Modeling Software Developer Expertise and Inexpertise to Handle Diverse Information Needs

Simple item page
dc.contributor.authorClaytor, Frank L.en
dc.contributor.committeechairServant Cortes, Francisco Javieren
dc.contributor.committeememberEdwards, Stephen H.en
dc.contributor.committeememberHuang, Berten
dc.contributor.departmentComputer Scienceen
dc.date.accessioned2018-06-09T08:01:21Zen
dc.date.available2018-06-09T08:01:21Zen
dc.date.issued2018-06-08en
dc.description.abstractExpert software developer recommendation is a mature research field with many different techniques being developed to help automate the search for experts to help with development tasks and questions. But all previous research on recommending expert developers has had two constant restrictions. First, all previous expert recommendation work assumed that developers only demonstrate positive expertise. But developers can also make mistakes and demonstrate negative expertise, referred to as inexpertise, and show which concepts they don't know as well. Previous research on developer expertise hasn't taken inexpertise into account. Another restriction is that all previous expert developer recommendation research has focused on recommending developers for a single development task or expertise need, such as fixing a bug report or helping with a change request. But not all expertise needs can be easily classified into one of these groups, and having different techniques for every possible task type would be difficult and confusing to maintain and use. We find that inexpertise exists, can be measured, and that it can be used to direct inspection effort to find potentially incorrect or buggy commits. Additionally we investigate how different expertise finding techniques perform on a diverse set of long and short expertise queries and develop new techniques that can get more consistent cross query performance.en
dc.description.abstractgeneralExpert software developers are a useful source of information. There have been many papers that research techniques for recommending expert developers for different tasks and questions. But all previous research on recommending expert developers has had two constant restrictions. First, all previous expert recommendation work assumed that developers only demonstrate positive expertise. But developers can also make mistakes and demonstrate negative expertise, referred to as inexpertise, and show which concepts they don’t know as well. Another restriction is that all previous work on recommending expert developers has focused on recommending developers for a single development task or question. But not all expertise needs can be easily classified into one of these groups, and having different techniques for every possible task type would be difficult and confusing to maintain and use. In our first chapter we show that inexpertise exists, can be measured, and that it can be used to help identify potentially buggy or incorrect code. In the second chapter we investigate how different techniques for finding expert developers perform when evaluated on different kinds of expertise finding tasks to find which technique works well on multiples types of tasks.en
dc.description.degreeMaster of Scienceen
dc.format.mediumETDen
dc.identifier.othervt_gsexam:15305en
dc.identifier.urihttp://hdl.handle.net/10919/83505en
dc.publisherVirginia Techen
dc.rightsIn Copyrighten
dc.rights.urihttp://rightsstatements.org/vocab/InC/1.0/en
dc.subjectExpertiseen
dc.subjectExpert Recommendationen
dc.subjectSoftware Engineeringen
dc.titleModeling Software Developer Expertise and Inexpertise to Handle Diverse Information Needsen
dc.typeThesisen
thesis.degree.disciplineComputer Science and Applicationsen
thesis.degree.grantorVirginia Polytechnic Institute and State Universityen
thesis.degree.levelmastersen
thesis.degree.nameMaster of Scienceen

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Claytor_FL_T_2018.pdf
Size:
4.22 MB
Format:
Adobe Portable Document Format
Download

Collections

Masters Theses