ARED-CK: an Automated Requirements Elicitation Approach Based on Decision-making with Complete Knowledge

Requirements elicitation, as the first phase of the software life cycle, its importance is becoming more and more prominent. How fast and accurate user requirements can be elicited, is a problem that people wants to solve for a long time. This paper proposes to establish an automated elicitation system based on domain knowledge on user requirements. By gradually perfecting and refining the original requirements description, we can acquire requirements effectively and ultimately improve the quality of requirements and the efficiency of the elicitation process. The decision-making based approach for automated requirements elicitation, ARED-CK, assumes that a fairly complete domain knowledge base is in place, based on which inquiry cycles for eliciting requirements is formulated into a decision-making problem. Through continuous interactions with the user, the requirements decision-making process use requirements models in the domain knowledge base matching with specific instances of requirements model, and ultimately identify a requirements model fit the specific purpose of the target user. Moreover, the paper put forward the proposed decision process for optimization, so that within least steps of user interaction an optimal precise requirements model can be identified.