Simplicity-oriented lifelong learning of web applications
| dc.contributor.advisor | Steffen, Bernhard | |
| dc.contributor.author | Bainczyk, Julius Alexander | |
| dc.contributor.referee | Hähnle, Reiner | |
| dc.date.accepted | 2024-01-29 | |
| dc.date.accessioned | 2024-04-15T14:04:59Z | |
| dc.date.available | 2024-04-15T14:04:59Z | |
| dc.date.issued | 2023 | |
| dc.description.abstract | Nowadays, web applications are ubiquitous. Entire business models revolve around making their services available over the Internet, anytime, anywhere in the world. Due to today’s rapid development practices, software changes are released faster than ever before, creating the risk of losing control over the quality of the delivered products. To counter this, appropriate testing methodologies must be deeply integrated into each phase of the development cycle to identify potential defects as early as possible and to ensure that the product operates as expected in production. The use of low- and no-code tools and code generation technologies can drastically reduce the implementation effort by using well-tailored (graphical) Domain-Specific Languages (DSLs) to focus on what is important: the product. DSLs and corresponding Integrated Modeling Environments (IMEs) are a key enabler for quality control because many system properties can already be verified at a pre-product level. However, to verify that the product fulfills given functional requirements at runtime, end-to-end testing is still a necessity. This dissertation describes the implementation of a lifelong learning framework for the continuous quality control of web applications. In this framework, models representing user-level behavior are mined from running systems using active automata learning, and system properties are verified using model checking. All this is achieved in a continuous and fully automated manner. Code changes trigger testing, learning, and verification processes which generate feedback that can be used for model refinement or product improvement. The main focus of this framework is simplicity. On the one hand, it allows Quality Assurance (QA) engineers to apply learning-based testing techniques to web applications with minimal effort, even without writing code; on the other hand, it allows automation engineers to easily implement these techniques in modern development workflows driven by Continuous Integration and Continuous Deployment (CI/CD). The effectiveness of this framework is leveraged by the Language-Driven Engineering (LDE) approach to web development. Key to this is the text-based DSL iHTML, which enables the instrumentation of user interfaces to make web applications learnable by design, i.e., they adhere to practices that allow fully automated inference of behavioral models without prior specification of an input alphabet. By designing code generators to generate instrumented web-based products, the effort for quality control in the LDE ecosystem is minimized and reduced to formulating runtime properties in temporal logic and verifying them against learned models. | en |
| dc.identifier.uri | http://hdl.handle.net/2003/42438 | |
| dc.identifier.uri | http://dx.doi.org/10.17877/DE290R-24274 | |
| dc.language.iso | en | de |
| dc.subject | Learning-based testing | en |
| dc.subject | Automata learning | en |
| dc.subject | Test automation | en |
| dc.subject | Continuous quality control | en |
| dc.subject | Domain-specific languages | en |
| dc.subject.ddc | 004 | |
| dc.subject.rswk | Domänenspezifische Sprache | de |
| dc.subject.rswk | Softwaretest | de |
| dc.subject.rswk | Qualitätssicherung | de |
| dc.title | Simplicity-oriented lifelong learning of web applications | en |
| dc.type | Text | de |
| dc.type.publicationtype | PhDThesis | de |
| dcterms.accessRights | open access | |
| eldorado.secondarypublication | false | de |