RAUMM: A Model for Requirement Ambiguity and Uncertainty Management in Software Requirement Specification

The Requirement specification document serves as a contract between the developer and the various stakeholders of the system under development. For better quality control, every requirement needs to be verified before being based line. According to the report of Systems Sciences Institute at IBM, it cost 6 times more when fixing a bug in implementation phase than fixing the same bug during the design phase. Furthermore, it costs 15 times more to fix bugs in testing phase if left during the design phase. This paper attempts to propose a model Requirement Ambiguity and Uncertainty Management Model (RAUMM) to deal with requirement ambiguity and uncertainty at early stage of software development. A tool has been developed to detect four types of ambiguities i.e. lexical, syntax, syntactic and referential ambiguity in requirements specification document. A compilation of potential ambiguous words is done from the taxonomy present in Ambiguity Handbook. This proposed tool will help the analysts to detect potential ambiguities. Several requirements writing rules have been taken from literature. Ambiguity Assessment has been performed by implementation of Adaptive Fuzzy Neural Network and ambiguities detected by tool have been taken as input. Relationship between Uncertainty and ambiguity has been analyzed finally.