Multi-Agents System Based Software Error Identification, Reporting And Fixing

Abstract

Software systems are being used in various sectors of the economy. Software is never perfect and would always have a defect that need to be passed to the developers to be investigated and then fixed. A numbers of software vendors have incorporated a feature in their software that automatically generates an error report and sends it to the developers. This is not the case for majority of other software that relies solely on the users to report the error. This poses a great challenge on the functionality of the system as some of errors are never reported, while others are reported with incomplete information. In this study, we designed a prototype for a multi agents-based error identification and reporting system that automatically identify errors as they occur on a software and reports them. A multi-agent architecture is presented, involving a team of agents that identify and report the errors. Furthermore the agents automatically download fixes that are already available, and apply them. We tested the system with a number of known errors and all the errors were reported successfully. Further we tested the error fixing component with a number of fixes and all of them were applied successfully. We did an online survey to evaluate the user response to the system and the expected impact of the system to an organization. The data showed that the positive impacts outweigh the negative and the user response was good