Previous by Date Index by Date
Threaded Index
Next by Date

Previous by Thread Next by Thread

Re: Usage Statistics

In our case, our current statistics for locally maintained databases count
the number of logins.  This of course is only a rough approximation of the
number of users, since even with a timeout we often have several users at
a public station using the same session.  With our web server we haven't
come up with an equivalent to a "session", and the one set of statistics
we got from an online web vendor was a difficult to read mass of data much
like our web statistics. 

What matters most to us in interpreting the data is consistency.  I'd like
the same data calculated each month, and I'd like it to match our old
figures as much as possible. 

To most closely approximate the old "session" I'd like to have the number
of users who use the resource (retrieve an article or articles) with the
understanding that the same user coming back over an hour later would be
counted as a new "session". 

Usually the online providers only have a computer IP number, not a name,
for the user, but that parallels our experience with our public terminals
that have no password.

Of course validation by IP number and domain name is also a great problem
for us, since users of our local data bases come in from a great variety
of IP numbers and domain names.  If we can arrange a different validation
method with the vendors cooperation, we'll also have much better ways to
count the sessions ourselves.

Jim Morgan 
Ruth Lilly Medical Library
Indiana Unversity
© 1996, 1997 Yale University Library
Please read our Disclaimer
E-mail us with feedback