Abstract
Natural language interfaces to Web data services are emerging as an enabling technology to grant non-expert users the ability to access huge data repositories. However, meeting the goal of natural interaction involves solving the complex problem of recognizing a relevant service or service composition given a potentially ambiguous and ungrammatical natural language question. In this paper, we discuss the requirements of natural language processing over data services and describe a lightweight process to for the semantic annotation of services. We then illustrate and evaluate a process to support natural language queries over a composition of services annotated according to the above process.