[provenance-challenge] Re: clarifying queries

[Paul Groth]

some responses in line...

On May 27, 2009, at 10:34 AM, Joe Futrelle wrote:

> Not sure OPM structure is enough by itself to identify corresponding  
> nodes across traces of the "same" process given that OPM admits  
> various levels of granularity, specificity, etc.

I think this why Luc suggested that each team identify the artifacts  
for their opm trace.

>
> The challenge definition could have specified some annotations that  
> each team was supposed to produce and then we could use descriptions  
> like "the process whose X property has the value Y" as in PC1.

I think this is a problem that we should definitely take up, are there  
ways to commonly refer to values...? I mean right now it's hard to  
determine what content is available in the value field of each  
artifact, hence, your suggestion of using annotations. Does something  
like this need to be part of OPM.

Paul



> --
> Joe Futrelle
> Cyberenvironments and Technologies
> National Center for Supercomputing Applications
> http://www.ncsa.uiuc.edu/People/futrelle
>
> ----- "Paolo Missier" <pmissier at cs.man.ac.uk> wrote:
>
>> Luc,
>>  absolutely -- personally, this is a last but significant hurdle for
>>
>> me: coming up with query formulations on third party traces.
>> well, isn't this an example of "semantic queries" that we hope to
>> answer
>> using a purely structural approach? maybe good for discussion
>>
>> Cheers, -Paolo
>>
>>
>> Luc Moreau wrote:
>>> Hi,
>>>
>>> Running queries over other teams' opm traces is definitely a
>> challenge.
>>>
>>> To simplify the task, I thought it would be helpful if teams would
>>> publish
>>> the ArtifactIds/ProcessIds for the inputs that Paul specified.
>>> ... and why not, the Ids where results can be found.
>>>
>>> What do you think?
>>>
>>> Cheers,
>>> Luc

-------------------------------------------------------
pgroth at gmail.com http://thinklinks.wordpress.org