Finding Clones in UML Sequence Diagrams and Web Services

Abstract

UML Sequence diagram is used to express the behavioral aspect of Design Models. Finding similarity in UML Sequence diagram is required in order to avoid redundancy of similar fragments, thereby reducing the resource consumption. The units chosen for comparison in UML Sequence diagrams are fragments representing conversation. A conversation is nothing but a Behavioral Execution Specification (BES) having sequences of Message Occurrence Specifications (MOS) and hence caused events embedded within it. For comparing various conversations, frequently occurring patterns are detected and reported as clones. Similarity reporting in our work has been accomplished by a text based clone detector Nicad which basically performs on XMI codes. Nicad can not _nd similarity directly from XMI codes because the XMI representation of sequence diagram is very complex to understand and process. It needs a step by step procedure like Identification, Consolidation and Contenxtualization to add clear meaning to the context. The restructuring is done with the help of Txl a transformation language. It has its own grammar and transformation rules to rede_ne the XMI codes to a pretty printing form. Nicad can work only on the Contextualized XMI codes. We have utilized same tool Nicad in finding similarity in various WSDL _les as WSDL itself is written in XMI. As we are approaching towards an era of Service Oriented Architecture (SOA), services are the most important things to be concentrated. In web services, the unit of comparison is the service operations enclosed inside the WSDL file. During the down time or heavy traffic period, if a service is found unavailable, we can get our work done lending the service from another service provider identified previously. This is called web service tagging.