[Context] Significant changes are occurring in the scientific scenario. In particular, there is increasing collaboration among researchers, which has led to the increasing use of processing techniques and the need to share results and observations. Researchers usually need to compose experiments using available services to fulfill their requirements. However, finding and/or specifying scientific applications is not a trivial task and many scientists lack the computational knowledge necessary to compose their experiments. [Objective] Therefore, the objective of this work is to develop an infrastructure in which researchers can work with heterogeneous information, accessing scientific communities according to their interests. As a result, they can create new experiments with the help of compositions of semantic web services. [Methods] Considering the infrastructure model, a prototype was specified and evaluated in a real-world context. [Results] This evaluation found evidence for the feasibility of the use of this infrastructure in scientific scenarios. Therefore, this paper proposes a distributed platform where scientists can collaborate in order to specify an experiment and share its results, considering semantic information. This is in agreement with the ideas of “collaboratories”.