Framework for analysis of software requirements

Abstract

newline Software engineering projects need high quality requirements to succeed. It is common newlinepractice to use natural language to develop software requirements specifications or other newlineartifacts to document requirements. However, deliverables in natural language tend to be newlineambiguous, and, sometimes, even inaccurate. Since requirements are an input to ensuing newlinetasks in the software development life cycle, it is very important to be able to perform newlineanalysis for detection of defects, wrong information or missing information. Manual newlinechecking of requirements for consistency and completeness is tiresome and error-prone, newlinemore so, when there are a large number of requirements. This PhD thesis proposes a Petri newlineNet and Natural Language Processing based approach to achieve improved software newlinerequirements through analysis leading to reduction in cost and schedule overruns.