Yes, they always worked for the stat. backend at least. With the latest
code (you should probably build from source or download the jar in the demo
folder), you can pass a confidence= value (0.5 is default). With lower
confidence values, you will get more spots. But note that this is still
quite experimental.
Jo
On Wed, Apr 16, 2014 at 10:02 AM, Arantxa Otegi <[email protected]>wrote:
> Hi,
> Does the spotter_thresholds.txt file work for the old version 0.6?
>
> Cheers,
>
>
> Arantxa
>
>
>
> og., 2014.eko apiren 03a 10:23(e)an, Joachim Daiber(e)k idatzi zuen:
>
> Hi Arantxa,
>
> the current spotters in the statistical backend have a lower limit
> because the most common use case is to not have too many spots. As Alex has
> suggested, you can manipulate the spotters by changing the weights in the
> spotter_thresholds.txt file. Alternatively, you can provide you own spots
> to Spotlight and let it handle the disambiguation.
>
> By the way, is this Eneko as in Eneko Agirre?
>
> Best,
> Jo
>
>
> On Tue, Apr 1, 2014 at 3:33 PM, Arantxa Otegi <[email protected]>wrote:
>
>> Hi again,
>>
>> We would like to use a more promiscuous spotter for the statistical
>> engine (v0.6), which tries to link as many words in a sentence as
>> possible. We have tried to change the parameters in the configuration
>> files with no avail. For instance, we tried the following set of
>> parameters:
>> Support 0; Confidence 0; Coref. Resolution False
>>
>> For example, if we call to "spot" with this text (and the mentioned
>> parameters)
>> "President Barack Obama and Michelle Obama were living in the White
>> House"
>> we obtain this result:
>> <annotation text="President Barack Obama and Michelle Obama were
>> living in the White House.">
>> <surfaceForm name="Barack Obama" offset="10"/>
>> <surfaceForm name="Michelle Obama" offset="27"/>
>> <surfaceForm name="White House" offset="61"/>
>> </annotation>
>> which is OK for us.
>>
>> But if we try with this other sentence:
>> "President Barack Obama and Michelle were living in the White House."
>> we get this result:
>> <annotation text="President Barack Obama and Michelle were living in
>> the White House.">
>> <surfaceForm name="Barack Obama" offset="10"/>
>> <surfaceForm name="White House" offset="55"/>
>> </annotation>
>>
>> As we want to link as many words as possible, we would like to annotate
>> Michelle in this last example too.
>>
>> Do you have any hints to obtain this behaviour?
>>
>> Thanks!
>>
>>
>> Arantxa, Ander, Eneko, Jokin, Aitor
>>
>>
>> ------------------------------------------------------------------------------
>> _______________________________________________
>> Dbp-spotlight-users mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/dbp-spotlight-users
>>
>
>
>
------------------------------------------------------------------------------
Learn Graph Databases - Download FREE O'Reilly Book
"Graph Databases" is the definitive new guide to graph databases and their
applications. Written by three acclaimed leaders in the field,
this first edition is now available. Download your free book today!
http://p.sf.net/sfu/NeoTech
_______________________________________________
Dbp-spotlight-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dbp-spotlight-users
