In this paper, a new algorithm for web service business protocol discovery is presented. It is based on an information retrieval and document indexing technique, the TF-IDF. The latter’s formula has been modified and adapted for computing importance of edges of the graphs corresponding to business protocols in question.
Log files describing the execution history of web services are used without any a priori information. The proposed approach has been tested on synthetically generated log files and found to be very efficient in discovering business protocols