Pattern-based Model-to-Model Transformation: Long Version

We present a new, high-level approach for the specification of model-to-model transformations based on declarative patterns. These are (atomic or composite) constraints on triple graphs declaring the allowed or forbidden relationships between source and target models. In this way, a transformation is defined by specifying a set of triple graph constraints that should be satisfied by the result of the transformation. The description of the transformation is then compiled into lower-level operational mechanisms to perform forward or backward transformations, as well as to establish mappings between two existent models. In this paper we study one of such mechanisms based on the generation of operational triple graph grammar rules. Moreover, we exploit deduction techniques at the specification level to generate more specialized constraints (preserving the specification semantics) reflecting pattern dependencies, from which additional rules can be derived. This is an extended version of the paper submitted to ICGT'08, with additional definitions and proofs.

💡 Research Summary

The paper introduces a novel, high‑level method for specifying model‑to‑model (M2M) transformations using declarative patterns defined on triple graphs. A triple graph consists of a source graph (Gs), a target graph (Gt), and a correspondence graph (Gc) that links nodes of the source and target via two functions cs and ct. Patterns are constraints that describe which sub‑structures must (or must not) appear simultaneously in the source and target. Three pattern categories are distinguished: S‑Patterns (simple patterns) contain a positive graph Q together with optional negative pre‑conditions and negative post‑conditions; C‑Patterns (composite patterns) extend S‑Patterns with an additional positive pre‑condition graph C, thus capturing dependencies between patterns; N‑Patterns (negative patterns) consist solely of a negative post‑condition and are used to forbid certain configurations (e.g., duplicate objects).

A transformation specification S is simply the conjunction of a set of such patterns. Satisfaction of a pattern is defined in both forward (source‑to‑target) and backward (target‑to‑source) directions. In forward satisfaction, if none of the negative pre‑conditions are present, the positive graph Q must be found in the source‑target triple, and no negative post‑conditions may be present. Backward satisfaction mirrors this definition on the target side. A triple graph satisfies the whole specification if it satisfies every pattern in both directions.

The core contribution lies in compiling the high‑level declarative specification into executable operational mechanisms based on Triple Graph Grammar (TGG) rules. This compilation proceeds in two stages:

1. Deduction and Annotation – Deduction rules analyse the set of patterns to discover implicit dependencies and to generate additional, more specialized patterns. The main technical tool is the Maximal Intersection Object (MIO), the largest common sub‑graph that can be injectively embedded in two triple graphs. By intersecting patterns, the system automatically creates composite patterns that encode that “pattern A must be present before pattern B can be applied,” and it enriches existing patterns with negative post‑conditions derived from N‑Patterns (e.g., a rule that prevents creating a second foreign‑key when one already exists). This step also transforms N‑Patterns into positive pre‑conditions for other patterns, thereby turning negative constraints into constructive guidance.

2. Rule Generation – For each (now possibly enriched) pattern, a corresponding operational TGG rule is produced. Forward rules map the positive graph Q onto the source graph and, when necessary, create the required target elements; backward rules do the opposite. Because the patterns already encode dependencies, the generated rule set can be applied without an external control flow: the operational mechanism enforces the patterns in an order that respects the derived dependencies. A third operational function, the “relating” transformation, uses the same rule set to relate two existing models, yielding a triple graph that satisfies the specification without modifying either model.

The paper also sketches analysis techniques that can be applied to the specification level: checking for consistency (i.e., whether the set of patterns can be simultaneously satisfied), detecting potential non‑termination, and verifying that the generated operational rules preserve the semantics of the original declarative specification. The use of N‑Patterns to express duplication‑prevention constraints is highlighted as a way to guarantee that rule applications do not introduce unintended redundancy.

Compared with existing declarative M2M approaches such as classic TGGs, the presented method has several advantages. First, it is purely declarative: the specification contains only patterns, no explicit causality or control information. Second, pattern dependencies are inferred automatically, removing the burden on the modeler to order rules manually. Third, the formal foundation (triple graphs, pattern satisfaction, MIO‑based deduction) enables rigorous reasoning about the transformation, something that many ad‑hoc declarative languages lack. Finally, because the operational semantics are expressed as ordinary TGG rules, the approach can be integrated with existing TGG execution engines.

The paper is organized as follows: Section 2 introduces triple graphs and the three pattern types; Section 3 presents the deduction rules and the definition of MIO; Section 4 explains how operational TGG rules are derived from patterns; Section 5 discusses analysis techniques for specifications; Section 6 surveys related declarative M2M formalisms and positions the contribution; and Section 7 concludes with future work directions.

In summary, the authors propose a comprehensive framework that bridges high‑level declarative specifications and low‑level operational transformations. By leveraging pattern‑based constraints on triple graphs and a systematic deduction process, they achieve automatic generation of correct, bidirectional transformation rules while preserving a solid formal semantics. This work advances the state of the art in model‑driven engineering by offering a more expressive yet analyzable way to define and execute model‑to‑model transformations.