Discovering Conversations in Web Services Using Semantic Correlation Analysis

Abstract

Composite business applications typically combine characteristics of workflow and transactional applications. In order to handle multiple concurrent sessions and conversations across different nodes and over an extended time period, these applications pass dedicated identifiers inside their messages such as "OrderNumber". Debugging and understanding these conversations and the flow of conversation identifiers can be challenging in realistic business applications. We present a new heuristic algorithm to find these conversation identifiers and a method to reverse-engineer the conversations from the content of traced messages. Visualizing the results of this analysis provides a deeper understanding of the data flow, the business process and the interactions between business partners in composite business applications.