SERG talk

My presentation will focus on the paper titled “Does End-To-End Test Scenario Carving Lead To Meaningful Generated Unit Tests?”, as well as my current research.

Abstract

Automatic unit test generators such as EvoSuite are able to automatically generate unit test suites with high coverage. This removes the burden of writing unit tests from developers, but the generated tests are often difficult to understand for them. In this paper, we introduce our approach that generates unit tests starting from manual or scripted end-to-end (E2E) tests. Using carved information from these E2E tests, we generate unit tests that have meaningful test scenarios and contain actual test data. When we apply our approach, we observe that 85% of the generated tests are executable, and through a case study on 3 subject systems we get indications that the generated tests are relatively easy to understand.