semantic analysis to attain exact word matches, they result in nonideal recommendations, thereby bringing complexity to the research. Therefore, to overcome this issue, the proposed work introduces a Semantically Enhanced Context‐Aware Graph Attention Network (SEC‐GAN), which provides more significant and relevant recommendations by calculating and ranking the recommendations. Therefore, with the proposed research work, enhanced contextual understanding with more precise and meaningful bibliographic recommendations is attained. The implementation results illustrate that the proposed work performs better than existing techniques in terms of accuracy of about 82%, precision of about 90%, recall of about 84%, and F1‐score of about 88%.