On investigating feature interactions described in the international benchmark for detecting feature interactions, published in 2000, it has been found that they contain many feature interactions that do not actually occur, and that these are the result of ambiguity of service specification descriptions. Such interactions are called "seeming interactions" in this paper. If large numbers of seeming interactions are included in the results of detecting interactions, too much extra time is required for solving interactions. One of the ambiguities of service specification descriptions is a lack of description of timing for triggering each service (called the trigger point: TP). To avoid such seeming interactions, this paper proposes a trigger point model where TPs are defined independently from individual services.