Hey Amir,
this sounds like a bug. The counts are automatically skewed a bit in cases
like Google (cases that often occur with other words Google $X). Could you
please open an Issue on Github for this?
Best,
Jo
On Fri, Feb 14, 2014 at 12:07 PM, Amir Mohammad Saied <[email protected]>wrote:
> Hi,
>
> We're using a self-hosted instance of statistical backend, without
> context.mem (so to be able to run it with 4GB of RAM). And even without
> context.mem spotlight does an amazing job in resource disambiguation.
>
> There's a tiny problem though. With default spotter_thresholds values (1.0
> 0.2 -0.2 0.25), spotlight never annotates "Google" but annotates all Google
> products (e.g., Google Drive), and "Google Inc.". Reading spotlight's
> documents I came across this wiki page:
> https://github.com/dbpedia-spotlight/dbpedia-spotlight/wiki/ISemantics-2013-Supporting-Dataand
> the Challenges section.
>
> The part that made me thinking it might be relevant is:
>
> "After further investigation, we found that these were all cases where
> the surface form is a substring of another surface form."
>
> Could it be the reason?
>
> It's worth mentioning that increasing the last parameter in
> spotter_thresholds values (0.25) to 0.3 will annotate Google, but it does
> annotate some other words as well which are not desirable (e.g., it'll
> always annotate II as World_War_II).
>
> Another thing that I noticed is that even Lucene backend doesn't annotate
> Google unless the spotter is "CoOccurrenceBasedSelector".
>
> Cheers,
>
> amir
>
>
>
> ------------------------------------------------------------------------------
> Android apps run on BlackBerry 10
> Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
> Now with support for Jelly Bean, Bluetooth, Mapview and more.
> Get your Android app in front of a whole new audience. Start now.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk
> _______________________________________________
> Dbp-spotlight-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/dbp-spotlight-users
>
>
------------------------------------------------------------------------------
Android apps run on BlackBerry 10
Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
Now with support for Jelly Bean, Bluetooth, Mapview and more.
Get your Android app in front of a whole new audience. Start now.
http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk
_______________________________________________
Dbp-spotlight-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dbp-spotlight-users
