Abstract Summary/Description
Building on prior research, this study extends and improves an algorithm originally designed to identify flood-related ontology concepts, evaluating its applicability across multiple fields. The primary objectives are to: (1) demonstrate the algorithm's effectiveness in various domains (urban system, flood with urban system, and flood with fashion) and (2) illustrate how community detection enables each cluster’s “centroid” to anchor the ontology hierarchy and represent core topics. The improved algorithm shows greater efficiency, completing in 0.35 seconds compared to 0.6 seconds for the original algorithm on datasets with up to 1,000 rows, with both algorithms delivering consistent results. The findings confirm the algorithm's adaptability across different domains and demonstrate the use of cluster centroids, derived through community detection, as representatives of key topics in domain-specific ontologies. This approach enhances information retrieval, supports coherent data integration across systems, and provides a framework to aid decision-making and analysis in complex domains.
