[pet] [itsdb] [incr tsdb()] and fsc test items

[Katya Alahverdzhieva]

Dear Stephan,

Thank you for your prompt reply!

I'd like to ask you for some further assistance concerning fsc test 
items in [incr tsdb()].

Here's what I've done: my item file is a single test item containing 
feature structures (see below). Then I copied the two functions as 
defined in

   http://wiki.delph-in.net/moin/PetInputFsc -> Integration [incr tsdb()]

to $LOGONROOT/dot.tsdbrc (I also modified the path to my .grm file 
appropriately).  I assume that to run the fsc test suite instance, it is 
not necessary to load my grammar in lkb since I'm using PET instead. So, 
with the root directory and the skeleton directory loaded, and with the 
test suite created, I tried processing the test item. The results 
contained nothing but empty cells.

I hope I'm not missing any relevant piece of information!

There's obviously something that I am not doing right, but it's hard for 
me to say what it is. I'd be great if you could help me solve this.

Thanks,
Katya
 
1 at author@@@1 at _external_argument_accented@<?xml version='1.0' 
encoding='utf-8'?><fsc version="1.0"><chart id="fsc-test-1"><lattice 
init="v0" final="v2"><edge source="v0" target="v1"><fs 
type="speech_token"><f name="+FORM"><str>she</str></f><f 
name="+FROM"><str>0</str></f><f name="+TO"><str>3</str></f><f 
name="+START"><str>33.15591</str></f><f 
name="+END"><str>33.26119</str></f><f name="+PHON"><fs type="lnr"><f 
name="+DOM"><str>she</str></f></fs></f></fs></edge><edge source="v1" 
target="v2"><fs type="speech_token"><f 
name="+FORM"><str>called</str></f><f name="+FROM"><str>4</str></f><f 
name="+TO"><str>8</str></f><f name="+START"><str>33.15591</str></f><f 
name="+END"><str>33.26119</str></f><f name="+PHON"><fs type="p-word"><f 
name="+DOM"><str>called</str></f></fs></f></fs></edge><edge source="v1" 
target="v2"><fs type="gesture_token"><f 
name="+FORM"><str>neverunify</str></f><f name="+GESTURE"><fs 
type="deictic-concrete-right-hand"><f name="RIGHT-HAND-SHAPE"><fs 
type="open-flat"></fs></f><f name="RIGHT-PALM-ORIENT"><fs 
type="towards-up"></fs></f><f name="RIGHT-FING-ORIENT"><fs 
type="forward"></fs></f><f name="RIGHT-LOC"><fs 
type="c_coord"></fs></f><f name="RIGHT-MOVE"><fs 
type="to-side"></fs></f></fs></f><f name="+FROM"><str>17</str></f><f 
name="+TO"><str>22</str></f><f name="+START"><str>33.03304</str></f><f 
name="+END"><str>34.60127</str></f></fs></edge></lattice></chart></fsc>@1 at 49@@KatyaSA at 25-Sep-2011



Stephan Oepen wrote:
> hi katya,
>
> i believe this should just work; have you tried?  [incr tsdb()] can store arbitrary strings in its ‘i-input’ field and will (unless configured otherwise) just send those strings verbatim to PET for processing.  in case you had ‘@’ signs (\s), newlines (\n), or backslashes (\\) in your inputs, these would have to be escaped in the tsdb(1) ‘item’ file, using the conventions indicated in parentheses above.
>
> i recall there used to be problems in sending PiC-format inputs to PET from [incr tsdb()], where i believe one had to pad the ‘i-input’ field with two final newlines; otherwise, the parser would hang inreading the PiC XML from the string received from [incr tsdb()].  i don't think this problem is there with the FSC format (though i have not tried myself, but believe peter adolph tends to use the exact setup you sketch).  if it were, please let us know (and we should fix it :-).
>
> i'm copying the PET list, in case other developers or users have further comments or experience reports.
>
> best, oe
>
>
>
> On 24. sep. 2011, at 12:33, Katya Alahverdzhieva <K.Alahverdzhieva at sms.ed.ac.uk> wrote:
>
>   
>> Dear All,
>>
>> I was wondering whether the test items in [incr tsdb()] can be feature structures instead of strings? So, instead of having "She barked" as a test item, I'd like the the test item to be augmented with arbitrary feature structures (like those inputted to PET), e.g.,
>>
>>   <edge source="v0" target="v1">
>>      <fs type="speech_token">
>>        <f name="+FORM">
>>          <str>she</str>
>>        </f>
>>        <f name="+PHON">
>>          <fs type="lnr">
>>            <f name="+DOM">
>>              <str>she</str>
>>            </f>
>>          </fs>
>>        </f>
>>      </fs>
>>    </edge>
>>   <edge source="v1" target="v2">
>>      <fs type="speech_token">
>>        <f name="+FORM">
>>          <str>barked</str>
>>        </f>
>>        <f name="+PHON">
>>          <fs type="p-word">
>>            <f name="+DOM">
>>              <str>barked</str>
>>            </f>
>>          </fs>
>>        </f>
>>      </fs>
>>    </edge>
>>
>> If you could give me some pointers to existing samples, that would be great! Thanks for the help!
>>
>> Cheers,
>> Katya
>>
>> -- 
>> The University of Edinburgh is a charitable body, registered in
>> Scotland, with registration number SC005336.
>>
>>     
>
>   


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.