Defeasible DL-Lite under Rational Closure: Tractable CQ Answering

Giovanni Casini, Umberto Straccia and 2 other authors submitted an arXiv paper on 23 Jun 2026 that shows reasoning and conjunctive query answering under Rational Closure for DL-Lite can be performed efficiently. The paper, arXiv:2606.24279, runs 108 pages and includes 2 figures and 1 table.

What does the paper show?

The paper demonstrates that Rational Closure (RC) can be applied to the core and horn variants of the DL-Lite family and that both entitlement (instance checking) and Conjunctive Query (CQ) answering under RC can be implemented with little extra cost. The authors state their main contribution is a plug-in architecture that builds upon existing standard classical reasoners, and they establish that reasoning and CQ answering under RC for DL-Lite can be done efficiently with minimal computational overhead.

Casini and Straccia frame the work within defeasible knowledge in Description Logics, treating RC as the non-monotonic formalism. Their evaluation focus is on the lightweight DL-Lite fragments, specifically the core and horn variants, and on two reasoning tasks named explicitly in the paper: instance checking (entitlement) and Conjunctive Query answering.

How does the plug-in architecture work?

The paper provides a plug-in architecture that leverages standard classical reasoners to perform RC-based reasoning and CQ answering, rather than replacing or reimplementing existing reasoners. In short: a DL-Lite knowledge base under Rational Closure is fed into a plug-in layer which delegates classical reasoning tasks to an off-the-shelf reasoner and composes results to yield RC-compliant answers.

That architecture is described as a bridge between defeasible DL-Lite knowledge and classical reasoners, allowing the authors to claim tractability and low overhead. The abstract and accompanying material say the design builds on "existing standard classical reasoners," and the work analyzes both instance checking and CQ answering under RC for the targeted DL-Lite fragments.

How did the paper document its work?

The submission to arXiv is detailed: 108 pages, 2 figures and 1 table are listed in the paper metadata. The authors present theoretical analysis and the plug-in design across the document, anchored to the formal properties of Rational Closure and to the syntax and semantics of the DL-Lite family. The arXiv identifier for the submission is arXiv:2606.24279 (submitted 23 Jun 2026).

Why it matters

Applying Rational Closure to lightweight description logics matters because DL-Lite is widely used where tractable reasoning is required, such as ontology-based data access. Showing that RC-based non-monotonic reasoning and CQ answering remain tractable when layered on top of existing classical reasoners lowers the barrier to adopting defeasible reasoning in systems that already use DL-Lite. If the plug-in approach truly imposes minimal computational overhead, practitioners can add defeasible behavior without replacing their reasoning stack.

What to watch

Look for an implementation or distribution of the plug-in and for follow-up work that reports empirical performance when paired with concrete classical reasoners. The paper on arXiv provides the theoretical design; the next concrete milestone will be integration tests or benchmarks plugging the architecture into standard reasoners and measuring actual overhead.

References: arXiv:2606.24279, "Tractable Reasoning and Conjunctive Query Answering for Defeasible DL-Lite under Rational Closure" (submitted 23 Jun 2026), Giovanni Casini (CNR - ISTI; University of Cape Town), Umberto Straccia (CNR - ISTI), 108 pages, 2 figures, 1 table.