Reconstructing Input Parameter Models from Combinatorial Test Suites

Combinatorial interaction testing (CIT) is a black box testing approach which generates a set of (abstract) test cases from a given input parameter model. Existing algorithms try to minimize the number of test cases while preserving a certain coverage of the input parameters. See #1 and #2 for more information.

Usually the input parameter model (IPM) is handcrafted. Afterwards, a combinatorial strategy is selected and the test suites are generated (See #3). However, the creation of the IPMs is complicated and error-prone. Identifying possible parameters and parameter values is a manual and creative task and might lead to invalid or missing combinations.

Goal

In this thesis we would like to try the other way round. The goal of this thesis is to reconstruct input parameter models from a given set of (handcrafted) test cases.

Possible Tasks

Design an approach to reconstruct input parameter models
Develop a prototype tool which (semi-)automatically reconstructs IPMs- Identify possible open source projects and apply your approach and tool to them

Literature

Yilmaz, Cemal, et al. “Moving forward with combinatorial interaction testing.” Computer 47.2 (2014): 37-45.
Kuhn, R. Introduction to Combinatorial Testing (Presentation), 2011
Grindal, Mats, and Jeff Offutt. “Input parameter modeling for combination strategies.” Proceedings of the 25th conference on IASTED International Multi-Conference: Software Engineering. ACTA Press, 2007.APA 4.
Nie, Changhai, and Hareton Leung. “A survey of combinatorial testing.” ACM Computing Surveys (CSUR) 43.2 (2011): 11.
Pairwise Independent Combinatorial Tool by Microsoft: https://github.com/Microsoft/pict