hybrid-ontology-matching-62eb7d25·1 events·first seen Aliases: Hybrid Ontology Matching
Researchers introduce Hybrid Ontology Matching (HOM), a new task formulation that unifies equivalence and subsumption discovery in ontology matching, previously treated as separate problems. They propose AgentMap, an LLM-based multi-agent framework that combines semantic retrieval, hierarchical search, and collaborative reasoning to identify concept correspondences across ontologies. The system is evaluated on four extended OM datasets and outperforms baselines on both equivalence-only and subsumption-only settings in addition to the new hybrid setting.