Asynchronous Distributed, Scalable Algorithms for Intelligent Dispersed, Hybrid Knowledge Bases

摘要

Integrating data and knowledge from multiple heterogeneous sources, each one possibly with a different underlying data model, is not only an important aspect of automated reasoning, but also of retrieval systems where queries can span such multiple sources. These sources can be as different as relational or deductive databases, object bases, constrained data (e.g., knowledge bases), structured files (e.g., spreadsheets), or even arbitrary program packages encapsulating specific knowledge, often in a hard wired form accessible only through function calls. Many queries can only be answered if data and knowledge from these different sources are available. In 1991-92, Gio Wiederhold proposed the pioneering concept of a mediator; a program that integrates multiple databases. The principal goal of this project was to develop a platform for the creation of mediated applications. Such a platform would provide a mechanism within which mediators may be developed for a variety of applications. The platform itself would be application independent, and would provide a variety of underlying technology and services that would be critical to the success of any specific application involving the use of mediation technology. The product of this research is a system called WebHERMES (heterogeneous Reasoning and Mediation System).