Open NLP based Refinement of Software Requirements

Abstract

Software requirements are usually written in natural language (NL) or speech language which is asymmetric and irregular. This paper presents a suitable method for transforming user software requirement specifications (SRS) and business designs written in natural language into useful object oriented models. Here a neoteric approach is proposed to generate object oriented items from SRS. For NL processes like sentence detection, tokenization, parts of speech tagging and parsing of requirement specifications we incorporate an open natural language processing (OpenNLP) tool. It provides very relevant parts of speech (POS) tags. This parts of speech tagging of the SRS is quite useful for further identification of object oriented elements like classes, objects, attributes, relationships etc. After obtaining the required and relative information, Semantic Business Vocabulary and Rules (SBVR) are applied to identify and to extract the object oriented elements from the NL processed requirement specifications.