Predicting directed links using nondiagonal matrix decompositions
Predicting directed links using nondiagonal matrix decompositions
We present a method for trust prediction based on nondiagonal decompositions of the asymmetric adjacency matrix of a directed network. The method we propose is based on a nondiagonal decomposition into directed components (DEDICOM), which we use to learn the coefficients of a matrix polynomial of the network's adjacency matrix. We show that our method can be used to compute better lowrank approximations to a polynomial of a network's adjacency matrix than using the singular value decomposition, and that a higher precision can be achieved at the task of predicting directed links than by undirected or bipartite methods.
Decomposition into directed components, Trust
948-953
Kunegis, Jérôme
066b7173-f5a6-4a0e-9656-873af0821799
Fliege, Jörg
54978787-a271-4f70-8494-3c701c893d98
10 December 2012
Kunegis, Jérôme
066b7173-f5a6-4a0e-9656-873af0821799
Fliege, Jörg
54978787-a271-4f70-8494-3c701c893d98
Kunegis, Jérôme and Fliege, Jörg
(2012)
Predicting directed links using nondiagonal matrix decompositions.
In Proceedings of the 12th IEEE International Conference on Data Mining, ICDM 2012.
IEEE.
.
(doi:10.1109/ICDM.2012.16).
Record type:
Conference or Workshop Item
(Paper)
Abstract
We present a method for trust prediction based on nondiagonal decompositions of the asymmetric adjacency matrix of a directed network. The method we propose is based on a nondiagonal decomposition into directed components (DEDICOM), which we use to learn the coefficients of a matrix polynomial of the network's adjacency matrix. We show that our method can be used to compute better lowrank approximations to a polynomial of a network's adjacency matrix than using the singular value decomposition, and that a higher precision can be achieved at the task of predicting directed links than by undirected or bipartite methods.
Text
kunegis-fliege-predicting-directed-links-using-nondiagonal-matrix-decompositions.pdf
- Version of Record
Restricted to Repository staff only
Request a copy
More information
Published date: 10 December 2012
Venue - Dates:
12th IEEE International Conference on Data Mining, ICDM 2012, , Brussels, Belgium, 2012-12-10 - 2012-12-13
Keywords:
Decomposition into directed components, Trust
Organisations:
Operational Research
Identifiers
Local EPrints ID: 342984
URI: http://eprints.soton.ac.uk/id/eprint/342984
ISSN: 1550-4786
PURE UUID: b3e65b6d-abfa-4c06-9194-2a38d9579235
Catalogue record
Date deposited: 19 Sep 2012 13:51
Last modified: 06 Jun 2024 01:46
Export record
Altmetrics
Contributors
Author:
Jérôme Kunegis
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics
