RE: NeuronDB RDF and OWL

> It would be immensely helpful, though, if these terms were to be used
> with their accepted meanings, as these are quite exact, relatively
> easy to define, widely understood, and have been standard in the
> relevant technical literature for about 50 years now. 

[VK] It is quite likely that other communities have different but yet similar
notions of various things. For example, Information Retrieval notions of
precision and recall are similar to notions of soundness and completeness
applied to information retrieval "algorithms". In general, notions soundness and
completeness are probably applicable across many contexts other than different
types of reasoners and theorem provers.

It is interesting to work at the intersection of different communities,
Knowledge Representation, Healthcare, Life Sciences and as it gives us an
opportunity incorporate and synergize different perspectives, even though
imperfectly initially, but as in all fields, subsequent observations such as
yours lead to further refinement and synthesis.

> At the very
> least, if you are using them with different meanings, please 'flag'
> this by adding a qualifier, along the lines of 'sound and complete
> with respect to our database' or some such, to give the reader a hint
> that you are intending to convey a divergent meaning.

[VK] This is an important suggestion to prevent confusion as we attempt to
synthesize two or more perspectives. The intention of course is to convey a
divergent but at the same time a "similar" meaning.

> Do you have some independent criterion for what is a 'correct
> answer'? If not, the claim of completeness is vacuous. If you do,
> what is it? Can you share it with us? To establish completeness is
> often quite tricky, and it would be good to see exactly what is being
> claimed.

[VK] The point of an "independent" criterion for a correct answer is very
crucial and honestly, we haven't reached that point of analysis. But as
we work on the demo for data integration to be presented at WWW2007, I presume
it would depend on the underlying semantics of the query language constructs we
use. And of course we are "closing the world locally" as discussed in an earlier
e-mail on this list.

To your point, it is interesting to note that precision/recall metrics in
Information Retrieval are typically based on user judgements (e.g., TREC)

Thanks,

---Vipul





The information transmitted in this electronic communication is intended only for the person or entity to whom it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of or taking of any action in reliance upon this information by persons or entities other than the intended recipient is prohibited. If you received this information in error, please contact the Compliance HelpLine at 800-856-1983 and properly dispose of this information.

Received on Friday, 16 March 2007 16:39:07 UTC