User Acceptance Testing is typically the final phase in a software development process in which the software is given to the intended audience or domain experts. These domain experts know the functional requirements of the application and write user acceptance tests (UAT) in their natural language. A normal UAT test case in English typically follows an imperative sentence structure, i.e. a sentence that gives advice or instructions, or that expresses a request or command.

We propose a methodology to write UAT test automation code using natural language processing techniques on test scripts written in free form English text by using the assumption that test cases are written in an imperative style. We have also built a proof of concept tool, the Autotestbot, to demonstrate the feasibility of our idea. In addition, with the help of Autotestbot, we also demonstrate the feasibility of our proposed approach to semi-automate the time consuming and cumbersome manual UAT test code generation process. The scope of this thesis is restricted to automating Web applications.