Metadata and The Semantic Web problem

drillchinchillaInternet και Εφαρμογές Web

21 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

72 εμφανίσεις

Metadata and The Semantic Web problem

The semantic web breaks the internet model, it demands too high a level of precision and it
tries to wrest control from the user and place it in the hands of the model maker. Wont work.

It presumes that whoever builds
the taxonomy in the first place, knows and can define to
everyone’s satisfaction, what they mean by the taxonomy. Before we can come close to Tim
Lee’s idea, we have to have a taxonomy of taxonomies that defines how taxonomies
work and what their c
ommon terms mean and how they develop, define and disseminate their
uncommon terms. Sounds to me rather like TMBW (too much bloody work) for no effective
return, and don’t get me started on ontologies.

THEN they have to define the content not only in terms

that make current sense, but that
leave open the certainty that uses will be found for the information at some future date, that
have not even been thought of yet.

THEN we have the problem that creators of information can only vaguely tell us what that
formation is about. Here’s a conundrum; my wife delivered her PhD this year and one of the
biggest problems is finding people to evaluate it. At that level, it is reasonable to expect that
the candidate is working on something about which most of the world

hasn’t a clue and which
will require even highly informed people to learn significantly new material tangential to the
speciality that qualifies them to evaluate the paper in the first place. No doubt we all feel that
our work is seminal and suitably impr
essive to anyone who matters, but we are to say the
least biased about its quality and totally ignorant about how others will view it or use it in the

This is a problem because I see metadata as a container. The more perfectly you define it,
the na
rrower the container with the fewest holes. That makes it wonderful right now and a
closed book in a year. The value, even the purpose of the information is defined by other
people’s use of it and their opinion of it, not by the writer. Metadata cannot dea
l with that, nor
can it anticipate its importance in the future. The Gettysburg address includes this, “The world
will little note nor long remember what we say here”

wrong, what if Lincoln had been
responsible for the metadata?

THEN we have the very human problem that people lie. Metadata depends for its validity on
the owner being both accurate

honest about the content and anyon
e who relies on that
shops on the TV Shopping channel. Plenty of them, but would you depend on them?

From my bit at QuestNet:

Give Us the Tools

When Doc Searls & David Weinberger launched they used a telling
phrase. “Take the value out of t
he centre and you enable an insane flowering of value
among the connected end points.” Yes please, can we have that?

The watchwords are annotation, reputation and collaboration. We have to create
educational value at the ends; more learning focused, useful

tools that take
advantage of the characteristics of connectivity and networking, and let the bandwidth
demands follow.

Give us tools that manage information, not documents, they need to be reputation
aware annotation systems, with flexible tools for teach
ers and learners to assemble
arguments and interact with each other. And I want the control of them in my hands
as easily as I control a piece of chalk and a blackboard.

Annotation is more important than metadata. There’s a lot of talk about metadata,
logies and the semantic web. I don’t listen to much of it because it hardly matters
what you think your document or resource is about, or how you think it can be used,
or how good it is. What matters is what I think about it and how I use it, then what the

people I respect, and the people they respect, think of it and how they used it. That’s
why annotation is crucial, I need to be able to be able to make my own annotations
on, link them to the annotations of others and understand how the work enhances or
iminishes your reputation in my chosen field.

Tools like that, attached to everything would be good, and ways for them to draw
horizons defined as I require and then to maintain those horizons in the way that
Kazaa and other file sharing P2P system work is


Google gets it. Google doesn’t understand particle physics, rare plant physiology or
the life of Van Gogh, but if you ask it a question on that subject, it returns very good
results. That is because Google understands the internet economy of l
inks and
opinions. That is why Google has bought Blogger, because the Blogosphere is a
snake pit of densely linked opinions and it is from that very dense web that Google
draws for its services. It doesn’t matter what the content of a document is, beyond
ome basic keywords, Google doesn’t even read the meta tags. What it does is read
its relationship with other documents on the same subject and, using the economy of
the net, figure out how respected that information or informant is. Then it publishes
the r

As Blogs experiment with Trackback and other linking concepts, the possibility is that
the web will turn into an isometric network with ranking and rating built into the links
themselves. At that stage the emergent intelligence of the internet will

take off and
Google will be there to use it.

Ontologies and taxonomies are observations that arise from thinking about language, they are
so difficult because language is an abominably complex process that lugs around information,
manages relationships, e
ncodes emotions, reveals, conceals, confounds and transmits secret
messages in plain text. BUT, it is so simple that a child can learn it, in fact, any language that
can’t be picked up in all its complexity by a child will die. Now, if you believe like Noa
Chomsky, that we are born with an innate language module or set of processes that guides
our acquisition of language, maybe the idea of extracting these into some formal structure
and then trying to shove them back in makes sense. Not to this pixie.

On t
he other hand, if you go with Terence Deacon in The Symbolic Species, language and
the brain have co
evolved, then formal structures are useful for figuring out what happened,
but they are no use for making something happen.

Emergence, keep the eye on emer
gence, and give us tools to help make it easier.