CMOMgen: Complex Multi-Ontology Alignment via Pattern-Guided In-Context Learning

TL;DR

CMOMgen tackles the challenge of Complex Multi-Ontology Matching by generating complete, semantically sound mappings that span arbitrary numbers of target ontologies. It combines lexical and embedding-based class selection with retrieval-augmented generation and in-context learning to produce OWL equivalentClass mappings, guided by reference patterns and examples. The approach introduces a semantic evaluation framework based on graph edit distance and demonstrates improvements over baselines on three biomedical phenotype tasks, with expert validation indicating substantial fidelity in non-reference mappings. This work advances automated CMOM, enabling richer, more traceable knowledge graphs across disjoint yet related domains, with practical implications for biomedical informatics.

Abstract

Constructing comprehensive knowledge graphs requires the use of multiple ontologies in order to fully contextualize data into a domain. Ontology matching finds equivalences between concepts interconnecting ontologies and creating a cohesive semantic layer. While the simple pairwise state of the art is well established, simple equivalence mappings cannot provide full semantic integration of related but disjoint ontologies. Complex multi-ontology matching (CMOM) aligns one source entity to composite logical expressions of multiple target entities, establishing more nuanced equivalences and provenance along the ontological hierarchy. We present CMOMgen, the first end-to-end CMOM strategy that generates complete and semantically sound mappings, without establishing any restrictions on the number of target ontologies or entities. Retrieval-Augmented Generation selects relevant classes to compose the mapping and filters matching reference mappings to serve as examples, enhancing In-Context Learning. The strategy was evaluated in three biomedical tasks with partial reference alignments. CMOMgen outperforms baselines in class selection, demonstrating the impact of having a dedicated strategy. Our strategy also achieves a minimum of 63% in F1-score, outperforming all baselines and ablated versions in two out of three tasks and placing second in the third. Furthermore, a manual evaluation of non-reference mappings showed that 46% of the mappings achieve the maximum score, further substantiating its ability to construct semantically sound mappings.

Figures (5)

• Figure 1: The logical definition of HP_0008163 (decreased circulating cortisol level) in both a simplified graphical version and the original OWL construct which we replicate with CMOMgen. The OWL version includes labels in the form of comments in the same line as the respective entity.
• Figure 2: Overview of the methodology for complex multi-ontology matching.
• Figure 3: Classes selection through two complementary strategies. (A) Lexical classes selection recursively finds target names that provide full coverage of the source name without overlap. (B) Language model-based classes selection uses recursive subtraction of the target name embeddings against the source name embeddings.
• Figure 4: Example of the graph transformation of the mapping in Figure \ref{['fig:LD']}.
• Figure 5: Distribution plot of fidelity across 50 non-reference complex mappings.

Theorems & Definitions (3)

• Definition 1: Ontology Matching
• Definition 2: Complex Ontology Matching
• Definition 3: Complex Multi-Ontology Matching