[Pellet-users] A lot of descendant queries

[Alan Ruttenberg]

Tidbit: GO cellular component branch: 800 explicit pairwise part_of  
relationships, ~20K inferred
(that one doesn't run out of memory before finishing)
-Alan

On May 3, 2007, at 2:03 PM, Alan Ruttenberg wrote:

> BTW, the reason I am doing this is to supplement the rdf we are  
> feeding the naive rdf store we are using, so that we are still able  
> to answer interesting questions.
> Another question: wouldn't this amount to computing role inclusions  
> on two transitive roles on instances?
>
> part-of < part-of o is-a
> +
> part-of < is-a o part_of
>
> (one of these rewritten to use inverses so it fits)
>
> -Alan
>
> On May 3, 2007, at 1:38 PM, bijan at clarkparsia.com wrote:
>
>> On Thu, May 3, 2007 1:20 pm, Alan Ruttenberg  
>> <alanruttenberg at gmail.com> said:
>>
>>> This would be equivalent to the calls I made *if* the sparql engine
>>> implemented TBOX queries, which I understand not to be the case in
>>> the current version.
>>
>> Though, we hope, it will be implemented, at least preliminarily,  
>> soon.
>>
>> I need to think a bit as to whether we'd need to special case this  
>> sort of query to use something like the classification code (a la  
>> Evren's suggestion) or not.
>>
>> Hmm. *Have* to is strong. Usefully do so, seems right. Basically,  
>> you'd want to *order* some of the values in one half of a join and  
>> use results from the prior row's evaluation to compute subsequent  
>> ones. Interesting. The sorting part is a bit like a sort join, but  
>> I don't know if any db technique tries to avoid computation of a  
>> row because it can figure it out from a prior row.
>>
>> The memory issues is separate of course.
>>
>> Cheers,
>> Bijan.
>>
>