Clustering-Based Project-Community Similarity Matching for Area-Based Research Site Selection Using Historical Institutional Data
Year: 2026
Type: conference-paper
Abstract: Area-Based Research (ABR) serves as the primary mechanism through which regional universities translate academic knowledge into local community benefits. At Kalasin University (KSU), the selection of research sites has historically relied on informal, experience-driven judgments rather than systematic data analysis. This paper proposes a Clustering-Based Project-Community Similarity Matching framework that leverages 86 completed ABR project reports spanning three fiscal years (2022–2024) to support evidence-based site selection decisions. We develop an automated feature extraction pipeline for Thai-language PDF reports and construct an 18-dimensional feature vector per project covering research domain scores, output type indicators, faculty affiliation, and project characteristics. K-Means clustering at K = 6 (Silhouette = 0.240, DBI = 1.443) produces six interpretable project clusters, each dominated by a single faculty, revealing a systematic faculty-silo effect in research organization. A cosine-similarity matching function recommends target districts for new project proposals, augmented by a Relative Coverage Gap score that identifies under-served communities. Gap analysis across 18 districts exposes a 7× disparity between the most and least researched areas (35 vs. 5 projects). The proposed framework delivers actionable, data-driven insights for university research governance and is directly replicable at other regional institutions.
Institution: Kalasin University
Language: English
Subfield: Management of Technology and Innovation
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7214718008