AIana: an AI planning system for test data generation

  • Authors:

  • Stefan J. Galler;Christoph Zehentner;Franz Wotawa

  • Affiliations:

  • Graz University of Technology, Graz, Austria;Graz University of Technology, Graz, Austria;Graz University of Technology, Graz, Austria

  • Venue:
  • Proceedings of the 1st Workshop on Testing Object-Oriented Systems
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Testing object oriented software is even for human beings a challenging task. Automating it is therefore very helpful for developers but by no means trivial. In this paper we present AIana, an approach for automatically generating complex objects used as test input data that satisfy a given precondition in terms of Design by Contract™ specification. AIana transforms the existing Design by Contract™ specification of the parameter type and the precondition of the method under test to PDDL (plan domain description language). Based on it, existing AI planners can be used to create a plan, i.e., a method sequence that transforms the object to the goal state. The goal state is given by the precondition of the method under test. AIana is evaluated on two case studies: a student developed stack based calculator, and a real-world event based application developed by our industry partner. AIana outperforms a random approach significantly in terms of methods tested and line coverage on both case studies.