Re: [igraph] bipartite networks question

[Gábor Csárdi]

Jose,

well, I think that the Page Rank in its current form does not really
make sense on bipartite networks, because in Page Rank the adjacency
matrix is normalized and it is assumed that every node could have the
same number of connections (i.e. to all the other vertices). This is,
however, not true if the network is bipartite, because you have two
kinds of nodes. (Unless you have the same number of nodes for both
kinds.)

It does not seem too difficult to me to generalize it to bipartite
networks, by normalizing differently, and perhaps if you want to
follow the "random surfer" model, then you want to allow jumps only to
an opposite kind of node.

Whether you will still have the favorable practical aspects of the
measure, I mean in terms or calculating it quickly for huge graphs, is
a different question.

Btw. if you search for 'page rank bipartite', this is the first hit,
it might be interesting for you:
http://portal.acm.org/citation.cfm?id=1426424

Best,
Gabor

On Sat, Mar 14, 2009 at 6:02 PM, Jose Quesada <address@hidden> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi,
>
> I'm using the lates igraph 0.6 beta off the bazaar repositories and
> the new bipartite networks functions.
> We have some strange results when using pagerank on large bipartite
> networks.
> I may need to reproduce them with an small example to post them here,
> but first, does it make sense to apply pagerank on bipartite networks?
> I couldn't find any paper that did this.
>
> Thanks,
> - -Jose
>
> - --
> Jose Quesada, PhD.
> Max Planck Institute,
> Center for Adaptive Behavior and cognition,
> Berlin
> http://www.josequesada.name/
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (MingW32)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iQIVAwUBSbvjGGMobUVGH+HKAQJIYQ//czBSkuwCJ7tGYTt1gWiCG3P4ueDFmK8j
> sGx9nvdKoA3fFWfshvVIcXFBglz1KW7ftYrxpm5vp0dBcyNW9nLgiFDa5mbnXRAo
> JMNAZqL8YT7VH10/W7jgwVnCD1ZC5MUiiBWIw+m/2zw5Csi3Vw6e7yGeFliWYpoP
> MP4IXL3EyRGFVmJvBc3H1gEC13hDn5fcaezwcb17axgRqSUw2dInS65E90ldcGq5
> d5thdFBNN+JTtlyamct2ygh9cCDhxl4/1FCgT8Q5AWNLTAWPeGGVLopTvVQIoIRu
> UUtycTuNxyhE1Gdupl1IFjZ1OQ215ARy3Y6cUyGzK1AFtlveuASDFW30VMDJlVr2
> kvCA9y/t2RR1qUvSopNUMTv+E56mlxR7eOdB6/vHMjmBQYrbY0zJ9IS7Lktvo0BV
> k/sEyRSTgHReTWiMZbbPsTa/Rq7HW9TmDz/X8igdxZCuhgNYFCyRbu/rlIH/qUIi
> 17tBsl0UxVjrwz1nHJej7RVrPPXsYaFyD9Aes+BqyPj4M+xf3D0CRan7uxjmZn2E
> t0bDNlO4gmQLnFr5BQ/tFc4v9rPCq2Q8sShx0uNB5WOMsfRLImffykfvbTMi6yNl
> x9jHE07EyN4XFusL3XMNNAyGcN0E26pgSP9dvqmNSnNQ+CLPRTJMrwN9Cp3cVneN
> exBh367jxM4=
> =VmH/
> -----END PGP SIGNATURE-----
>
>
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help
>



-- 
Gabor Csardi <address@hidden>     UNIL DGM