Ali Alqahtani MSc Dissertation 2015/16

Building a dictionary of words related to a Java class

Supervised by G.Fraser

Abstract

Several tools exist for automated test generation which is an emerging aspect of software development. These tools analyse the code and automatically generate tests as well as random test data values for input parameters and assertions, which are randomly generated and not related to the class members contextually or semantically. This makes generated tests unreadable and difficult to understand.

The purpose of this project is to build a tool which, when given a Java class file (i.e. byte code), extracts words from class member names and uses natural language processing techniques i.e. tokenization, stemming and lemmatisation to get root forms of those words. The root words will then be checked in WordNet for generating a dictionary of words and their synonyms for the input class. This tool can then be used by EvoSuite for generating test data values which are contextually and semantically related to the class.