The study described in this paper is a part of a broader project on E-Tourism that aims at providing mobile users with context-aware personalised services. This paper focuses on issues raised by service discovery and automated composition. Given mobile users' requirement (e.g., buying airplane tickets, booking a hotel room, renting a car, etc.) expressed in a free-text query, the framework we describe in the paper deals with discovery and composition services to be executed in order to fulfil users' needs. First, behind the service discovery module is the idea to match the users' queries with a set of documents in a corpus such that each document contains a service description. To achieve the design of this module we have proposed a family of IR models. Second the module which is responsible with on-the-fly composition of the operations of the retrieved services is based on AI planning and produces an executable business process model whose each task correspond to a call to a service operation. The whole framework has been partially evaluated: we give in the paper experimental evaluations that show that our discovery model, based on query expansion via a co-occurrence thesaurus, outperforms the effectiveness of all models we have reviewed.