| Abstract [eng] |
Information systems are designed to facilitate and speed up work within an area of activity. The developed systems must meet the requirements of the customer, but usually not all requirements are known at the beginning of the system development process. This happens due to the lack of communication between customers and information system developers. Business vocabularies and rules, expressed in natural language, are used to ensure a smoother communication process between customers and system developers. SBVR is a standard that allows describing business vocabulary and rules using structured natural language. The analysis of the existing solutions revealed that there are no automated tools that would be able to identify errors and contradictions in the business vocabulary and rules and would not limit the expressiveness of the SBVR standard. To avoid human factor errors in describing the business vocabulary and rules, a new method for contradiction detection was developed which later was implemented as a method prototype for error and contradiction checking within SBVR specifications. The prototype was implemented using contradiction detection algorithms, described in this work. The results of testing and experimentation of the prototype show that errors and contradictions in the SBVR specifications are mostly correctly identified. For future development of the system, it is recommended to supplement the detection of contradictions with algorithms that can trace the hierarchies of concepts and use them to detect the relationships of the concepts. |
