A Note on Semantic Web Services Specification and Composition in Constructive Description Logics

The idea of the Semantic Web is to annotate Web content and services with computer interpretable descriptions with the aim to automatize many tasks currently performed by human users. In the context of Web services, one of the most interesting tasks is their composition. In this paper we formalize this problem in the framework of a constructive description logic. In particular we propose a declarative service specification language and a calculus for service composition. We show by means of an example how this calculus can be used to define composed Web services and we discuss the problem of automatic service synthesis.

💡 Research Summary

The paper addresses the challenge of automatically composing Semantic Web services by grounding the problem in a constructive description logic (CDL). Traditional Semantic Web service frameworks such as OWL‑S, WSMO, or SAWSDL rely on ontological descriptions and external planners, which often separate the logical consistency check from the generation of executable workflows. The authors argue that a constructive logic, where proofs correspond directly to constructive procedures, is better suited for service composition because the proof itself can be interpreted as an executable plan.

First, the authors introduce the syntax and semantics of CDL. Unlike classical description logics that are model‑theoretic, CDL is proof‑theoretic: concepts are built from atomic names using conjunction, existential restriction, and a constructive implication operator. Proof rules such as implication‑introduction, conjunction‑elimination, and the cut rule are defined, and each rule has a computational reading. This foundation allows the authors to treat a service specification as a logical implication: an input concept entails an output concept.

Based on this, a declarative service specification language is proposed. A service S is written as “InputConcept → OutputConcept”, where both sides are CDL concept expressions. For example, a book‑search service might be specified as Query → BookList, and an ordering service as BookList → OrderConfirmation. The language is intentionally minimal: it does not embed procedural details, but it captures pre‑ and post‑conditions in a form that can be directly used by the proof system.

The core contribution is a calculus for service composition. Given two services S₁: A → B and S₂: C → D, composition is permitted when B subsumes C (i.e., B ⊑ C). In that case a new service S₁∘S₂: A → D is defined. The composition is justified by constructing a CDL proof that starts from A, applies the implication‑introduction rule for S₁ to obtain B, then uses the subsumption B ⊑ C to bridge to S₂, and finally derives D. The authors show that this composition operator is associative and has an identity element, which means that complex service pipelines can be built modularly without ambiguity.

An illustrative example demonstrates the whole process. The authors model a “search‑then‑order” workflow by composing the BookSearch and Order services. They translate the two specifications into CDL formulas, build a proof tree that witnesses the entailment Query → OrderConfirmation, and explain how each inference step corresponds to a concrete data flow (the BookList produced by the first service becomes the input for the second). This example highlights that the proof itself serves as an execution plan, eliminating the need for a separate planning algorithm.

Finally, the paper discusses the automatic synthesis problem: given a set of atomic services and a desired overall specification (Input → DesiredOutput), does there exist a composition proof? The authors observe that this question reduces to the existence of a CDL proof, which is a decidable problem in many fragments of CDL. However, they acknowledge that current CDL tooling does not yet provide efficient, scalable proof search, and they identify open research directions such as heuristic proof‑search strategies, pruning of the composition space, and optimization of the resulting workflow (e.g., minimizing intermediate data transfers).

In summary, the work proposes a clean, logic‑driven framework for specifying and composing Semantic Web services. By leveraging constructive description logics, it unifies specification, verification, and execution planning within a single formalism. The calculus is shown to be mathematically well‑behaved, and the example illustrates its practical applicability. While the approach currently lacks mature automated tooling, it opens a promising avenue for future research on provably correct, automatically synthesized web service compositions.