Optimization of Bounded Continuous Search Queries Based on Ranking Distributions

Dirk Kukulenz, Nils Hoeller, Sven Groppe, Volker Linnemann

Abstract

A common search problem in the World Wide Web concerns finding information if it is not known when the sources of information appear and how long sources will be available on the Web, as e.g. sales offers for products or news reports. Continuous queries are a means to monitor the Web over a specific period of time. Main problems concerning the optimization of such queries are to provide high quality and up-to-date results and to control the amount of information returned by a continuous query engine. In this paper we present a new method to realize such search queries which is based on the extraction of the distribution of ranking values and a new strategy to select relevant data objects in a stream of documents. The new method provides results of significantly higher quality if ranking distributions may be modeled by Gaussian distributions. This is usually the case if a larger number of information sources on the Web and higher quality candidates are considered.
Original languageEnglish
Title of host publicationWeb Information Systems Engineering -- WISE 2007
EditorsBoualem Benatallah, Fabio Casati, Dimitrios Georgakopoulos, Claudio Bartolini, Wasim Sadiq, Claude Godart
Number of pages12
Volume4831
Place of PublicationBerlin, Heidelberg
PublisherSpringer Berlin Heidelberg
Publication date01.12.2007
Pages26-37
ISBN (Print)978-3-540-76992-7
ISBN (Electronic)978-3-540-76993-4
DOIs
Publication statusPublished - 01.12.2007
Event8th International Conference on Web Information Systems Engineering - Nancy, France
Duration: 03.12.200707.12.2007
Conference number: 71211

Research Areas and Centers

  • Research Area: Intelligent Systems
  • Centers: Center for Artificial Intelligence Luebeck (ZKIL)

DFG Research Classification Scheme

  • 409-04 Operating, Communication, Database and Distributed Systems

Fingerprint

Dive into the research topics of 'Optimization of Bounded Continuous Search Queries Based on Ranking Distributions'. Together they form a unique fingerprint.

Cite this