ITEE seminar: John Zakos, 02.00PM, Thu 23 Oct 2003

A Novel Concept-based Approach for Web Information Retrieval

Speaker: John Zakos, Griffith University

When: 02.00PM, Thursday 23 Oct 2003

Venue: 78-420

Host: Dr Xue Li

Abstract:


  With so much information on the WWW, we need adequate tools to be
  able to retrieve, analyse and visualise this information. From an
  information retrieval point of view, search engines allow us to
  quickly find information on the Web. Typically, a search engine
  performs retrieval based on a keyword-matching model. Consequently,
  a web document will only appear in search results if it contains the
  same keywords that were also used in the query. While a keyword
  matching approach may produce satisfactory search results for some
  queries, it will fail to retrieve relevant documents that use terms
  that do not appear in the query.

  The presentation will introduce a novel concept-based approach to
  information retrieval that has the ability to outperform
  keyword-matching systems. The main idea at the core of the approach
  is to treat a word as a concept, rather than a string of
  characters. This is a fundamental characteristic that allows the
  approach to adequately cater for the nature of language by taking
  word semantics into account.

  Based on this, we are developing a novel approach that effectively
  encodes and interprets conceptual information in documents. The
  conceptual encoding technique generates concise, logical and
  descriptive conceptual document representations by combining word
  semantics. These conceptual document representations, that are
  stored in a conceptual index, are then interpreted by an intelligent
  search algorithm to perform retrieval. The main characteristics of
  the approach are:

  1.  Exploits and incorporates conceptual information in
      ontologies/knowledge-base for information retrieval.
  2.  Fast, efficient and intelligent concept-based search.
  3.  Suitable for application to huge document collections such as
      the Web.
  4.  Independent of global document statistics.
  5.  Documents can be added or removed from the collection without
      requiring re-indexing.
  6.  Avoids erroneous word sense disambiguation.

Biography:

 
  John Zakos is a Ph.D. student under the supervision of Dr Brijesh
  Verma at the Computational Intelligence Research Lab (CIRL) within
  the School of Information Technology at Griffith University, Gold
  Coast Campus. He completed a B.Info.Tech (hons) in 1998 prior to
  joining IXLA Limited as a Software Engineer in 1999. He worked for
  over 2 years at IXLA and then returned to Griffith University in
  2002 to pursue a Ph.D. in the area of intelligent web information
  retrieval. In addition to his Ph.D. research, John is also involved
  with the development of web search technology in industry and has a
  general research interest in computational intelligence, pattern
  recognition and digital imaging. He has published 1 journal and 2
  conference papers.

Type:

D&KE

Contact:

Dr Xue Li, seminar host (xueli@itee.uq.edu.au)
or Guido Governatori (ITEE seminar co-ordinator) (guido@itee.uq.edu.au)

ITEE seminar web page: http://www.itee.uq.edu.au/~seminar

[All seminars]