چكيده لاتين
The failure of software projects is the result of misunderstanding the requirements during the development and requirements analysis process. In this regard, graphic models are used to draw system processes so that all people involved in the projects of managers, employers and also software developers reach a common language (with minimal ambiguity) to express and understand system requirements. One of the most widely used symbols for business process modeling is the Business Process Model and Notation (BPMN), which has become a de-facto standard, both in IT and other business domains. As mentioned before, due to the very basic and important role of business process models regarding system requirements, the accuracy and correctness of these models is vital. However, the breadth and to some extent the complexity of BPM symbols and features lead to structural errors in the models developed by users.
One of the most reliable methods of verifying the created models is the use of formal verification, which can use the well-combined methods of the model expressed by formal languages and then verify their evaluation using appropriate tools. According to the characteristics of BPMN symbols, in this research, the official object adverbial language (OCL) has been used to express the structural rules of this symbol. First, with the help of natural language processing techniques, the rules expressed in the standard document of this symbol are automatically translated into their equivalent OCL constraints, then by adding these constraints to the current metamodel of these symbols, this metamodel is more accurate and complete in recognition. Structural errors in BPMN models have been improved.
To evaluate this research, there are two stages of design. In the first stage, the proposed algorithm is evaluated in terms of the amount of sentence structure coverage and also the correctness of the production method. The evaluation results of this stage have shown the coverage of 84.4% of the sentence structure and also the correctness of 100% of the constraints in terms of syntax. In the second stage, the semantic accuracy and performance of the produced constraints were evaluated. First, using the sample of designed tests, the semantic correctness of the produced was checked and 84.8% of the constraints were semantically correct. Then, 18 objective business process models from the industry were collected, and after the official verification of these models, it was found that 84.2% of the errors in them were correctly detected.
