A type of services that require human physical actions and intelligent decision making exists in various real fields, such as nursing in hospitals and caregiving in nursing homes. The authors' group call such services "physical and adaptive intelligent services," and is developing an IT-based system that is aims to assist cooperation and knowledge sharing among staffs. In this paper, we propose a new method for analyzing changes in the behavior of staffs before/after introducing the system. The method is based on probabilistic modeling of staffs' behavior. Using event-logs recorded by the system, behavior models are learned in the form of N-gram models. The method is also used for detection of unusual human behavior in the event-logs.