[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [help-GIFT] Re: Region query for MRML/GIFT

From: Wolfgang Mueller
Subject: Re: [help-GIFT] Re: Region query for MRML/GIFT
Date: Mon, 4 Mar 2002 14:49:24 +0100

Hi again,
> Hence, I believe that you agree with me about my protocol, so how about the
>  wording ? should I change any words ? If everything is ok, I will
> submit my DTD soon, after discuss it (again) with David.
Yes, I think it would be good to change some names (I hope you agree).

How about get-segment-by-url instead of user-segment-query (or some changes 
to that effect?)? I guess it would make clear that it is rather a short 
lookup than a full scale similarity query (even if both are, in fact, 

In terms of the query paradigm I am asking myself if it should't be rather

<collection collection-id="some-id".>
     <query-paradigm index-type="inverted-file" type="segment-query"/>

Rather than

<collection collection-id="some-id-2".>
      <query-paradigm index-type="inverted-file"> 
      <query-paradigm type="segment-query"> 

Imagine you are using some extended version of viper. That uses the classic 
inverted file accessor for accessing the index, and you put on top of that 
some region querying extension that does exactly the type of negotiation 

Now imagine, I manage to do the same thing using bothrops. In this we could 
imagine that I would like to specify in the configuration file some 
collection that is not indexed using the classic accessor, but rather the 
bothrops accessor, and that allows for region queries. If I'd specify that as

<collection collection-id="some-id-3".>
       <query-paradigm type="segment-query" index-type="bothrops">

Some algorithm written for bothrops 

<algorithm algorithm-id="some-algo-id">
         <query-paradigm type="segment-query" index-type="bothrops">

some other algorithm written for bothrops (your style).

<algorithm algorithm-id="some-algo-id-2">
         <query-paradigm type="segment-query" />
         <query-paradigm type="bothrops"/>

some-algo-id-2 WOULD MATCH some-id-2, although they are not indexed the same 

However, some-algo-id would match some-id-3, but not some-id, as the 
index-type attributes are conflicting. This is exactly the behaviour we want. 

Suggestions for other query-paradigm type and index-type attributes are most 

BTW I find your DIA documentation of the interaction really cool. 


reply via email to

[Prev in Thread] Current Thread [Next in Thread]