Asserting subclasses of open ranges or domains

I had an interesting exchange on Semantic Web list on this issue last week. You can browse the whole thread, but I would recommend answers by Pat Hayes and John Sowa.
Below are extracts of my final answer.

1. There is quite a difference to make between concepts in ontologies strongly defined by domain experts, and targeted at feeding reasoners (e.g., bio-medical or legal ontologies), and lightweight ontologies such as FOAF, VCard, Dublin Core, Geonames ... which are mainly targeting interoperability of data, and of which meaning (if not formal semantics) emerge from usage and population. I can't define formally what a Person is, but I can say that you and I are some instances.

2. For the latter said ontologies, the main objective is to provide guidelines for applications harvesting and managing data. The actual formal semantics of those models is next to nothing, but implementations can reasonably leverage them on the basis of a common sense interpretation. For example the thousands of different data models to represent a person can be re-engineered as so many specifications of the generic class foaf:Person, therefore allowing a shallow, but efficient level of data interoperability.

3. There is no more, no less semantics nor potential usability in declaring skos:Concept to be in the range of dcterms:subject, than to declare foaf:Person to be a subclass of foaf:Agent. "Dont acte"

For some other ill-defined property ranges in the Semantic Web popular ontologies, another path would be to use enumerated classes, or in a more flexible way, to indicate a published vocabulary maintaining a reference enumeration. For example when LoC publishes later this year the authoritative ISO 639-2 list of languages as a SKOS Concept Scheme, the range of dcterms:language could be restricted to the values in such a list (using e.g., a restriction on the value of skos:inScheme). This would avoid Dublin Core to go through the painful task of defining formally the class dcterms:Linguistic System which is the current specified range - with the same lack of definition as foaf:Agent. Referring to some authority is certainly the best way to deal with the issue here. We (DC) don't know what a language is, go ask ISO 639-2 folks, apparently they know because they are able to provide a list.

From meaningless to ambiguous

A previous post looked at concepts as names species. Several threads about URI meaning and ambiguity later, including on www-tag@w3.org list this post from Karl Dubost, and with in mind again names and concepts as living things, here is yet another variation on morning, noon, and evening mountains.

Morning names, meaningless, useless

Names at noon, meaningful, useful

Evening names, abused, ambiguous

This basically is the life cycle of names from no use to use and eventually abuse. Use leads to meaning, then abuse leads to meaning overloading. This is the natural course of things. And since URIs are names, this is also the natural URI life cycle. So let us use what is meaningful while it is if we want meaning. Be prepared to ambiguity at the end of the day, but if evening ambiguous names have been abused, it does not mean they are useless.

URI species

The debate about proliferation of URIs representing the same thing keeps on rolling on various Semantic Web lists, going back again and again to the same questions. How does one discover existing URIs for a thing, if any? Is it a good or bad practice to mint a new URI for a thing which already has one? How do one link URIs identifying the same thing? Many smart and conflicting answers have been given, largely depending on the viewpoint on Web architecture and the main use of URIs in the mind of their authors. Web pragmatists and linked data evangelists tend to consider that proliferation of URIs is not necessarily a good idea, but something we are bound to live with, whereas experts in knowledge representation tend to consider it should be avoided by all means. Trust, persistence, quality of resource descriptions, use and abuse of owl:sameAs have been discussed over and over, with no obvious technical answer.

Since life provides the oldest, proven, efficient ways to store, maintain, replicate and use information, I've tried to figure if we could not learn from biology. Interestingly enough, biologists are not more able to come to a consensus about what a species is than Semantic Web gurus to agree on what is behind a URI. Somehow, the two issues are very similar. They deal with persistence of information over time. With the disclaimer that I am not a biologist, let me assume here the definition of a species as the set of individual expressions of some common genetic pool. Protection and persistence of the species genetic pool is the main occupation of any form of life. Strategies to achieve this goal present an awesome diversity, but in this variety one can find some constants. Among those are the basic facts that individuals are bound to a short life span, so the protection of the genetic pool is best achieved by assuming mortality of individuals, and ensuring duplication and replication of the information in as many individuals as possible. Not by defending a single representation behind firewalls.

How does that apply to the Semantic Web? A URI, along with the resource description it provides, can be seen as an individual expression of a species concept. As any human artefact, or any living individual, or any physical manifestation in this world, this expression is bound to be a transient. The agent who created and maintain the URI is bound to disappear, among other things. It will be less costly, as life tells us, to have copies of the information in as many expressions as possible all over the place, than to protect this specific one. Consider a URI not as the unique representation of a thing, but as an individual expression of a species.

Common Tag

Based on Common Tag specification and various APIs around, there are certainly a lot of easy next steps towards interfacing more efficiently the free tagging and linked data universe. For example it should not be too difficult to build an interface allowing the mapping of Twitter hashtags to DBpedia URIs, based on both Twitter and Zemanta APIs. Faviki could open this path.

Everything is a thing, everyone is many

So I'm now on Twitter, following a couple a people, hoping some interesting bits and pieces will float around to my shore. And well, some have. For example this interesting piece on Multiple Personas. What Silona writes there on the essential multiplicity of a person can certainly also apply to things and representations. We know more or less from inside our multiplicity and ambiguity, and the importance of keeping multiple personas revolve around our fundamental and essential emptiness. But we've lost most of the time the capacity to look at things the same way. As Jean Rousselot put it, we are not simple enough any more to "enter things as things can enter things"

Il faudrait pouvoir entrer sans frémir
Dans les choses
Comme les choses
Entrent dans les choses.

The poets are the ones able to enter things, experience their mutiplicity, and show how they appear to us as multiple personas.
We should look at the necessary convergence of the social and semantic web(s) with this paradigm of multiple personas in mind.

sameas.org

sameas.org is quite an implementation of hubjects for the linked data universe. It relies still a bit too much on owl:sameAs, but I begin to believe that owl:sameAs eventual semantics will be the one applications make of it. Only reasoners in closed universes will apply owl:sameAs for what it is in the standard (strict identity). Open Web and linked data cloud will use it as "follow-your-nose" hubs to switch from one representation to another and aggregate information.

WE are the hubs

We've spoken a lot here about identifying and linking things and data about things. But 2008 will stay as the year of linking people. Clay Shirky exposed that brilliantly in his book "Here Comes Everybody"
As hundreds of millions of people, I've engaged in a few social networks this year. Not that I was not in some on-line communities before. In 1999-2000 I was Open Directory editor, since 2001 I've been a lunatic Wikipedian. But it was not those fancy Web *.0 social networks making billions of dollars. I've never been in Facebook or MySpace or the like, never going over this a priori reluctancy to join in noisy, aimless social chatter, exchanging images, music and recipes. No thanks.
Nevertheless ... I joined LinkedIn in January 2008, mostly to update my address book. At the time there were 17,000,000 people "In". Now the figures are over 30,000,000, or so they say. At the end of 2009 it's bound to be 1% of mankind. But LinkedIn is mostly an ego-booster. The main purpose of being "In" is very selfish : look at how efficient and professional, how many people I'm connected with etc. There are groups, but they have no real social activity as far as I can tell. So I keep my LinkedIn account for what it's worth : a free human resources directory.
In spring I had a try at Twine. With good marketing, they attracted a lot of smart, brilliant people. It was an interesting experience, but mostly unfocused and time-consuming. Twine has no particular purpose beyond conversation. I was quickly fed up with that, so I left.
My favourite these days is WiserEarth. I was very impressed by the clarity of purpose, the quality of information, and the spirit of the community. Of course the numbers are still small, less than 20,000 to-date. But you're most welcome to join.