Discussion about the semantic web

Submitted by naught101 on Tue, 04/15/2008 - 20:49

The following is a discussion from #swig on irc.freenode.org - the Semantic Web Interest Group. It's logged here if you don't believe me: http://chatlogs.planetrdf.com/swig/2008-04-15#T10-32-11. Edited slightly for clarity.

I think the semantic web is an extremely useful tool, but as I mention down the bottom, I probably would have agreed with Francis Bacon that cutting up animals in the name of science was a good thing at the time. For the record, I don't believe this.

Reading the comment first might help.

naught101: check http://www.semanticfocus.com/blog/entry/title/5-problems-of-the-semantic-web/

naught101: my comment down the bottom, would love feedback from anyone here

bengee: simplification is a feature, not really a problem

bengee: URIs and triples reduce the complexity to a level that computers can do useful things with it

bengee: e.g. <#product> :rating "***"; :rating "****"; :rating "**". what might seem contradictory to you may be very useful to an app

naught101: na, I wasn't talking about that kind of information bengee

naught101: Say philosophy for instance... let me find a nice quote

naught101: Only when the last tree has died and the last river been poisoned and the last fish been caught will we realise we cannot eat money.

naught101: * Cree Indian Proverb

naught101: Obviously a computer could use this sentence, but would it be able to use it a a way useful to humans?

naught101: Obviously it's not TRUE, as most of us already know we can't eat money

bengee: <#only1> a :CreeIndianProverb; rdf:value "Only when..." .

bengee: that triple could be useful for programs that list proverbs

naught101: yeah, sure, but that's triplification ABOUT the proverb, not about the information contained within the proverb

bengee: well, then you have to increase the granularity if your app wants to provide richer functionality

naught101: how do you mean?

bengee: extract more triples from the human-readable text

naught101: but what triples could you extract from a one-sentance text that has no quantitative truth, but which holds more qualitative truth than many many paragraphs of, say, a science text book?

bengee: exactly

naught101: huh?

bengee: you may misunderstand what the semweb is mainly for

naught101: sure, that's a true sentence :)

bengee: it's not for implementing automated philosophers, or compete with humans with respect to intelligence

bengee: well, OWL folks might disagree with me here ;)

bengee: the more rewarding approach (IMHO) is to think about use cases that semweb tech *can* enable/simplify, not to think hard about things that are near-impossible for computers in general

naught101: I agree. but I'm not talking about what the semantic web should be

naught101: I'm talking about possible problems with what it currently is

naught101: I mean, I don't want bite-sized chunks of information taking over the world of ideas

naught101: I think the philospohy or the public is degraded enough without chopping it into bits even more

bengee: oh, semweb tech can clearly improve the distribution and discovery of ideas

bengee: just like the web did

naught101: but it could also hide them

naught101: I don't think the 'web did, neccesarily

bengee: you just google'd WRT, no?

naught101: correct

naught101: I don't think finding an accronym compares to finding meaning in life.

naught101: (if I sound like I'm attacking the semantic web, I'm not, I'm just exploring ideas)

bengee: yeah, don't think I can contribute too much here, sorry.

naught101: no worries :)

naught101: I see it something like baconian/descartian science. it's useful for finding out the little bits of information, but it's not particularly useful for figuring out the interrelationships, or looking at the information holistically

naught101: I mean, for example, the semantic web can take information from a wikipedia article, but it couldn't write a wikipedia article

kjetilkWork: right

kjetilkWork: I don't think it is a very significant goal of the semweb to produce that kind of information

kjetilkWork: we have a billion people out there that can do that much better

bengee: the semweb can be a great aid in helping you write the article, though

naught101: sure, but thinking of the possibility the semantic web a large part of the web should probably include thinking about what it can't do, and how to not impede that work

kjetilkWork: rather than the AI world of natural language analysis to reason and infere relation, I think the semweb is much more about using the collective intelligence of all its users, i.e. real intelligence

naught101: bengee: yes, it could. it could also be a hinderance (information overload)

naught101: kjetilkWork: good. I like that. I just hope we're collectively intelligent, and not collectively stupid :)

kjetilkWork: well, that's what it means to me, at least

kjetilkWork: hehe, yeah

kjetilkWork: I think semweb can help us be collectively intelligent rather than stupid, though... :-)

naught101: I haven't got that far yet

naught101: :)


Why are we taking such a long journey to unravel a simple matter?

The "Semantic Web" was a poor moniker choice for the Web as a Strucuted Database. Oracle, DB2, SQL Server, MySQL are strucutred databases, but It isn't natural across these databases to JOIN records across DBMS instance boundaries (where the DBMS doesn't support federation) or across DBMS vendor and/or DBMS engine type boundaries. And all of this is before we look at LAN, WAN, and transport protocol matters.

Add the Linked Data Web (Layer 1 of the Semantic Web technology stack) and you get the Web as a Federated DBMS (Graph Data Model) that is endowed with HTTP based Record IDs (URIs/IRIs) that enable remote Record Referencing and Record JOINS across DBMS instance, DBMS vendor type, LAN, WAN, and transort protocol boundaries.

In short, via the Linked Data Web I can point to records in your Database from mine (or vice versa) without any adverse effect. The worst that can happen is that we agree to disagree about the claims (Triples) that exist in my database i.e. we just arrive at perspectives withing the context of open Discourse.

The Semantic Web is not about replacing human intelligence, it's about encouraging the application of Human Intelligence to what it does best: Reasoning (typically :-) ).

What you say is true. I'm just trying find holes in the apparent goodness of the semantic web. Any technology has unintended side-effects, and it's important to find out what those side effects might be.

For example, the humble shovel initially just allowed us to dig holes more easily, but then we realised that we could also dig bigger and better holes. Then we decided that the shovel wasn't big enough, and repeated the process until we finally came to the point where we are now, with huge disasters like the Aswan Dam.

I'm not saying that the semantic web WILL be a problem, or have problematic side-effects, I'm just asking how it COULD be. A question that should be asked of any new technology.