Hi Eric,

Thanks for digging in to this, its very helpful. And sorry to leave you
hanging without a response.

At the time, when geoIP was unable to determine what city / country /
continent, the system would fail and spit out a null-pointer-exception (NPE)
and be generally unfriendly.

So it was an improvement to keep the system working, and say "unknown
country" instead of failing. However, all of these unable-to-be-determined
don't indicate that a secret unknown civilization has been visiting your
site, such as Atlantis, but that the system couldn't figure it out (timeout
for reverse DNS too short -- better in 1.7
DS-628<https://jira.duraspace.org/browse/DS-628>,
no data for IP's reverse geolocation, ...) .

However, it might be an improvement to just skip from showing you how many
not-available's there are affecting an entry in the stats, then to slap an
unknown on it.




--
Peter Dietz
Systems Developer/Engineer
Ohio State University Libraries



On Thu, Oct 28, 2010 at 11:27 AM, Eric Luhrs <[email protected]> wrote:

> I have done some digging, and feel that I have resolved all but one issue.
>
> The remaining point has to do with blank rows that show up under "Top
> city views".  In every case, the number of hits reported on this blank
> line matches the number of hits shown for "Unknown Country" under "Top
> country views."  I am not a programmer, but after looking at
> LocationUtils.java, I see that i18n messages are being pulled for
> unknown-country and unknown-continent, but not for unknown-city
> (805-821):
>
>
> http://scm.dspace.org/trac/dspace/browser/dspace/trunk/dspace-stats/src/main/java/org/dspace/statistics/util/LocationUtils.java
>
> In fact, there doesn't seem to be any code that deals with
> unknown-city, and there is no entry for it in Messages.properties.
>
> I found a few other sites that have the same problem, though most do
> not.  Here is one example:
>
> http://si-pddr.si.edu/dspace/displaystats?handle=10088/6582
>
> Which is very similar to what I get here:
>
> http://dspace0.lafayette.edu/handle/10385/516/statistics
>
> Could someone who is more of a programmer than me determine if this is
> related to the following bug reported back in March?
>
> https://jira.duraspace.org/browse/DS-509
>
> Eric
>
>
>
>
> On Fri, Oct 15, 2010 at 11:43 AM, Eric Luhrs <[email protected]> wrote:
> > I have noticed some oddities in the statistics I just imported into
> > SOLR.  I hesitate to compare results from Google Analytics to SOLR but
> > these are the only tools to which I have access.  Here are a few of
> > the oddities that I have noticed so far (from the same item in both
> > reporting tools):
> >
> > - Google shows no hits from Korea, SOLR shows Korea as the country
> > with the second highest number of hits.
> > - Google shows the country with the second highest number of hits is
> > Taiwan, but Taiwan doesn't even register in SOLR.
> > - All SOLR records that show "Unknown Country" also show a blank row
> > with the same number of hits under "Top cities".
> >
> > How do I verify that hits attributed to Korea are not actually coming
> > from Taiwan?  Any idea what's going on with with the blank City row
> > with hits that match "Unknown Country?"
> >
> > A related issue is granularity of geographic results.  For instance,
> > "Clichy" shows up in many of our records, which I assume refers to the
> > suburb of Paris.  Has anyone tried replacing GeoLiteCity with
> > GeoIPCity?  If so, did you get noticeably more accurate geographic
> > results?  And what changes were required to get GeoIPCity to work with
> > SOLR?
> >
> > Other than these issues, I am EXTREMELY happy with the SOLR approach
> > to statistics.  If I can resolve these issues, I think our faculty
> > will be very pleased.
> >
> > Eric
> >
>
>
> ------------------------------------------------------------------------------
> Nokia and AT&T present the 2010 Calling All Innovators-North America
> contest
> Create new apps & games for the Nokia N8 for consumers in  U.S. and Canada
> $10 million total in prizes - $4M cash, 500 devices, nearly $6M in
> marketing
> Develop with Nokia Qt SDK, Web Runtime, or Java and Publish to Ovi Store
> http://p.sf.net/sfu/nokia-dev2dev
> _______________________________________________
> DSpace-tech mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/dspace-tech
>
------------------------------------------------------------------------------
Nokia and AT&T present the 2010 Calling All Innovators-North America contest
Create new apps & games for the Nokia N8 for consumers in  U.S. and Canada
$10 million total in prizes - $4M cash, 500 devices, nearly $6M in marketing
Develop with Nokia Qt SDK, Web Runtime, or Java and Publish to Ovi Store 
http://p.sf.net/sfu/nokia-dev2dev
_______________________________________________
DSpace-tech mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dspace-tech

Reply via email to