Meaning for the masses: Theory and applications for Semantic Web and Semantic email systems.

摘要

The Semantic Web envisions a portion of the World-Wide Web in which the underlying data is machine understandable and can thus be exploited for improved querying, aggregation, and interaction. However, despite the great potential of this vision and numerous efforts, the growth of the Semantic Web has been stymied by the lack of incentive to create content, and the high cost of doing so.; The goal of this dissertation is to enable and motivate non-technical people to both utilize and contribute content for the Semantic Web. As the foundation for our work, we identify three design principles that are essential for producing a successful Semantic Web system: (1) Instant Gratification---provide an immediate, tangible benefit to users. (2) Gradual Adoption---offer such benefit even when the system has few users. (3) Ease of Use---be simple enough for a non-technical person to use.; We then design mechanisms and theory that support these principles in the construction of two novel systems: MANGROVE, a community Semantic Web system, and Semantic Email, a system for leveraging declarative content to automate email-mediated tasks.; First, we describe MANGROVE's architecture and explain how its explicit publish and feedback mechanisms can provide instant gratification to content authors. In addition, we describe several novel semantic services that motivate the annotation of HTML content by consuming semantic information. We show how these services can provide tangible benefit to authors even when pages are only sparsely annotated. Furthermore, we demonstrate how seeding and inline annotation with our lightweight annotation syntax can bolster gradual adoption in MANGROVE.; Second, we introduce a paradigm for Semantic Email and describe a broad class of semantic email processes (SEPs). In support of instant gratification, these automated processes offer tangible productivity gains on a wide variety of email-mediated activities. To manage these processes, we define two formal models for specifying the desired behavior of a SEP. We show that computing the optimal message handling policies for these models is intractable in general, but identify key restrictions that enable these problems to be solved in polynomial time while still enabling a range of useful functionality. We then address a number of significant problems related to SEP usage by non-technical people. In particular, we design a high-level language for SEP templates that greatly simplifies the process of specifying and invoking a new SEP. In addition, we show that it is possible to verify, in polynomial time, that a given template will always produce a valid instantiation, and demonstrate how to generate explanations for the SEP's behavior in polynomial time. Finally, we describe how to meet our principles of gradual adoption and ease of use via a template-based semantic email server that functions seamlessly for participants with any mail client and with no a priori knowledge of semantic email.; Both systems have been fully implemented and deployed in a real-world environment, allowing us to report on practical experience gained with actual users. Overall, this work produces two novel, usable systems, as well as insights and techniques that can direct future Semantic Web systems.