Object Oriented Visualization of Natural Language Requirement Specification and NFR Preference Elicitation

摘要

Requirements engineering is where the formal meets the informal. Application of natural language understanding to requirements gathering remains a field that has only limited explorations so far. Further, automation of requirement gathering is still in its infancy. There are three module proposed in this paper. In the first module, an approach for automatic requirements capture from natural language requirements specification is proposed. This approach starts by subjecting the natural language text for identification of parts of speech of the words in each sentence by applying sentence tagging techniques. The text thus tagged is normalized to simple sentences. Further, to resolve the ambiguity posed by the pronoun, the pronoun resolution is performed on the simplified text. Then, the elements of the object oriented system namely the classes, the attribute, methods and relationships between the classes, sequence of actions, the usecases and actors are identified by mapping the ‘part of speech tagged’ words of the natural language text onto the object oriented modeling language element, using some mapping rules based on the classical noun-verb analysis, thus eliciting the system requirements. In the second module, from the elicited object oriented elements, a semi-automatic approach of design and development of ontology for the requirement specification is proposed. Ontologies are especially useful for the development of high-level reusable software, like domain models and frameworks. They provide an unambiguous terminology that can be shared by all involved in the development process. Modeling of software requirements as ontology is done by, converting the object oriented elements elicited from the requirement specification as Resource Description Framework. By storing the ontology in database, the user can query and acquire the domain knowledge. This way, the same ontology can be used to guide the development of several applications, diluting the cost of the initial stage and allowing knowledge sharing and reuse. Requirement engineering plays a vital role in the development of the software. The quality of the software being developed depends on the Non-Functional Requirements(NFR), which are still not derived effectively due to the conflicts between NFRs. In the third module of this paper, a new approach is proposed to identify the NFRs for a given usecase description from the domain model such as Unified Modeling Language (UML) class diagram, and goal based questionnaires. This approach makes use of the domain model to find out the behavior of the system and possible constraints for actors in the system. The NFR taxonomy and the user preference are used to analyze the conflicts, which is resolved based on trade-off analysis by prioritizing the preference. The prioritization depends on the dominating NFRs from the inference engine