Querying web-based applications under models of uncertainty

• Authors: Daniel Deutch;Tova Milo
• Venue: Proceedings of the VLDB Endowment
• Year: 2008

Many businesses offer their services to customers via Web-based application interfaces. Reasoning about execution flows of such applications is extremely valuable for companies. Such reasoning must often operate under terms of uncertainty and partial information, due to partial tracing, effects of unknown external parameters, and more. The objectives of this research are (1) to define models for capturing Web application executions, with partial information and uncertainly of various flavors, (2) to design algorithms that allow for efficient reasoning over applications/execution traces under these models, and (3) to provide practical implementations that exploit these sound theoretical foundations for effective optimization of Web applications. We identify a restricted class of models that capture realistic scenarios, while allowing for an efficient query-based applications analysis. Hardness results indicate the necessity of such restricted models. We describe these results, highlight open problems, and consider directions for future research.