Answers are grounded in this paper and cite their sources.
Universities need practical methods to integrate internal scholarly data for collaboration, discovery, and institutional decision-making. However, research information is often distributed across institutional records, ORCID profiles, and Scopus indicators, limiting the ability to identify meaningful internal collaboration opportunities. This paper presents an AI-supported explainable semantic enterprise research network for institutional collaboration recommendation. The proposed system integrates internal university records, ORCID-linked information, and Scopus-related indicators into a unified research intelligence platform. Researcher profiles are constructed from publication metadata, keywords, topics, citation indicators, publication-quality measures, and institutional attributes. The system uses OpenAI text-embedding-3-small to represent researcher profiles and applies cosine similarity to retrieve candidate collaborators. A structured AI-enrichment stage then generates shared topics, a recommendation rationale, a suggested project idea, and initial collaboration steps using pseudonymized researcher profiles. The platform was evaluated using institutional case study data, expert-based relevance assessment, additional baseline comparisons, ablation analysis, sensitivity analysis, and a preliminary user-acceptance assessment. The deployed system included 89 users, including 86 teaching and research staff, with 71% ORCID coverage and 67% Scopus coverage. The expert evaluation covered 45 recommendations for 9 target researchers and achieved a mean relevance score of 4.28, Precision@5 of 0.89, and NDCG@5 of 0.90. The proposed method performed better than keyword-overlap, BM25, and dense-embedding baselines. A preliminary System Usability Scale assessment with 15 users produced a mean score of 78.6. The results indicate that semantic profiling, institutional indicators, and explainable recommendations can support internal collaboration discovery and research monitoring.
No takes yet. Share an insight, caveat, or question.
Shaheen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: