ai4st SLR

Research on AI for Software Testing Research, 2020-2025

TabularCC BY 4.0 InternationalIntroduced 2025-06-17

To check the validity of the ai4st ontology, an adapted, lightweight systematic literature review (SLR) was conducted to analyse related research. This SLR protocol was followed:

Review title: Initial SLR on the application of the ai4st taxonomy.

Objectives of the review:

  • To test the validity of the ai4st taxonomy with an initial research selection.

  • To determine a classification of this initial research selection.

Research questions

  • RQ1: Which standardized terms are being used in \textit{ai4st} related research?

  • RQ2: Which alternative terms are being used in \textit{ai4st} related research?

  • RQ3: Are the \textit{ai4st} taxonomy dimensions useful to classify the pre-selected research?

Database: Research papers from the conference proceedings of the latest International Conference on Software Engineering, ICSE 2025 in Ottawa, Canada and its co-located conferences and workshops; and referenced papers for backward snowballing the research. Forward snowballing was in this case unnecessary, as ICSE 2025 represented the most recent research publications at that time. A complementary search of the IEEE and ACM digital libraries has added further software testing and AI-related research papers published between 2020 and 2025.

Inclusion criteria

  • Peer-reviewed original research.

  • Online available.

  • Research on AI for ST.

Exclusion criteria:

  • Meta-research such as evaluations, benchmarking, comparisons, surveys, taxonomies, roadmaps.

  • Testing of software-based systems like IoT, cloud, vehicle, etc.

  • Research on ST for AI.

  • Posters and tutorials.

Selection process:

  • Title and abstract screening for the pre-selection of unique research candidates by use of the concept map resulting from the stc ontology to identify ST-related research, and the concept map resulting from the ai4st dimensions 'AI type' to identify AI related research in the ST-related subset of research.

  • Full text review and assessment of the research contributions for the final selection of unique research.

Tools:

  • Online research libraries, including dblp, ACM DL, IEEE Xplore, and Google Scholar to identify related work; and

  • Python for text analysis and post-processing of finally selected research, supported by MS Visual Studio, Google AI Studio, and LibreOffice

Synthesis process:

  • Review of the new and synonym candidate terms for inclusion into the stc ontology.

  • Classification of the selected research for inclusion into the stc ontology.