Test'n'Mo: A Collaborative Platform for Human Testers and Intelligent Monitoring Agents

Many software bugs have disruptive consequences, both in financial terms and in loss of life. Software Testing is one widely used approach to detect software bugs and ensure software quality but the testing activity, conducted either manually or using testing frameworks, is repetitive and expensive. Runtime Monitoring, differently from Software Testing, does not require test cases to be designed and executed and - once the property to be monitored has been specified - it does not rely on human beings performing any further actions, unless a violation is detected. However the property to be monitored, that must feed the monitor along with the trace or stream of observed events, may be very hard to identify and specify. In this extended abstract we present the Test'n'Mo vision which goes in the direction of exploiting Artificial intelligence and Machine Learning as enabling techniques for a hybrid platform for Software Testing and Runtime Monitoring. In Test'n'Mo, human testers and software agents of different kinds - 'Learning Agents' and 'Runtime Monitoring and Testing Agents' - collaborate to achieve their common testing goal. Although Test'n'Mo is meant to address User Interface testing of web/mobile apps, the Test'n'Mo approach may be adapted to other software testing activities.