Group Research Recommendation Using Ontology-Driven Knowledge Graph Mining and Metaheuristics

The rapid expansion of scholarly data on the Web has amplified the need for efficient data mining and knowledge-driven systems to support research. While article recommendation systems currently focus on individual users using collaborative or content-based filtering, they often struggle to capture collective research interest and semantic relationships within research groups. This paper presents a framework for research articles and authors, targeting student research groups. It leverages metaheuristic optimization and ontological representations. The proposed recommendation system models authors, institutions, papers, and research concepts as a structured knowledge graph that enables semantic mining of group interests and relationships through data mining. The recommendation framework employs a hybrid fitness function, which is a combination of textual relevance, group consensus and citation influence. It is optimized through a novel integration of Hill Climbing and Grey Wolf Optimizer(GWO) algorithms. The framework is stateless and does not rely on historical user data or offline training. For each group query, the fitness-function weights are re-optimized from scratch using only the group's current stated interests. Results obtained through experimentation demonstrate that the proposed hybrid of GWO and Hill Climbing attains superior fitness values and faster convergence when compared with the conventional optimization baselines. Further evaluation against various recommendation approaches presents consistency in improvement of relevance, actionability and diversity. The framework highlights the advantages of integrating metaheuristic optimizations with knowledge-graph mining for ontology-aware group recommendation in scholarly information systems.