----- Original Message -----
Sent: Wednesday, October 07, 2009 9:50 PM
Subject: Ontology Goverance
A discussion over a glass of wine last Monday
reminded me of a query that I hadn't time to put to the forum previously -
presented below as a vignette:
Site Bloodnock asserts both A and B
Site Moriarty asserts NOT A and M (but he is lying
about NOT A)
Site Eccles provides query answering services, in
which it takes a query and redirects it to sources that may be able to answer
Bluebottle is standing on a narrow bridge, high
above the seething waters of some obscure African river. The guardian asks him
"A and B and M?"
Bluebottle sends the query to Eccles who then
forwards it to Bloodnock and Moriarty, consolidates their answers and then
returns the answer to Bluebottle. There are three cases here:
1) Eccles sends "A AND B AND M?"
Bloodnock does not reply, since
he does not know M
Since Moriarty asserts NOT A, he
does not need to check B (or M), and so can reply FALSE
2) Eccles sends Bloodnock "A and B?" and gets the
Eccles sends Moriary "A and M?"
and get the reply FALSE (since NOT A)
3) Eccles sends Bloodnock "A?" and "B?", and sends
Moriarty "A?" and "M?"
In case 1 and 2, Eccles sends Bluebottle the reply
In case 3, Eccles can may do a simple
evaluation, and since (A AND NOT A) = FALSE, he sends false. However, he is in
the position to observe that Bloodnock and Moriarty are inconsistent. What
should he tell Bluebottle?
Question A: Is there any formal policy on whether a
site should check for inconsistencies and what it should do when one is found?
What would be the forum to raise such questions? Does this forum have any
views on this?
Question B: Can I lay claim to the discovery of the
first ontology virus? And how should the semantic web police itself to detect
and eliminate such viruses?
Sean Barker, Bristol
PS It is likely that someone from the UK and over
50 will spot the subtext, but ontologies don't infer subtexts, and even if they
did, it is up the knowledge user to infer as much or as little as they