[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Semantic Web Hackings
- From: Danny Ayers <firstname.lastname@example.org>
- To: email@example.com, firstname.lastname@example.org
- Date: Tue, 02 Jan 2001 00:40:26 +0600
> Now, the URI mailto:email@example.com tells me absolutely
> nothing about
> you (except, maybe, that I can send a message to that address and
> hope that
> you get it) nor does it tell me how to get more information about you.
> Likewise, mailto:firstname.lastname@example.org tells you nothing about me. You can
> make all the assertions you want about your URI but if I don't have access
> to those assertions then they won't do me any bit of good.
> You keep saying that all I have to do is "query the Semantic Web." Again,
> pardon my practicality, but in order to do that, I'm going to need a host,
> port, path, some parameter names, and some inkling of an idea on what to
> expect as the result of that query so that I'll know how to parse it.
I wouldn't disagree that 'query the semantic web' is a little vague, but
this only means the details need filling in. I do think there is some merit
in the general idea of using mailto: to locate a person/information about a
person, and there are a lot of different ways a system using this could be
implemented. Ok, you've got mailto:email@example.com - you want a host,
what about injektilo.org, a port - call it port x, some parameter names -
how about something like 'lookup mailto:jason', for a return from the query
how about a url? So you would need a service running on injektilo.org to
handle all this - why not? Maybe it wouldn't be practical to try and extend
DNS for this purpose, but a more generic directory service along the lines
of LDAP (which might also include DNS) could do a lot of useful work for the
semantic web - far beyond domain name resolution. In any case, retrieving a
url from a mailto: name on an individual host would be pretty trivial,
though setting up a distributed system would be a lot trickier, but there
are analogeous systems around (like DNS, or even mail relaying) and so this
would probably be feasible.
> Today, my solution would be to use HTTP and RDF. It can't get much simpler
> than that. Assuming, of course, that I know the http URL that I can
> dereference to get your RDF.
But there isn't a problem if you know the URL, and if you don't have it then
this isn't a solution.
You could however hack it, by having one or more servers that would take a
http request, something like http://mailnamer/lookup?mailto:@injektilo.org
and responding by redirecting to a url. Directory services by stealth. How
do you get the data in there? For the sake of perversion, how about a
modified mail server that recognises messages with a particular string as
subject line as submissions, with the message body as a load of RDF? Ok,
I'll work out the security issues later...
If being able to locate people and structured information about people has
its uses (which it probably is, even if just to give bibliographic info),
then some standardised naming will be required, and some way of getting from
the name to useful information. There is already a partially standardised
form of naming in mailto: . The alternative would be to come up with some
kind of naming from scratch first and then sort out the rest. This may be a
better approach, but using mailto: probably means less work - that to me
sounds like a practical approach. So take the mailto:, query a directory
server with it to obtain a url, got to the url and retrieve your rdf.
> Now, if you want to add RDF RRs to DNS so that we could translate
> your email
> address into a a URL then you have my full support. But good luck
> the rest of the world to use them.
(maybe not DNS but same principle) If there is any benefit to such a system
no-one will need convincing.
> Or, if you'd rather use the out-of-band approach and just tell me your URL
> so that I can configure my Semantic Web browser to get all assertions for
> your mailto URI from such-and-such http URL, then I'll accept that, too. I
> think it's a terrible idea but it'd be usable for at least a handful of
> people. But don't we want the whole world using this?
Getting computers and internet connections to the whole world will be the
hard part ;-)