Natural Language Based Object-Oriented Software Modelling

Abstract

Deriving useful information from natural language has been a task of much relevance for fields ranging from machine translation, software modelling, and artificial intelligence and so on. Sufficient literature is available on utilisation of grammatical inference in object oriented software modelling. The major advancements in this field along with the challenges faced by researchers as well as practitioners have been outlined. An amalgamation of ideas taken from existing theories and models along with proposed methodology has been worked out so as to utilise natural language text in the field of object oriented analysis and design. The very first step of Natural Language (NL) text processing is Parts-of-Speech (POS) tagging. Grammatical rules, some already existing and some deduced through careful observation of NL structures has been extensively discussed and implemented. After appropriate tagging the words to their respective parts of speech the objective is to recognise the classes among them. The classes along with their attributes and methods were listed out. These classes essentially are identified as part of the major functionalities in an information system. The information system consists of requirement specification given by clients for their target software. Comprehending client specification is a time consuming process. Therefore analysing classes from the specification provided becomes mandatory. Several ambiguities and redundancy in class identification were faced and were effectively resolved. Final classes from the given requirement specification were found out. Subsequently the knowledge acquired from the same is put to use while analysing these functionalities through various UML (Unified Modelling Language) diagrams. There are several UML tools that serve the purpose of drawing these diagrams. But the motive is to make the entire process of deriving the UML diagrams in a logical and automated manner.