SimRank: a measure of structural-context similarity
G. Jeh, and J. Widom. KDD '02: Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining, page 538--543. New York, NY, USA, ACM Press, (2002)
DOI: 10.1145/775047.775126
Data: WWW, Scientific Papers, Recommender Systems Task: find similar nodes / similarity measure for nodes Method: SimRank is based on Google's PageRank. They convert the weighted graph structure G to another weighted graph G2, where each v2 = (v,v) and a link e2 = (v2, v2) = ((v', v''), (v''', v'''')) exists if G contains the two edges (v', v''') and (v'', v''''). This SimRank measure can be applied to homogenous (e.g. papers) and bipartite (e.g. user-item) domains Motto: Two objects are similar if they relate to similar objects.
%0 Conference Paper
%1 citeulike:349900
%A Jeh, Glen
%A Widom, Jennifer
%B KDD '02: Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining
%C New York, NY, USA
%D 2002
%I ACM Press
%K clustering context, information-retrieval link-mining measure, ranking, similarity, text-mining
%P 538--543
%R 10.1145/775047.775126
%T SimRank: a measure of structural-context similarity
%U http://dx.doi.org/10.1145/775047.775126
@inproceedings{citeulike:349900,
added-at = {2008-02-10T02:19:38.000+0100},
address = {New York, NY, USA},
author = {Jeh, Glen and Widom, Jennifer},
biburl = {https://www.bibsonomy.org/bibtex/29b901eee9786a06b6ef7203e7a920608/brightbyte},
booktitle = {KDD '02: Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining},
citeulike-article-id = {2162785},
comment = {Data: WWW, Scientific Papers, Recommender Systems Task: find similar nodes / similarity measure for nodes Method: SimRank is based on Google's PageRank. They convert the weighted graph structure G to another weighted graph G2, where each v2 = (v,v) and a link e2 = (v2, v2) = ((v', v''), (v''', v'''')) exists if G contains the two edges (v', v''') and (v'', v''''). This SimRank measure can be applied to homogenous (e.g. papers) and bipartite (e.g. user-item) domains Motto: Two objects are similar if they relate to similar objects.},
description = {stuff from citeyoulike},
doi = {10.1145/775047.775126},
interhash = {a6d4531690305dc44937118df813b4b5},
intrahash = {9b901eee9786a06b6ef7203e7a920608},
keywords = {clustering context, information-retrieval link-mining measure, ranking, similarity, text-mining},
pages = {538--543},
priority = {2},
publisher = {ACM Press},
timestamp = {2009-01-23T09:58:50.000+0100},
title = {SimRank: a measure of structural-context similarity},
url = {http://dx.doi.org/10.1145/775047.775126},
year = 2002
}