[Dspace-general] Statistics

[Mark H. Wood]

On Thu, Aug 28, 2008 at 08:40:11AM +1000, Scott Yeadon wrote:
> While jumping ahead a bit and not completely relevant to the context of 
> this discussion, it's important in any solution to separate out event 
> capture and statistics. Web server level statistics will only get you so 
> far. Having recently been through an exercise in building a prototype 
> statistics aggregator, the fundamentals in producing "good" statistics 
> (i.e. the reported information) is the *targetted capture* of events 
> (i.e. the raw event data) typically by the application (i.e. in the 
> DSpace code). We found the majority of reports which people want (or 
> rather the accuracy and granularity thereof) can only be provided where 
> the application has captured the event information rather than the more 
> general-level web container app. If you couple the DSpace 1.5.x event 
> producer/consumer feature with something like the De Minho front-end or 
> a Manakin stats aspect, that would make a pretty neat default stats package.

I agree. :-)  A start on that:

  http://sourceforge.net/tracker/index.php?func=detail&aid=2025998&group_id=19984&atid=319984

The Event System seems focused on changes to the repository, and I
recall that there was some resistance to expanding it to cover
references that don't change the model.  The above is a separate event
mechanism focused on reference events.

I've made considerable progress on adapting the University of
Rochester statistics package to take cases from this UsageEvent stream
instead of custom patching, and an XMLUI Aspect to make the resulting
per-object stat.s available for theming, but it's not quite ready for
daylight yet.

It's my understanding that the Minho package is one of those which
take cases from periodic digestion of log files.  Once an event stream
is available, it should be simple to create an adapter which appends
event records to a file in a suitable format, without clutter and with
the data you need.  The above patch demonstrates this with a plugin
which appends to a simple XML-like file.

-- 
Mark H. Wood, Lead System Programmer   mwood at IUPUI.Edu
Typically when a software vendor says that a product is "intuitive" he
means the exact opposite.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
Url : http://mailman.mit.edu/pipermail/dspace-general/attachments/20080828/9edd1894/attachment.bin