Home Messages Index
[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index

Re: Semantic Searches - Knowledge Engines

__/ [Davémon] on Monday 10 October 2005 18:14 \__

> Roy Schestowitz wrote:
> Roy, it's nice to see some blue-sky thinking here.

Thanks for the compliment. The former follow-up interpreted my words as a
/personal/ grand plan that's intended to infiltrate people's privacy. That
was clearly a misleading misinterpretation.

>> As we all (should) begin to realise, data and software are
>> migrating to the Web so no more will we need to handle things locally,
>> i.e. on/from our hard-drives.
> Can't say I agree with that, for a whole host of reasons.

That'll probably be a major battleground for the desktop giant and the
alliance of Web giants.

>> Web 2.0 also comes to mind, but it's not related
>> directly to search engines. Now, here's a crazy thought...
> The way I see it (if I understand you correctly) is that the real
> problem is that computers cannot understand complex linguistic
> constructs ('natural language query'), nor human behavioral patterns
> ('who i am' in a purely deterministic way) in any kind of meaningful
> way, you're assuming that more processing power would solve that, but
> I'd argue that its not computationally possible because there are too
> many gaps in behaviour and communication that the human brain ignores.

Practically, it is more viable than most people imagine. Translating
sentences to first order logic is not too hard a task and in excess of
training data/input you can compensate for abundance of complex natural
language. In my days as a computer science student, we actually wrote some
programs to scan text and translate it to logic, predominantly using
compiler technologies. Issues then become languages, homonyms, synonyms,
subtle meanings and cultural factors.

Also remember that we transform into a _semantic_ Web so it contains more
information than we see. What about XFN, for example? Look at the
implication of links on ranks, reliability and relevancy.

> Thats not to say that the current state of artificial intelligence,
> along side folksonomy and contributor lead webservices could lead to
> some very interesting research.
> Re: http://iuron.com/ - why not a wiki (it's more web2.0!)?

Great idea. I'll do that once I'm done with that proposal. I have about 5
documents lying around at the moment (mainly the academic stuff), so it may
take time. I should probably stick to phpWiki because I have experience
with it.

> I've also been toying with an idea of DMOZ as a wiki - no corrupt
> editors, just people power. Or is that just plain silly?

Because of spammers, it would require even more maintenance. The idea of
self-maintenance relies on morals or a system that penalises the crooks.


Roy S. Schestowitz      | Useless fact: 85% of plant life in in the oceans
http://Schestowitz.com  |    SuSE Linux    |     PGP-Key: 74572E8E
  2:25am  up 46 days 14:39,  2 users,  load average: 0.49, 0.59, 0.49

[Date Prev][Date Next][Thread Prev][Thread Next]
Author IndexDate IndexThread Index