Logical Coherence and Repair in Ontology Alignment

You are reading version 1 of this direction, frozen 2026-10-04. The latest version is version 2.

Version 1 of 214 papers22 conceptsfrozen 2026-10-04

When we integrate data from different sources—such as merging two medical databases or connecting different web schemas—we rely on ontology matching. This is the process of finding correspondences (mappings) between related entities in different vocabularies. However, as researchers like Antoine Zimmermann and Alfio Ferrara have documented through years of the Ontology Alignment Evaluation Initiative (OAEI), simply finding similar names is not enough.

The core challenge is that while a mapping might look correct in isolation, it can be "logically toxic" when added to the combined system.

The Problem of Unsatisfiable Classes

Imagine two ontologies. Ontology A defines a "Vegan" as someone who eats no animal products. Ontology B defines a "Cheese Lover" as someone who must eat dairy. If an automated matcher sees that both classes involve "Food Consumers" and incorrectly maps "Vegan" to "Cheese Lover" based on some shared context, the resulting integrated model contains a logical contradiction: a person who must eat dairy but cannot eat animal products.

In formal logic, this results in an unsatisfiable class. If a class is unsatisfiable, any automated reasoner trying to process the data will flag an error, rendering the entire integrated knowledge base unusable for formal inference. As noted in the Results of the Ontology Alignment Evaluation Initiative 2015 (OM 2015), automated matching often produces alignments that are locally plausible but globally incoherent.

Logical Repair as a Filter

To solve this, modern alignment systems incorporate logical repair facilities. Instead of treating matching as a simple similarity search, they treat it as a constraint satisfaction problem.

The process generally follows three steps:

  1. Candidate Generation: Use lexical or structural algorithms to find potential mappings.
  2. Reasoning: Use an automated reasoner to "test" what happens if these mappings are applied. The reasoner looks for logical contradictions or classes that become unsatisfiable.
  3. Repair: If a contradiction is found, the system identifies the minimal set of mappings that must be removed to restore coherence.

By integrating these reasoners directly into the alignment pipeline, systems can ensure that the "glue" they provide doesn't break the underlying logic of the data.

Evolution of the Field

The importance of this mechanism has been a recurring theme in the OAEI benchmarks. In the Results of the Ontology Alignment Evaluation Initiative 2014 (OM 2014), the community emphasized the integration of logical reasoning tools to detect contradictions early.

By the time of the Results of the Ontology Alignment Evaluation Initiative 2018 (OM@ISWC 2018), led by researchers such as Alsayed Algergawy and Michelle Cheatham, top-tier systems like LogMap and AML (AgreementMakerLight) had become dominant largely because of their sophisticated handling of these logical constraints.

Daniel Faria and other contributors have shown that while basic schema matching has reached a performance plateau, the real frontier lies in scalability—performing these complex logical repairs on massive ontologies with thousands of classes without crashing the reasoner. The goal has shifted from merely finding links to ensuring those links preserve the structural integrity of the global knowledge graph.

Papers in this version

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 18 papers
  2. Version 1 · 2026-10-04 · 14 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_logic_based_alignment_repair_8d482679,
  title = {logic-based alignment repair},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/logic-based-alignment-repair/rdr_8d4826796c9e428b99e6832d267e562b},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}

This exact version (version 1, frozen text):

@misc{lacuna_logical_coherence_and_repair_in_ontology_8d482679_v1,
  title = {Logical Coherence and Repair in Ontology Alignment},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/logical-coherence-and-repair-in-ontology-alignment/rdr_8d4826796c9e428b99e6832d267e562b/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-22},
}