The SphereSearch engine for unified ranked retrieval of heterogeneous XML and web documents

  • Authors:

  • Jens Graupmann;Ralf Schenkel;Gerhard Weikum

  • Affiliations:

  • Max-Planck-Institut für Informatik, Germany;Max-Planck-Institut für Informatik, Germany;Max-Planck-Institut für Informatik, Germany

  • Venue:
  • VLDB '05 Proceedings of the 31st international conference on Very large data bases
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents the novel SphereSearch Engine that provides unified ranked retrieval on heterogeneous XML and Web data. Its search capabilities include vague structure conditions, text content conditions, and relevance ranking based on IR statistics and statistically quantified ontological relationships. Web pages in HTML or PDF are automatically converted into XML format, with the option of generating semantic tags by means of linguistic annotation tools. For Web data the XML-oriented query engine is leveraged to provide very rich search options that cannot be expressed in traditional Web search engines: concept-aware and link-aware querying that takes into account the implicit structure and context of Web pages. The benefits of the SphereSearch engine are demonstrated by experiments with a large and richly tagged but non-schematic open encyclopedia extended with external documents.