Web services and personal assistants are used more and more in our everyday life. Unfortunately we are very limited, when it comes to using them in a domain of travelling. Web service compositions solve some of the problems the user faces, but also creates others. Most of the existing web service compositions just aggregate different services, which makes it hard to deal with temporal or spatial-based variables. This affects not only the run-time of the service, but also the delivered results to the users. The solution we propose uses different appropriate temporal models with every service, in order to avoid possible errors and undesired results. It also uses different spatial models in this regard to create a seamless communication between the service and the different APIs. Furthermore the proposed solution will compose the different responses in order to deliver a result, according to the users specification and requirements. The evaluation in this thesis provides evidence of the significantly lower run-time of the solution, which in most cases is twice as low compared to using conventional methods.