License
When quoting this document, please refer to the following
URN: urn:nbn:de:0030-drops-17876
URL: http://drops.dagstuhl.de/opus/volltexte/2008/1787/
Go to the corresponding Portal


Halpin, Harry

Information-Theoretic Models of Tagging

pdf-format:
Document 1.pdf (49 KB)


Abstract

In earlier work, we showed using Kulback-Leibler (KL) divergence that tags form a power law distribution very quickly. Yet there is one major observed deviation from the ideal power law distribution for the top 25 tags, a large "bump" in increased frequency for the top 7-10 tags. We originally hypothesized that the "bump" in the data could be caused by a preferential attachment mechanism. However, an experiment that tested both feedback and no-feedback conditions over tagging (200+ subjects) shows that the power law distribution arises regardless of any feedback effect. We hypothesize that an information-theoretic analysis of tags lead to a power law without feedback.

BibTeX - Entry

@InProceedings{halpin:DSP:2008:1787,
  author =	{Harry Halpin},
  title =	{Information-Theoretic Models of Tagging},
  booktitle =	{Social Web Communities },
  year =	{2008},
  editor =	{Harith Alani and Steffen Staab and Gerd Stumme},
  number =	{08391},
  series =	{Dagstuhl Seminar Proceedings},
  ISSN =	{1862-4405},
  publisher =	{Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik, Germany},
  address =	{Dagstuhl, Germany},
  URL =		{http://drops.dagstuhl.de/opus/volltexte/2008/1787},
  annote =	{Keywords: Tagging information theory feedback}
}

Keywords: Tagging information theory feedback
Seminar: 08391 - Social Web Communities
Issue Date: 2008
Date of publication: 22.12.2008


DROPS-Home | Fulltext Search | Imprint Published by LZI