An Application of Model Seeding to Search-Based Unit Test Generation for Gson

Abstract

Model seeding is a strategy for injecting additional information in a search-based test generation process in the form of models, representing usages of the classes of the software under test. These models are used during the search-process to generate logical sequences of calls whenever an instance of a specific class is required. Model seeding was originally proposed for search-based crash reproduction. We adapted it to unit test generation using EvoSuite and applied it to Gson, a Java library to convert Java objects from and to JSON. Although our study shows mixed results, it identifies potential future research directions.

Mitchell Olsthoorn

Assistant Professor

Mitchell Olsthoorn is an Assistant Professor researcher in the Software Engineering Research Group (SERG) at Delft University of Technology. He is also a member of the Computational Intelligence for Software Engineering lab (CISELab) and the Blockchain lab. Mitchell holds an Ph.D. degree in Computer Science – with a specialization in Computational Intelligence.