The scenarios or functional requirements constitute the first step of life cycle system that are usually vague and unclear, so a natural language processing model was created to solve this problem and to reduce time and effort spent by analysts. In addition, to transforming these texts after processing them into a structure related to the activities diagram. The major goal of this thesis is to develop new AI method that automatically extracts an activity diagram element from given functional requirements. The results are evaluated by using confusion matrix, which encompasses the calculating of number of factors, in which accuracy is one among them. The method has been manually tested, verified for validation and reliability. The accuracy registers approximately 81% for recognizing activity diagram element. Keywords: Activity Diagram, Generation Activity Diagram,I-CASE, NLP, Requirements Specifications.