Reverse Engineering Web Sales Configurators

Sales configurators are widespread Web applications. Although such applications have specific common characteristics (e.g., they manage options governed by constraints, they enforce a configuration process.), they are usually developed in an unspecific way, that is, like any other Web application. Proceeding this way leads to configurators that are sub optimal in reliability, efficiency and maintainability. This PhD thesis is concerned with the reverse-engineering of Web sales configurators. It aims to develop a consistent set of methods, languages and tools to semi-automatically extract configuration-specific data from a Web configurator. Such data is stored in formal models (e.g., variability models, process models). These models can later be used for verification purposes (e.g., checking the completeness and correctness of the configuration constraints) as well as input for generative techniques (e.g., to re-engineer legacy configurators). More precisely, our two main research questions are: (1) How to extract variability data from the unstructured or semi-structured Web pages of a sales configurator? (2) How to extract such data from the dynamic content created when the configurator is executing? The accuracy of the extracted data and the scalability of the delivered tools are major concerns. The PhD thesis is meant to be completed within the coming year.