From Logs to Regression Tests using AI: Increase Test Relevance and Speed-up Test Automation

Prof. Bruno Legeard, University of Franche-Comté / Smartesting / French Testing Board

Julien Botella, Gravity Product Manager - Smartesting

Functional regression testing, such as end-to-end testing of web applications, benefits greatly from good coverage of actual usage in production. Unfortunately, automated tests are written by test automation engineers who don’t know what exactly the product usage is. Thus, tests may cover some behavior that is not performed by end-users, and hence may not reveal bugs that would be observed by end-users.

This presentation will show how AI-assisted log analysis can be used to evaluate and complete usage coverage by tests. The machine learning techniques used are trained from operation logs, extracted from monitoring tools like Splunk, Dynatrace, Elastic or Datadog, and allow a semi-automation of the creation and maintenance of regression tests.