Schema Compliant Consistency Management via Triple Graph Grammars and Integer Linear Programming

Nils Weidmann,Anthony Anjorin

doi:10.1007/978-3-030-45234-6_16

Abstract

Triple Graph Grammars (TGGs) are a declarative and rule-based approach to bidirectional model transformation. The key feature of TGGs is the automatic derivation of various operations such as unidirectional transformation, model synchronisation, and consistency checking. Application conditions can be used to increase the expressiveness of TGGs by guaranteeing schema compliance, i.e., that domain constraints are respected by the TGG. In recent years, a series of new TGG-based operations has been introduced leveraging Integer Linear Programming (ILP) solvers to flexible consistency maintenance even in cases where no strict solution exists. Schema compliance is not guaranteed, however, as application conditions from the original TGG cannot be directly transferred to these ILP-based operations. In this paper, we extend ILP-based TGG operations so as to guarantee schema compliance. We implement and evaluate the practical feasibility of our approach.

Highlights

In the context of Model-Driven Engineering (MDE), software systems are represented as a collection of different models
We investigate the impact of graph constraints on runtime performance, considering scalability of consistency checking for growing model sizes with and without taking graph constraints into account, by two research questions: (RQ1) By which factor does the number of variables and Integer Linear Programming (ILP) constraints increase when introducing graph constraints to the ILP? How does this influence the runtime of pattern matching, ILP construction, and ILP solving?
A difference can be observed for the ILP solving step, whose runtime is negligible without constraints, but increases by a factor of 10 when including graph constraints

Summary

Introduction

In the context of Model-Driven Engineering (MDE), software systems are represented as a collection of different models. Often several semantically related models are involved and have to be kept consistent to each other. The process of maintaining consistency among multiple models is called consistency management and involves various operations including (unidirectional) transformation, synchronisation, and consistency checking. Practical applications of consistency checking occur in the industry automation domain, where multiple domain-specific languages (DSLs) are used to describe complex systems [4]. Triple Graph Grammars (TGGs) are a declarative rule-based approach to specifying a bidirectional consistency relation between two modelling languages. The main advantage of TGGs is the possibility to derive multiple consistency management operations from the same formal specification. In their roadmap for c The Author(s) 2020 H.

Methods

Results

Conclusion