[Corpora-List] Query-By-Example (QBE) -like GUI to query corpora ...

[Christian Chiarcos]

On Mon, 29 Nov 2010 14:23:14 +0100, John F. Sowa <sowa at bestweb.net> wrote:

> On 11/29/2010 5:51 AM, Christian Chiarcos wrote:
>>>  How on earth could you query texts semantically?
>>
>> Most query languages implement fragments of first-order predicate logic.
>> What's the difference ?
>
> The answer depends on what kinds of people are asking the question,
> for what purpose, with that tools, and with what theory of semantics.

Certainly true. I have the strong feeling this entire discussion arises  
out of a confusion between syntax/semantics of language (object language,  
which is subject to annotation) and syntax/semantics of query languages  
and corpus representation formalisms (meta language). Using OWL or Prolog  
to represent and to query corpora (on the level of meta language) is fully  
legitimate, and it doesn't necessarily imply that any semantic reasoning  
on the level of object language takes place.

The study of natural language semantics is, of course, a fascinating field  
on its own, and I absolutely agree in the assessment that it is essential  
to the ultimate goal of NLU/NLP, but this kind of question is rather  
independent from infrastructural aspects of corpus management and corpus  
querying. And as I understood that this was what Albretch's question was  
referring to.

And with respect to this, the difference between query languages lies in  
their expressivity, but not that much in their functionality. In this  
function, OWL/RDF just provides us with another formalism - but it's  
highly generic and that's what makes it particularly interesting for the  
representation and querying of multi-layer annotations.

Christian

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora