Search for content and authors
 

Scaling of human activity at several Web portal

Anna Chmiel 1Kamila Kowalska 2Janusz A. Hołyst 1

1. Warsaw University of Technology, Faculty of Physics and Cent.of Exc.for Complex Systems Research, Koszykowa 75, Warszawa 00-662, Poland
2. Gemius SA, Wołoska 7, Warszawa 02-675, Poland

Abstract

The cyberspace can be understood as a complex system of human actions, with one very popular kind of this activity corresponding to the information search in this virtual world. The set of web pages we choose, the browsing art, and the time we spend reading the contents reflects our pastime habits and interests in the real world.To study this issue we used the data from Polish portals, and considered the behavior of a user on a Web page. Our analysis is based on cookie statistic provided by Gemius company. For example the portal www.onet.pl consists of 515 sub pages which are visited by 4 milions cookie users during one day. We constructed a weighted network of sub pages defining link weights as the number of users moving from one sub page to another. The weights distribution is a power law with characteristic exponents a = 1.45 -1.65. We observed the trafic at the portal and we learnt about habits and patterns of behaviour of internauts by the analysis of network properties. Our data contains also the information about the time spent by a  ser on various sub pages. We investigated the distribution of total time a user spent on the portal, and the distribution of times spent on one sub page. This distribution is a power law (b= 1.19) for the www.gazeta.pl and (b = 1.31) for the www.onet.pl) over two decades. We also examined also properties of partial time distribution. A partial time is the time the user stayed on one sub page normalized by the total time spent on the portal for this user. The distribution of partial times is a power law with b = 1.10 for the www.gazeta.pl and b = 1.13 for the www.onet.pl). On the other hand we observed an exponential distribution of numbers of different pages visited by a single user. We constructed a model explaining some of features observed at sub pages networks.

 

Legal notice
  • Legal notice:
 

Related papers

Presentation: Oral at International Conference on Economic Science with Heterogeneous Interacting Agents 2008, by Anna Chmiel
See On-line Journal of International Conference on Economic Science with Heterogeneous Interacting Agents 2008

Submitted: 2008-04-04 17:17
Revised:   2009-06-07 00:48