Text mining with unstructured data from software repositories

Abstract

newline Exponential usage of internet resources and the need for centralization of the documents, has led to the increase in development of software projects also. Due to the complexity of projects and geographically diverse teams, software artifacts are maintained. Lot of artifacts are produced during the project lifecycle like requirement documents, design documents, source code, bug reports, etc. These software project artifacts help get a lot of information to help build up either the new project or to get the project management information for improving the software process. Nowadays nothing is recorded in the offline world. Everything has gone online. Software Repositories are used by almost all the software companies to store the artifacts so as to ensure the smooth working for a geographically dispersed team. newline newlineAutomatic Summarization is one of the most researched areas among the researchers performing research in the NLP field. The emergence of the World Wide Web and exponential growth of data, demands the need for ways for quick retrieval of the required task. The main focus of the thesis is to understand the techniques used for the task of summarization and apply the knowledge for the Bug Report Summarization and its related tasks. In order to achieve this task, the studies were divided into seven phases which included Study of NLP in Mining Software Repositories, Study of Extractive Summarization Techniques, Study of Abstractive Summarization, Study of Deep Learning Models, Study of Bug Report Summarization, Conduct of Experiments on generating the Bug Report summaries using different different techniques available and Study of how the duplicate bug reports hinder in project management during assignment, what are the various techniques being used for the task, how various similarity measures impact performing this task and Conduct of experiments on it. newline newlineThe first phase was to identify the tasks done in the area of software repository mining where the NLP techniques are used. After analyzing and