An Expressive Model for Comparing Tree-Structured Data

We study the problem of comparing tree-structured data. This problem has applications in several diverse elds, including molecular biology, natural language processing, image and pattern matching, program analysis, document management, and change management in databases [SZ90, NBR88, WZJS94, WCM94, Yan91, WSC97, C3]. Our interest in this problem stems from implementing a subscription service that detects database changes by periodically querying an external database and checking the results for new changes [CGL97]. For example, consider a Web site listing local entertainment events [EG]. Our prototype allows a user to subscribe to certain changes e.g., the cancellation or rescheduling of an event. The prototype periodically queries the event listings, and uses a tree-di erencing algorithm to detect changes in the results.