RE: peer-to-peer was Re: Distributed querying on the semantic web

Response to Jeremy's 3 models of architecture at the end.

On Fri, 23 Apr 2004, Jeremy Carroll wrote:

>James Lynn:
>> Jeremy Question/Suggestion at end -
>>
>> > -----Original Message-----
>> > 1: fully centralized
>> >
>> > 2: terms decentralized, meanings per term centralized
>> >
>> > 3: fully decentralized
>> >
>> > There are many who argue that (2) is the right way for the SemWeb.
>> > [Maybe I am going to argue myself into (2), I prefer the
>> > beauty of (3),
>> > but the engineer in me sees the attraction of (2)]
>> >
>> > Note that the choice of semantic architecture is orthogonal to the
>> > choice of distributed system architecture - it is possible to
>> > implement
>> > SemArch (3) on a mainframe serving the planet; it is possible to
>> > implement SemArch (1) on thousands or millions of peers
>> > circling the globe.
>>
>> How about a hybrid of (2) and (3). One would have the option of
>> availing themselves of "the beauty of (3)" while referring to
>> centralized definitions as appropriate or convenient. Isn't this
>> what we do in technical writing? We make up our own definitions
>> when truly neccesary but find it convenient and efficient to make
>> use of definitions from previously published papers.
>>
>That's nice. It seems, to me at least, to chime with Peter's view that
>authority and definition comes about in a social process. In general, such a
>social process *does* indeed create some places that are more authoritative
>than others, but is always at least a little anarchic.

I think in practice we live with (3) in the current SemWeb. MGET is an
attempt to provide a way of doing (2). But the linguist and historian in me
prefers (3). We seem not to have a clean architecture for the peer-to-peer
nature of the Semantic Web, but then we are prepared to substitute Google and
our private bookamrks for a real architecture to deal with getting
information from the ordinary Web. So we seem to be copying what we have
learned before, and seeing where it leads us.

As Jeremy said, this probably doesn't depend on whether the architecture
underneath is a single monolithic server for the planet, or millions of
distributed processors and data-stores on mobile phones. To that extent, I
think Jonathan's original question of whether the definitions would be better
off in one server or not is not quite the right one. Instead the relavant
question is the social policy and the expectation around controlling and
updating those definitions... (Just to try and more explicitly say what
Jeremy implied).

Cheers

Chaals

Received on Friday, 23 April 2004 13:39:34 UTC