Feature Interaction Detection Technique Based on Feature Assumptions

We present an approach for the detection of feature interactions. It is based on two concepts: Call-Context Model (CCM) and Feature Context Assumptions (FCA). CCM specifies the possible call-connection(s) related to the feature and FCA summarizes the “important” assumptions about the feature environment. By analyzing the CCM-FCAs of two or more given features, we determine the existence or absence of Common Call-Context(s) (CCC). In the case where a CCC exists, we concretize the FCAs and detect possible interactions by searching for conflicting assumptions. In the case where a CCC does not exist, there is no interaction between the features. We tested our approach with the Benchmark from Bellcore. We also detected new interactions.