An exploratory study on improving automated issue triage with attached screenshots

Abstract

Issue triage is a manual and time consuming process for both openand closed source software projects. Triagers first validate the issuereports and then find the appropriate developers or teams to solvethem. In our industrial case, we automated the assignment part ofthe problem with a machine learning based approach. However,the automated system's average accuracy performance is 3% belowthe human triagers' performance. In our effort to improve ourapproach, we analyzed the incorrectly assigned issue reports andrealized that many of them have attachments with them, whichare mostly screenshots. Such issue reports generally have shortdescriptions compared to the ones without attachments, which weconsider as one of the reasons for incorrect classification. In thisstudy, we describe our proposed approach to include this new pieceof information for issue triage and present the initial results.