Intelligent Adaptive Needs Characterization for System Engineering

Abstract

The first phase of system engineering process includes collecting and understanding of stakeholder needs. Each collected need requires many hours of discussion with stakeholeder to translate the original need to a formal need statement with sufficient context and details for system engineers to develop requirements. Numerous follow-up discussions are common. A tool (ReqVISor) that would increase the quality of the formal need statements and descrease the number of repeated interactions between stakeholders and engineers is desired but non-existent. The proposed tool model is based on stakeholder needs for a first-person shooter video game. The tool would (1) extract the original need statement provided in natural human language format; (2) assert it in the ontologies that represent needs and stakeholders, domain concepts, and entities and relationships inside the need description; and (3) utilize the facts asserted into ReqVISor’s knowledge base to identify potential issues and infer possible relationship with previously collected needs. In this project we developed all the ontologies and related rules using Web Ontology Language (OWL) and concepts from upper level ontologies. The needs were organized in the ontologies according to System Engineering Boby of Knowledge (SEBok 0.75). The natural language processing was implemented using Stanford Parser and OpenCalais tool. The simple ReqVISor prototype with limited functionality was developed using Java. Finally, we used this prototype to assert the original video game need statements and identify conflicts and other issues with previously collected needs.