[foaf-protocols] P2P FOAF search

Luke Maurits luke at maurits.id.au
Fri Jun 5 12:14:40 CEST 2009


> What about queries that no nodes know the answer to, but could be
> answered considering the sum of their knowledge.

That's a really interesting point and example situation.

I think this is a separate issue to the development of the actual
network protocol.  In order to ensure inter-operability of nodes, we
basically just need to agree on something like this:

"If you get a query and you can answer it, send the answer to the
originating node.  Otherwise, decrement the TTL counter and pass it on
to other nodes"

The "if you can answer it" part can be left deliberately vague and then
each implementation of the network node can have its own criteria for
"can I answer this query?".  The simplest implementations could apply
the obvious criteria of "I can answer a querry iff I have all the
required knowledge in my local store".  More complicated
implementations could take a more involved approach and, like your
example, use their own subqueries in an attempt to answer the original
query.

Perhaps in addition to a standardised protocol for the actual
distributed searching, we could specify a standard way of specifying
search heuristics like your example, and then concrete implementations
of the network protocol could also load search heuristics from a file
and use applicable heuristics in preference to the simple approach of
only answering what one knows directly.

A good arsenel of search heuristics like this might play an important
role in the success of an eventual P2P search network.

It might be hard to start work on actually specifying a format for
heuristics before more work has been done on the actual network
protocol, but this idea is definitely worth keeping in mind.

-- 
Luke Maurits <luke at maurits.id.au>
http://www.luke.maurits.id.au
http://www.luke.maurits.id.au/foaf.rdf
CompCogSci | Crypto | Maths | Python | Unix


More information about the foaf-protocols mailing list