A model for managing information flow on the World Wide Web
This thesis considers the nature of information management on the World Wide Web. The web has evolved into a global information system that is completely unregulated, permitting anyone to publish whatever information they wish. However, this information is almost entirely unmanaged, which, together with the enormous number of users who access it, places enormous strain on the web's architecture. This has led to the exposure of inherent flaws, which reduce its effectiveness as an information system. The thesis presents a thorough analysis of the state of this architecture, and identifies three flaws that could render the web unusable: link rot; a shrinking namespace; and the inevitable increase of noise in the system. A critical examination of existing solutions to these flaws is provided, together with a discussion on why the solutions have not been deployed or adopted. The thesis determines that they have failed to take into account the nature of the information flow between information provider and consumer, or the open philosophy of the web. The overall aim of the research has therefore been to design a new solution to these flaws in the web, based on a greater understanding of the nature of the information that flows upon it. The realization of this objective has included the development of a new model for managing information flow on the web, which is used to develop a solution to the flaws. The solution comprises three new additions to the web's architecture: a temporal referencing scheme; an Oracle Server Network for more effective web browsing; and a Resource Locator Service, which provides automatic transparent resource migration. The thesis describes their design and operation, and presents the concept of the Request Router, which provides a new way of integrating such distributed systems into the web's existing architecture without breaking it. The design of the Resource Locator Service, including the development of new protocols for resource migration, is covered in great detail, and a prototype system that has been developed to prove the effectiveness of the design is presented. The design is further validated by comprehensive performance measurements of the prototype, which show that it will scale to manage a web whose size is orders of magnitude greater than it is today.