Ohrndorf, Manuel (2023). A History-based Approach for Model Repair Recommendations in Software Engineering. (Thesis). Universität Bern, Bern
|
Text
23ohrndorf_m.pdf - Thesis Available under License Creative Commons: Attribution (CC-BY 4.0). Download (3MB) | Preview |
Abstract
Software is an everyday companion in today’s technology society that need to be evolved and maintained over long time periods. To manage the complexity of software projects, it has always been an effort to increase the level of abstraction during software development. Model-Driven Engineering (MDE) has shown to be a suitable method to raise abstraction levels during software development. Models are primary development artifacts in MDE that describe complex software systems from different viewpoints. In MDE software projects, models are heavily edited through all stages of the development process. During this editing process, the models can become inconsistent due to uncertainties in the software design or various editing mistakes. While most inconsistencies can be tolerated temporarily, they need to be resolved eventually. The resolution of an inconsistency affecting a model’s design is typically a creative process that requires a developer’s expertise. Model repair recommendation tools can guide the developer through this process and propose a ranked list of repairs to resolve the inconsistency. However, such tools will only be accepted in practice if the list of recommendations is plausible and understandable to a developer. Current approaches mainly focus on exhaustive search strategies to generate improved versions of an inconsistent model. Such resolutions might not be understandable to developers, may not reflect the original intentions of an editing process, or just undo former work. Moreover, those tools typically resolve multiple inconsistencies at a time, which might lead to an incomprehensible composition of repair proposals. This thesis proposes a history-based approach for model repair recommendations. The approach focuses on the detection and complementation of incomplete edit steps, which can be located in the editing history of a model. Edit steps are defined by consistency-preserving edit operations (CPEOs), which formally capture complex and error-prone modifications of a specific modeling language. A recognized incomplete edit step can either be undone or extended to a full execution of a CPEO. The final inconsistency resolution depends on the developer’s approval. The proposed recommendation approach is fully implemented and supported by our interactive repair tool called ReVision. The tool also includes configuration support to generate CPEOs by a semi-automated process. The approach is evaluated using histories of real-world models obtained from popular open-source modeling projects hosted in the Eclipse Git repository. Our experimental results confirm our hypothesis that most of the inconsistencies, namely 93.4%, can be resolved by complementing incomplete edits. 92.6% of the generated repair proposals are relevant in the sense that their effect can be observed in the models’ histories. 94.9% of the relevant repair proposals are ranked at the topmost position. Our empirical results show that the presented history-based model recommendation approach allows developers to repair model inconsistencies efficiently and effectively.
| Item Type: | Thesis |
|---|---|
| Dissertation Type: | Single |
| Date of Defense: | 14 July 2023 |
| Subjects: | 000 Computer science, knowledge & systems |
| Institute / Center: | 08 Faculty of Science > Institute of Computer Science (INF) |
| Depositing User: | Sarah Stalder |
| Date Deposited: | 16 Aug 2023 11:29 |
| Last Modified: | 19 Aug 2023 17:36 |
| URI: | https://boristheses.unibe.ch/id/eprint/4488 |
Actions (login required)
 |
View Item |