[p2p-hackers] Interest based locality and query distribution

[Etienne Riviere]

Hi Alexander,

I think you should have interest in these two publications that give 
some hints about your questions. I'm interest in users behaviour 
modeling also, and i'll try to match your hypothesis against some real 
trace (from FastTrack (Kazaa)) i have.

[1] is dealing about a trace gathered from a crawl of thousands of 
edonkey clients (what is obtained is cache contents). This can exhibit 
some interest for your questions. What is shown is the effect of 
removing 'generous uploaders' from the network, among other 
measurements. On-going work in our lab is on fine-grain exploitation of 
semantic clustering.

[2] is about proximity of users in the same trace (both geographical and 
interest based)

Best regards.
Etienne

[1] : Sidath Handurukande, Anne-Marie Kermarrec, Fabrice Le Fessant, and 
Laurent Massoulié. *Exploiting Semantic Clustering in the eDonkey P2P 
network*. In /SIGOPS European Workshop/, Leuven, Belgium, pages 109-114, 
September 2004. 
http://www.irisa.fr/paris/Biblio/Papers/Kermarrec/HanKerLefMas04EWSIGOPS.pdf 


[2] : Fabrice Le Fessant, Sidath Handurukande, Anne-Marie Kermarrec, and 
Laurent Massoulié. *Clustering in Peer-to-Peer File Sharing Workloads*. 
In /3rd International Workshop on Peer-to-peer systems (IPTPS 04)/, San 
Diego, CA, February 2004.
http://iptps04.cs.ucsd.edu/papers/le-fessant-clustering.pdf


Alexander Löser a écrit :

>Hi all
>recently several unstructured peer-to-peer systems dynamically select
>neighbours for each peer using the principle of interest-based locality,
>e.g. [1,2,3,4].  To model and simulate the behaviour in such networks
>information about the distribution of interests of each user, e.g. the
>queries each user issues and the content each user publishes, is
>required.  I'm interested to prove the following hypotheses:
>
>1.) Queries follow a Zipf distribution, only a few queries are highly
>popular, while the majority of the queries cover only rare topics.
>2.) A minority of users issues the majority of all queries.
>3.) It is very likely, that these users also stay for a long time in the
>network.
>4.) Most of the queries cover only own interests. What percentage of
>queries is issued to 'random' topics?
>5.) User interests follow a Zipf distribution, e.g. a user spends much
>of her queries to only one or two topics.
>
>Does anybody know citations from real file sharing trace, e.g. Gnutella,
>Kazaa that support or reject my hypotheses?
>
>Alex
>
>
>[1] V. Cholvi, P. Felber, and E.W. Biersack. Efficient Search in
>Unstructured Peer-to-Peer Networks.
>[2] Adriana Iamnitchi, Matei Ripeanu and Ian Foster, Small World File
>Sharing Communities.
>[3] J. Keller, D. Stern and F. Dang Ngoc. MAAY: A Self-Adaptive Peer
>Network for Efficient Document Search.
>[4] Efficient Content Location Using Interest Based Locality in
>Peer-to-Peer Systems http://www.ieee-infocom.org/2003/papers/53_01.PDF
>
>--
>___________________________________________________________
>
>  Alexander Löser
>  Technische Universität Berlin
>  http://cis.cs.tu-berlin.de/~aloeser/
>  office : +49- 30-314-25551
>  fax    : +49- 30-314-21601
>___________________________________________________________
>
>
>_______________________________________________
>p2p-hackers mailing list
>p2p-hackers at zgp.org
>http://zgp.org/mailman/listinfo/p2p-hackers
>_______________________________________________
>Here is a web page listing P2P Conferences:
>http://www.neurogrid.net/twiki/bin/view/Main/PeerToPeerConferences
>  
>

-- 
Etienne Riviere - PhD candidate - PARIS project
Office E208 - IRISA-INRIA, Campus de Beaulieu, 35042 Rennes cedex, France
Tél: +33 (0)2 99 84 71 48, Fax: +33 (0)2 99 84 71 71
http://www.irisa.fr/paris/pages-perso/Etienne-Riviere
MajecSTIC'05 : 16, 17 & 18 Nov., Rennes : http://majecstic05.irisa.fr