Tomasz Wegrzanowski wrote:
On Tue, Jan 07, 2003 at 03:38:09PM -0800, Brion Vibber wrote:
On mar, 2003-01-07 at 14:40, Tomasz Wegrzanowski wrote:
- it might be good idea to consider making our own searching system instead of relying on something provided by database
Hell, why rely on a database? Why not write our own storage system? And filesystem... and virtual memory... and process manager... :)
We shouldn't have to be a wheel factory; if a good, fast, flexible search mechanism is available in our database we should use it.
Searching doesn't seem like something "general", it's almost always specific to given usage.
Examples of magic that would be nice to have (nothing urgent):
- On Polish Wikipedia, if someone tries a query and no hits are generated, and it didn't contain any diactrics, try to search undiactricized version of whole database. Some people for some reason write ASCII Polish while on Internet.
- Dictionary searches would also need more magic. See: http://wiktionary.org/wiki/Wiktionary:Alternative_vision_of_multilingual_dic...
Wikitech-l mailing list Wikitech-l@wikipedia.org http://www.wikipedia.org/mailman/listinfo/wikitech-l
Hi there,
I'm VERY new here. Just read Alternative_vision_of_multilingual_dictionary and I think these are 'wise words'. Think I agree with most things said there. It is indeed a very fundamental problem. :-) The more I think about it, the less I want to say about it. I am just wondering whether we can learn from Kohonen's Self-Organising-algorithm. Can we use neural networks (fuzzy logic, etc) at all in Wiki software? ? (In general, neural networks don't perform well on 'symbolic' data, I know that but maybe,.. with a lot of preprocessing,... gathering statistical data, link-counts, and such,.... maybe then, the software could be made as 'self-organising' as humans are? I've played around a bit with Self-Organising-Maps and pattern-recognition in 'raw' data. But I'm not too familiar with linguistics,... or the 'symbolic-approach'.
As Wikis evolve, we'll need more and more sub-structuring, and more and more 'protection'. (Like in Kohonen's SOM, where 'neighbourhood-space' decreases during self-organisation, more and more (groups of!) people will start using their OWN Wiki-site. Because Wiki-SOFTWARE does not reorganise itself, humans will have to do it
Someone here on this list familiar with Kohonen-SOM-stuff?
This is a 'tricky' discussion here, on this list.... very exciting yes, on the one hand, but I also think it is a bit of an 'endless discussion', it's a very fundamental problem.... :-)
Just keep adding more and more features to the software will not solve the problem. Better think about removing 'options for the user' than adding new ones. Very curious how Wiki will develop in the future.
Does anyone know if there are C or C++, CGI or standalone-server versions of WikiSource?
Thanks, Pieter Suurmond