The theme of this thesis deals the problem of web services in the context of OpenData and Clouds Computing. We witness on one hand to the utilization of Clouds computing for application deployment and data storage, and on the other to the publication of OpenData by local authorities, governments, etc. The aim of this thesis is to propose a service oriented architecture for the discovery and the composition of services dedicated to the efficient utilization of OpenData, while taking into account the uncertain aspect related to OpenData environments particularly that concerning the semantization process. This service oriented architecture should be generic and able to be deployed in Clouds Computing. This thesis will also focus on the nature of the data services. Indeed, modeling, semantic annotation, and composition must take into account some fundamental features of OpenData that may be inaccurate, incomplete or contradictory.