Computer Science – Information Retrieval
Scientific paper
2011-10-27
Computer Science
Information Retrieval
16 pages, 8 figures
Scientific paper
Background: Viewing the World Wide Web as a river network in which web pages are the confluences, hyperlinks are the channels, and clickstreams showing the navigation of users from one website to another are the water flows. In this way we can model the flow of human attention in a weighted, direct network comparable to transportation networks in organisms or ecosystems. Methodology: We construct a network comprising of 980 websites and 12,008 clickstreams with the publicly accessible data on www.alexa.com. The total traffic to these websites accounts for 97% of the global Internet traffic. Three quantities of interest are defined on the network, including connectance k, characteristic flow length l, and recycling rate. To study the size-variant properties of the clickstream, we detect six communities and observe how the three quantities change with community size. Conclusions: The clickstream network can be divided into language-based website communities, across which a scaling relationship in the clickstream recycling is observed. From this scaling regularity we obtain as a novel indicator characterizing the level of collaboration between websites in attracting user traffic. When the size of a community increases, its stability (k) declines, its attractiveness (l) increases, and its efficiency in utilizing traffic (gamma) keeps unchanged.
Wu Lingfei
Zhang Jiang
No associations
LandOfFree
The Flow Structure on the WWW does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with The Flow Structure on the WWW, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and The Flow Structure on the WWW will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-376174