Problem Solving Recommendation Model based on Machine Learning for an Issue Tracking System

Abstract

Global electronically popularization has increased the reliance on software the demand for customized software is also increasing and facing the various transaction requests from different users project managers need to ensure that everything is accurately tracked and executed through the management tool Issue Tracking System (ITS) However ITS projects are all-encompassing and it is a cumbersome time-consuming and unproductive job to use manual methods to filter or classify these issues On the other hand each issue report contains many descriptive natural languages In the limited query conditions of ITS it is not easy to find similar problems thus causing a problem to be reported by many people not only increases the workload of the program developer but also increases the cost of the project In the past many studies have proposed various automated methods to classify or group issue reports but most of these studies focus on categorizing issues by severity or finding relevance between issues In fact the reply record of the issue report usually has the processing history and solutions which are useful information for the assignee This paper proposes a solution recommendation model for classifying issues and automatically summarizing the solution It can be known from the experimental results that the solution recommendation model can help the assignee to obtain the solution of similar issues thereby improving the processing efficiency

Date of Award	2019
Original language	English
Supervisor	Hei-Chia Wang (Supervisor)