Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.686921
Title: Analysing Java identifier names
Author: Butler, Simon
ISNI:       0000 0004 5920 8340
Awarding Body: Open University
Current Institution: Open University
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Identifier names are the principal means of recording and communicating ideas in source code and are a significant source of information for software developers and maintainers, and the tools that support their work. This research aims to increase understanding of identifier name content types - words, abbreviations, etc. - and phrasal structures - noun phrases, verb phrases, etc. - by improving techniques for the analysis of identifier names. The techniques and knowledge acquired can be applied to improve program comprehension tools that support internal code quality, concept location, traceability and model extraction. Previous detailed investigations of identifier names have focused on method names, and the content and structure of Java class and reference (field, parameter, and variable) names are less well understood. I developed improved algorithms to tokenise names, and trained part-of-speech tagger models on identifier names to support the analysis of class and reference names in a corpus of 60 open source Java projects. I confirm that developers structure the majority of names according to identifier naming conventions, and use phrasal structures reported in the literature. I also show that developers use a wider variety of content types and phrasal structures than previously understood. Unusually structured class names are largely project-specific naming conventions, but could indicate design issues. Analysis of phrasal reference names showed that developers most often use the phrasal structures described in the literature and used to support the extraction of information from names, but also choose unexpected phrasal structures, and complex, multi-phrasal, names. Using Nominal - software I created to evaluate adherence to naming conventions - I found developers tend to follow naming conventions, but that adherence to published conventions varies between projects because developers also establish new conventions for the use of typography, content types and phrasal structure to support their work: particularly to distinguish the roles of Java field names.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.686921  DOI: Not available
Share: