Section: New Results
Anti-Bouncing Model for Usage Data Streams
Participants : Chongsheng Zhang, Florent Masseglia, Yves Lechevallier.
This work takes place in the context of Chongsheng Zhang's Ph.D thesis.
The bounce rate (BR ) of a website is the percentage of visitors (or users)
who hit a given page and do not visit any other page on that website. It is
defined as 
with To the total number of visits viewing only
one page and Tv the total number of visits. According to Wikipedia:
it essentially represents the percentage of initial visitors to a site who
"bounce" away to a different site, rather than continue on to other pages within the
same site.
Bounce rate is very important for usage analysis and most commercial
websites would like to lower it.
We claim that, in some cases,
the observed bounce rate is higher than the real one, because of the data
stream model.
In [39] , [49] we introduce the ABS (Anti-Bouncing Stream) model, a new model relying on a novel point of view. The goal of ABS is to maintain a reliable representation of the recent data in the stream and to avoid breaking down the navigations. We show that with ABS, we observe lower bounce rates.
