Toward an Architecture for Model Composition Techniques

Abstract

Abstract-Academia and industry are increasingly concerned with producing general-purpose model composition techniques to support many software engineering activities, e.g., evolving UML design models or reconciling conflicting models. However, the current techniques fail to provide flexible and reusable architectures, a comprehensive understanding of the critical composition activities, and guidelines about how developers can use and extend them. These limitations are one of the main reasons why state-of-the-art techniques are often unable to aid the development of new composition tools. To overcome these shortcomings, this paper, therefore, proposes a flexible, component-based architecture for aiding the development of composition techniques. Moreover, an intelligible composition workflow is proposed to help developers to improve the understanding of crucial composition activities and their relationships. Our preliminary evaluation indicated that the proposed architecture could support composition tools for UML class, sequence, and component diagrams.

Toward an Architecture for Comparing UML Design Models

“…Many studies have been proposed in the field of the model composition such as [9], [11], [12]. Specifically, in [11], authors propose a flexible, strategy-based process for model composition approaches.…”

“…It is strategy-based because it composes elements based on the syntactic and semantic strategies. Similarly, the authors in [12] propose a modular and flexible architecture focused on model composition. The model comparison in this work is responsible for mapping the commonalities that will be integrated into the composed output model.…”

“…This process considers four criteria to calculate the similarity degree. These criteria specifically are: (1) Syntactic [11], the technique evaluates the structure of the visual language; (2) Semantic [6], the technique evaluates the meaning of the terms; (3) Structural [12], the approach evaluates the similarity of the aspects of the hierarchy, such as the kind of relationship and neighbors; and (4) …”

Evaluating the Effort of Integrating Feature Models: A Controlled Experiment (S)

“…However, developers may end up not producing the FMAB. Instead, developers often produce an output-composed model, FMCM, with problems (i.e., FMCM  FMAB) [1] [11]. This happens because developers are usually unable to properly detect and resolving integration problems, such as conflicts and inconsistencies, given the problem at hand.…”