S. Adafre, und M. de Rijke. Proc. of LinkKDD 2005, Chicago, IL, USA, (August 2005)
Zusammenfassung
In this paper we address the problem of discovering miss-
ing hypertext links in Wikipedia. The method we propose
consists of two steps: first, we compute a cluster of highly
similar pages around a given page, and then we identify can-
didate links from those similar pages that might be missing
on the given page. The main innovation is in the algorithm
that we use for identifying similar pages, LTRank, which
ranks pages using co-citation and page title information.
Both LTRank and the link discovery method are manually
evaluated and show acceptable results, especially given the
simplicity of the methods and conservativeness of the eval-
uation criteria
%0 Conference Paper
%1 adafre05discovering
%A Adafre, Sisay F.
%A de Rijke, Maarten
%B Proc. of LinkKDD 2005
%C Chicago, IL, USA
%D 2005
%K similarity, web-graph, wikipedia
%T Discovering missing links in Wikipedia
%U http://www.science.uva.nl/\~mdr/Publications/Files/linkkdd2005.pdf
%X In this paper we address the problem of discovering miss-
ing hypertext links in Wikipedia. The method we propose
consists of two steps: first, we compute a cluster of highly
similar pages around a given page, and then we identify can-
didate links from those similar pages that might be missing
on the given page. The main innovation is in the algorithm
that we use for identifying similar pages, LTRank, which
ranks pages using co-citation and page title information.
Both LTRank and the link discovery method are manually
evaluated and show acceptable results, especially given the
simplicity of the methods and conservativeness of the eval-
uation criteria
@inproceedings{adafre05discovering,
abstract = {In this paper we address the problem of discovering miss-
ing hypertext links in Wikipedia. The method we propose
consists of two steps: first, we compute a cluster of highly
similar pages around a given page, and then we identify can-
didate links from those similar pages that might be missing
on the given page. The main innovation is in the algorithm
that we use for identifying similar pages, LTRank, which
ranks pages using co-citation and page title information.
Both LTRank and the link discovery method are manually
evaluated and show acceptable results, especially given the
simplicity of the methods and conservativeness of the eval-
uation criteria},
added-at = {2009-08-06T15:16:38.000+0200},
address = {Chicago, IL, USA},
author = {Adafre, Sisay F. and de Rijke, Maarten},
biburl = {https://www.bibsonomy.org/bibtex/22e43ba5585b31f32eaf6da0ed77ce97e/chato},
booktitle = {Proc. of LinkKDD 2005},
citeulike-article-id = {558261},
citeulike-linkout-0 = {http://www.science.uva.nl/\~{}mdr/Publications/Files/linkkdd2005.pdf},
citeulike-linkout-1 = {http://www.isi.edu/LinkKDD-05/Download/Papers/linkkdd05-13.pdf},
interhash = {4e320a2ee75ba8f52d8e7bdc3af25e5a},
intrahash = {2e43ba5585b31f32eaf6da0ed77ce97e},
keywords = {similarity, web-graph, wikipedia},
month = {August},
posted-at = {2006-03-21 11:46:18},
priority = {0},
timestamp = {2009-08-06T15:16:53.000+0200},
title = {Discovering missing links in Wikipedia},
url = {http://www.science.uva.nl/\~{}mdr/Publications/Files/linkkdd2005.pdf},
year = 2005
}