A Hybrid AI Approach for Recommending Collaborators in Research Projects
Piermichele Rosati, Michela Quadrini, Emanuele Laurenzi
Communications in computer and information science
Problems Identified (4)
Research Collaborator Identification: Forming a research project consortium requires identifying adequate research collaborators, which is described as highly challenging.
Limited Traditional Recommendation Signals: Traditional collaborator recommendation methods relying only on social networks or citation counts are limited in efficacy.
Research Collaborator Identification: Forming a research project consortium requires identifying adequate research collaborators, which is described as highly challenging.
Limited Traditional Recommendation Signals: Traditional collaborator recommendation methods relying only on social networks or citation counts are limited in efficacy.
Proposed Solutions (4)
Agentic Graph RAG Collaborator Recommendation: The paper proposes an Agentic Graph Retrieval-Augmented Generation method for contextual, explainable collaborator recommendations tailored to researcher expertise and project relevance.
Hybrid KG-LLM Recommendation: The proposed method combines Knowledge Graph and Large Language Model capabilities for collaborator recommendation.
Agentic Graph RAG Collaborator Recommendation: The paper proposes an Agentic Graph Retrieval-Augmented Generation method for contextual, explainable collaborator recommendations tailored to researcher expertise and project relevance.
Hybrid KG-LLM Recommendation: The proposed method combines Knowledge Graph and Large Language Model capabilities for collaborator recommendation.
Results (3)
Improved Recommendation Effectiveness:
LLM-Based Evaluation:
Improved Recommendation Effectiveness:
Research Domain
AI-based research collaborator recommendation