Research On Technology Of Direct Manipulation In Semantic Feature Modeling System

In current feature modeling systems, support for direct manipulation of features is not commonly available. This is due to the lack of speed of current feature model constraint solvers. For the above problems, this paper presents the new technology for direct manipulation in semantic feature modeling system.Firstly, this paper presents a method which generates a feature model from business process model by analyzing the process of generating feature model in semantic feature modeling system. This method is from the user's point of view, introduces the theory of software product line, divides the steps of generating feature model into: business process analysis, product plan analysis, use case analysis and feature analysis. The method by making use of the algorithm of folding operate organizes a use-case group that has the highly related and similar functionality, explicitly shows commonality and variability in business processes. The study of the method has practical significance to direct manipulate feature model, because in the selection phase of direct manipulation, if generating feature model accords to the user's intent, the work of feature manipulation will be greatly reduced in interaction phase.Secondly, this paper presents the ideas of specific region and describes the separation steps of specific region. In order to achieve direct manipulation of feature model, it is to have real-time feedback which requires system to be fast enough for constraint solving. But in the interaction phase of direct manipulation, most of the manipulations change only one or a few feature parameters, the most parameters of model don't change. Therefore, we can separate the region which caused changes by feature manipulation from model as specific region,This will limit the manipulation of feature in the specific region. In this specific region the number of constraints solving is very small, so that the constraint of feature model can be optimized for solving constraints. Finally, this paper presents the method of specific region constraint solving. Because the number of feature in this specific region is not very large, we use an undirected graph to represent the constraint system and give the definition of the method as well as the detailed steps.