The issue of e-Maintenance, i.e. the remote maintenance of devices, is more and more important as the loose-coupling between center of competencies and productions sites is increasing. Maintenance operators have to know procedures for all devices they are in charge of. Whispering possible interactions to operators when they operate could ease their task and increase their productivity. In this paper, we propose to model maintenance procedures (therefore sequences of actions of human operators) using the Semantic Web standards. The system can infer the possible next actions from the human operator, and assist him by suggesting the next operation steps. An analysis of the use's trace is also proposed in order to hint operators on improvements in his processes.