Category: Semantics

Marc Cantor recently sent an email to the RDFWeb-Dev mailing list regarding FOAF. I chatted with Marc about this offline, and also the concept of ‘verification’ of FOAF relationships.

The social aspects of the increased interest in FOAF, I’ll discuss in the Burningbird weblog, but there are some RDF components I wanted to touch on here.

For instance, Marc mentions setting some relationships as ’standard’, such as the following:

– they are acquainted with somebody
– they know somebody by reputation
– they know somebody in passing
– or that they don’t know somebody, but wish to

First, there is no ’standard’ with this effort – even RDF is a specification and not a standard, as with all W3C efforts. Secondly, when Marc discusses ‘extending’ the vocabulary, he’s not necessarily aware of the fact that the whole point on an XML vocabulary being based on the RDF model is that anyone can create their own vocabulary and combine it with FOAF if they so choose – the original FOAF vocabularly doesn’t have to be ‘extended’. FOAF, and in fact all RDF vocabularies are not the same as RSS 2.0.

In fact, any time there’s extensions in the world of XML, problems occur, so extensions and versions should be discouraged at all costs. Clarifications are good – extensions, and in particular, modifications and deletions are bad.

The relationships that Marc refers to are not coming from FOAF but from another vocabulary, what looks like a Relationship vocabulary. They’re being picked up through the increased use of several of the FOAF Tools including the Add-a-Friend.

What the FOAF folks are going to have to decide is exactly what it is that they’re describing. Just like with RSS, the proponents end up including items within the vocabulary just because it’s handy. For instance, there’s discussion about adding address and movement information – but what does this have to do with Friend-of-a-Friend?

The key to a successful RDF/XML vocabulary is to keep it small, and to the point, and focused on the data of the business of the vocabulary. The FOAF creators know this, and I know that this new found enthusiasm for FOAF won’t push them into rushing extensions into FOAF that are ill-thought. Dan Brickley’s already posted some good responses back to Marc, as has Morten Frederiksec.

It will be interesting to see what Marc has in mind for this, in particular from a ‘verification’ point of view. Whatever FOAF is, it isn’t the Web of Trust, and that’s got me a bit concerned at this point – folks trying to make it into such.
—–
PING:
BLOG NAME: Burningbird
URL: http://weblog.burningbird.net/fires/001367.htm
TITLE: I wanna hold your hand
IP: 69.10.138.64
DATE: 07/23/2003 07:07:08 AM
Networks of friends, or at least people that know each other, seem to be very popular lately – I had two invitations yesterday to two different ‘friends’ networks. While I appreciate the thought and the invitation – I really do – I declined both; no…
—–

Recovered from the Wayback Machine.

There seems to be a great deal of activity about FOAF lately, as pointed out by Danny Ayers:

The FOAF (“friend-of-a-friend”) page on the project Wiki is growing into a great link collection, so presumably there’s a fair bit of cross-project interest.

The FOAF project now has a new home page, a weblog, a Wiki and IRC channel (#foaf on irc.freenode.net).

It’s also getting additional use out at Technorati, according to Joi Ito:

Technorati reads the FOAF file from your blog and creates a profile. Your picture from your FOAF file and a link to your profile shows up when you appear in people’s cosmos listings.

These are all good activities, particularly from an RDF perspective. Unlike HTML, RDF isn’t going to come blowing through the front door – it will creep in quietly through the backdoor with different applications that are based on the RDF model, and use the RDF/XML syntax.

However, having said this…

FOAF, as with RSS, does not make the semantic web, all by them lonesomes. Marc Cantor, also getting heavily involved in the FOAF world, recently wrote:

The connections between (n)Echo-Atom and FOAF should be obvious. Now throw in some ThreadsML and RVW – mix it up and out comes a semantic web!

Uh, no.

FOAF is an excellent way of identifying people whom you say you know, with some assumptions, but no guarantees, that they know you back. There is no inherent basis of trust or indication of relationship with FOAF, until you expand on the understanding of what ‘knows’ means in FOAF.

If one were to make a decision on buying a car or a computer based on the fact that you know a person who knows a person who knows a person who knows another who made a recommendation about this car or computer – all relationships documented with FOAF – well then, you deserve what you get if you get a lemon. Trust is diluted with each level of ‘knows’, within the current FOAF vocabulary and existing implementations of same.

There is little semantics in FOAF beyond the fact that it helps to loosely tie together a network of people, and provide some additional information about the people. With this and something like ThreadsML, you can also eventually find out the conversations the person is involved in. Add in this some RSS/Echo/Atom/whatever information and you’ll have a better idea of what they’ve written lately – but all of this combined is not “the Semantic Web”. It’s all just a piece, and a small one at that, of what will eventually become the Semantic Web.

We need a Turing Test of the Semantic Web, a test by while you’ll know the Semantic Web exists when you can do _____ on the web. For instance, I’ll know the Semantic Web exists when:

I can search for a poem about the loss of freedom, and one that uses a closed door to represent this feeling, metaphorically. In addition, I only want to see poems that someone I know, directly or indirectly (3 levels down, show relationship) has either reviewed or recommended, or discussed at some time. My preference would be British poet, a Romantic. Perferably, I would prefer a style similar to Wordsworth.

When I can do this, and I don’t get a lot of crap back as a result, then I’ll know that the Semantic Web is here.