2009 | OriginalPaper | Buchkapitel
Discovering and Building Semantic Models of Web Sources
verfasst von : Craig A. Knoblock
Erschienen in: The Semantic Web: Research and Applications
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
To achieve widespread use of the Semantic Web depends on having a critical mass of Web data available with semantic annotations. Since there are a huge number of sources available today without any such annotations, the challenge is how to find and build semantic models for these sources. In this talk I will describe an integrated end-to-end approach that automatically discovers information-producing web sources, invokes and extracts the data from these sources, builds semantic models of the sources, and validates the results by comparing the data produced by the source with the model of the source. These techniques are implemented in a system called DEIMOS, which integrates a diverse set of technologies to completely automate this task. DEIMOS starts with a “seed” source and finds other similar sources online using data from a social networking web site. Next the system learns how to invoke these sources through experimentation and then extracts data from these sources with automatic wrapping techniques. Finally, DEIMOS learns a semantic model of a source, which identifies the semantic types of the data produced by a source as well as the function that maps the inputs to the outputs. I will describe the challenges in integrating the component technologies into a unified approach to discovering, extracting and modeling new online sources. I will also present an evaluation of the integrated system on three different domains to demonstrate that it can automatically discover and model new Web sources.