Change Management in Heterogeneous Semistructured Databases (Demonstration Description)

Abstract

This paper describes the design and implementation of the Stanford C3 system for managing change in an environment of heterogeneous, semistructured databases (e.g., web sites, legacy information systems with proprietary interfaces, news feeds, etc.). In particular, users can (1) detect and browse changes in such databases using the TDIFF component, (2) store and query historical data using the CORE component, and (3) subscribe to be notified when changes of a certain kind occur, using the QSS component. We use wrapper and mediator technology from the Tsimmis project to interface with diverse data sources, and the Lore system as a store for semistructured data. The TDIFF component uses tree-differencing algorithms for detecting changes in semistructured data. The CORE component implements the Delta-OEM (DOEM) data model and the Change Lorel (Chorel) query language. The QSS component ties together a variety of change detection and change reporting mechanisms in a common framework.