Ontology-Based Data Access: Query Abstractions and UCQ Extension

Michel Leclère, Marie-Laure Mugnier and Guillaume Pérution-Kihli submitted a paper titled "Abstractions of Queries in Ontology-Based Data Access" to arXiv on 23 Jun 2026 (arXiv:2606.24618, v1, 57 KB). The paper studies the problem of translating data queries to the ontology layer in an OBDA setting based on existential rules and the certain answer semantics, and it compares notions of minimally complete and maximally sound abstractions.

What problem do the authors address?

The paper tackles query abstraction in ontology-based data access: specifically, how to abstract data queries by translating them to the ontology layer under existential rules and the certain answer semantics. The authors note that a perfect abstraction may not exist, so they focus on two weaker notions: minimally complete abstractions and maximally sound abstractions. They position their work as a study of these abstractions and how to express them within a particular extension of unions of conjunctive queries.

The abstract states the setting precisely: multiple data sources are integrated via mappings to an ontology in OBDA, the technical framework uses existential rules, and query answers are considered under the certain answer semantics.

How do the authors extend UCQs and what does that buy you?

They study abstractions inside an extension of unions of conjunctive queries (UCQs) that adds a limited form of inequality and a special predicate to mark database constants. The extension is able to express minimally complete abstractions and therefore perfect abstractions when such perfect abstractions exist. The paper also claims this extension does not lead to an increased complexity for the decision problems the authors consider.

Concretely, the two syntactic additions are a restricted inequality construct and a predicate that flags constants coming from the database. Those additions are presented as sufficient to capture minimally complete abstractions without raising the computational complexity of the core problems the paper addresses.

How do the authors handle maximally sound abstractions?

The paper characterizes maximally sound abstractions by linking them to the notion of maximum recovery from the data exchange literature. The authors make a new connection between maximally sound abstractions in OBDA and maximum recovery, using that bridge to describe and understand the limits of soundness when translating queries to the ontology layer.

This characterization gives a formal handle on what can be soundly abstracted at the ontology level, and ties the OBDA abstraction problem to established concepts in data exchange.

Why it matters

Clarifying which queries can be moved from the data layer to the ontology layer matters because OBDA systems rely on such translations to let users query heterogeneous sources through a single ontology. By showing a UCQ extension that expresses minimally complete abstractions while preserving complexity, the paper provides a concrete language that captures otherwise elusive abstractions. The connection to maximum recovery anchors maximally sound abstractions in an existing theoretical framework, making it easier for database theorists to apply known tools and limits.

Researchers working on OBDA, existential rules, or query rewriting now have a formal avenue to distinguish what is expressible as a minimally complete abstraction and what can only be captured as maximally sound.

What to watch

This arXiv submission is an extended version of a paper published in the proceedings of KR 2025; the related KR DOI is https://doi.org/10.24963/kr.2025/43. Watch for how the paper’s UCQ extension and the maximum recovery connection are used in follow-up work, for instance in formalizing automated abstraction algorithms or in implementations that test whether the claimed complexity preservation holds in practice.

Bibliographic note: the paper was submitted to arXiv on 23 Jun 2026 as arXiv:2606.24618 and lists Michel Leclère, Marie-Laure Mugnier and Guillaume Pérution-Kihli as authors.