Author: Anjali Sreekumar
Programme: Doctoral Programme in Network and Information Technologies
Language: English
Supervisors: Dr Robert Clarisó Viladrosa & Dr Elena Planas Hortal
Faculty / Institute: Doctoral School UOC
Subjects: Computer Science
Key words: natural language processing, machine learning, software requirements engineering,feature oriented design and analysis,software product lines
Area of knowledge: Network and Information Technologies
Summary
A well-defined software requirements specification uniquely describes a functionality or part of a functionality of the software product and is consistent with the definitions and descriptions of the other functionalities in the product, without redundancy. Families of related products sharing common features among them are called software product lines. Engineering a software product line is a complex process. Gathering high quality software requirements and documenting them meticulously is a cumbersome task. Several types of human errors which can be introduced are proven to have a heavy cost on the success of the software project. The information collected will primarily be in the form of large volumes of textual information spread across multiple, mostly unstructured documents. The most critical task is to make sense of such a large text corpus. There is also no way to check the correspondence between the final requirements and the source documents. This thesis focuses on techniques and tools for the management of textual documentation in the engineering of a software product line. It aims to support the exploitation of natural language documents in the context of software product lines, providing automated mechanisms to extract Feature Models and check their quality.