Keyphrase-Based Literature Recommendation: Enhancing User Queries with Hybrid Co-citation and Co-occurrence Networks

The literature recommendation system addresses the issue of time-consuming literature searches for researchers. A scholarly literature recommendation system recommends related papers to the user's search query. Systems can improve the precision of user queries by generating relevant keywords. The proposed approach aimed to recommend research papers that align with the user's interests by analyzing the query and returning a set of relevant papers. By pulling relevant keyphrases from the keyphrase networks, this was possible. A novel hybrid approach was introduced, which combined co-occurrence and co-citation networks based on their unique connections. This hybrid method improved performance by making the user's query bigger and giving each keyphrase in the query set a certain amount of weight. The combination of co-citation and co-occurrence relations in the proposed method was able to capture co-occurring keyphrases with semantically similar keyphrases to the user's query. The results showed that when the top 40 or 50 articles were chosen for the user's query, the results were more relevant because the proposed method could capture more aspects of the user's query than traditional single network-based methods.