## SOCNET@LISTS.UFL.EDU

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font

Subject:

eigenvecor in Pajek - collection of responses -and THANK you!

From:

Milanov, Hana

Date:

Wed, 22 Nov 2006 12:22:55 -0500

Content-Type:

text/plain

Parts/Attachments:

 text/plain (66 lines)
 ```***** To join INSNA, visit http://www.insna.org ***** Dear all,   I wanted to thank you for all the responses I received to my recent inquiry on how to get the measure of eigenvector or Bonacich Power centrality in Pajek. I wanted to share the responses with everyone in case someone runs into a similar problem as I have (having very large matrices to compute such measures in Ucinet)   I am copy-pasting the responses below and thank you all again for all your time and effort!   In Pajek you have a measure close to Bonacich power centrality for one-mode networks available with command Net / Vector / Important vertices / 1-Mode:Hubs/Authorities. The measure was introduced by Kleinberg for analysis of internet networks      http://www.cs.cornell.edu/home/kleinber/auth.pdf      http://www.cs.cornell.edu/home/kleinber/ There is also the corresponding measure (based on the same approach) for 2-mode networks available with command Net / Vector / Important vertices / 2-Mode:Important vertices   In Pajek you can get (Laplacian) eigenvectors for (not too large -   up to some thousands vertices) one-mode networks also using in   the draw window the command      Layout / Eigenvalues   and then transfering coordinates to vectors      Net / Vector / Get coordinate   For analysis of two mode networks you could also consider two new   methods available in Pajek ( (p,q)-cores and 4-rings ), see:      http://vlado.fmf.uni-lj.si/pub/networks/doc/dagstuhl/NewProcs.pdf      http://vlado.fmf.uni-lj.si/pub/networks/doc/ECPR/ECPR07.pdf   Vlado -------------------------------------------- RE: eigenvector centrality in Pajek? Phil Bonacich [[log in to unmask]] Hana - Standard statistical software, like SAS, permits the analysis of very large matrices. ------------------------------ Hi Hana, Maybe your computer didn't really freeze, maybe it just needed more time. I had a bipartite graph with about 3,500 cases and it took my computer something like 15 hours to calculate eigenvector centrality (multiple measures), but UCINET did it without a crash. With 10,000 it will probably take days. I would advise you to write to the creators of the program directly just in case. Vladimir Batagelj [log in to unmask] . ------------------------------------------------- Mikko Jääskeläinen [[log in to unmask]] Hi Hana, If you need the standard Bonacich '87 centrality measure, there is a way to compute this with Matlab from Access dyad-list output. Format: Firm A Firm B strengh year/or time unit Firm A Firm C strengh year/or time unit ....   _____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.```