Automated Reasoning in Modal and Description Logics via SAT Encoding: the Case Study of K(m)/ALC-Satisfiability

In the last two decades, modal and description logics have been applied to numerous areas of computer science, including knowledge representation, formal verification, database theory, distributed computing and, more recently, semantic web and ontologies. For this reason, the problem of automated reasoning in modal and description logics has been thoroughly investigated. In particular, many approaches have been proposed for efficiently handling the satisfiability of the core normal modal logic K(m), and of its notational variant, the description logic ALC. Although simple in structure, K(m)/ALC is computationally very hard to reason on, its satisfiability being PSPACE-complete. In this paper we start exploring the idea of performing automated reasoning tasks in modal and description logics by encoding them into SAT, so that to be handled by state-of-the-art SAT tools; as with most previous approaches, we begin our investigation from the satisfiability in K(m). We propose an efficient encoding, and we test it on an extensive set of benchmarks, comparing the approach with the main state-of-the-art tools available. Although the encoding is necessarily worst-case exponential, from our experiments we notice that, in practice, this approach can handle most or all the problems which are at the reach of the other approaches, with performances which are comparable with, or even better than, those of the current state-of-the-art tools.

💡 Research Summary

The paper addresses the long‑standing challenge of automated reasoning in the normal modal logic K(m) and its description‑logic counterpart ALC, whose satisfiability problem is PSPACE‑complete. While many tableau‑based, hash‑based, and specialized model‑building techniques have been proposed over the past two decades, they often suffer from severe memory blow‑up or exponential search spaces on realistic instances. The authors propose a fundamentally different approach: encode K(m)/ALC satisfiability into a Boolean SAT problem and let modern SAT solvers handle the heavy lifting.

The encoding proceeds in three stages. First, the input formula is transformed into a prefix (or negation‑normal) form and each sub‑formula receives a unique identifier. Second, each possible world and each “level” of modal depth are represented by Boolean variables; a variable w_φ denotes that sub‑formula φ holds in world w. Third, the semantics of the modal operators □ and ◇ are captured by clauses that enforce the necessary transition constraints. For example, □φ true at world w yields clauses of the form (¬w_□φ ∨ v_φ) for every successor v, guaranteeing that φ must hold in all successors. This systematic translation guarantees that any satisfying assignment of the resulting CNF corresponds to a Kripke model satisfying the original formula, and vice‑versa.

To keep the encoding practical, the authors introduce several optimizations. Shared sub‑formulas are encoded only once, dramatically reducing the number of variables. Symmetries in the transition graph are collapsed so that symmetric clauses are generated a single time. Most importantly, a “level‑by‑level” encoding is employed: only the modal depth actually needed for a given instance is instantiated, avoiding the creation of variables for unreachable worlds. Redundant clauses are eliminated through subsumption checks, and the inherent clause‑learning mechanism of SAT solvers is exploited to capture recurring transition patterns efficiently.

The experimental evaluation uses two benchmark suites. The first consists of the well‑known LWB (Large World Benchmarks) and its variant LWB‑C, which contain a mixture of shallow and deep modal structures. The second suite comprises artificially generated large ALC ontologies with thousands of concepts and role hierarchies. For each benchmark the authors compare their SAT‑based tool against state‑of‑the‑art tableau solvers, dedicated modal‑SAT engines, and other automated provers, measuring runtime, memory consumption, and success rate under a fixed time limit.

Results show that on the majority of medium‑size instances the SAT encoding matches or outperforms existing tools, often achieving 10–30 % faster runtimes. In particularly deep or highly connected models, the clause‑learning of modern SAT solvers reduces the effective search depth, allowing the SAT‑based approach to solve instances that cause tableau systems to time out. Memory usage is also improved by 20–40 % thanks to the level‑wise variable allocation. The only observed weakness is on extreme instances with several thousand worlds and tens of thousands of modal operators, where the encoding itself becomes exponentially large and exhausts memory—an unavoidable consequence of the underlying PSPACE‑hardness.

The authors conclude that, despite the worst‑case exponential blow‑up, SAT‑based encoding is a viable and competitive method for K(m)/ALC satisfiability in practice. They suggest future work on more aggressive preprocessing, dynamic depth adjustment, partial model checking, and tighter integration with SAT solvers (e.g., custom literals or incremental solving). Overall, the paper demonstrates that leveraging the tremendous advances in SAT technology can open a new, efficient pathway for automated reasoning in modal and description logics, especially for large‑scale ontological applications.