Rob Church wrote
I don't know if it's quite what we're looking for, but we could inspect the category browser feature (for efficiency and load etc.) and perhaps switch it on...?
Rob Church
I'm working with Tim on making the CategoryTree extension fit for use on Wikipedia. It would help to make navigating categories easier - and would hopefully help to find messy bits. It does not solve the problem in principle, i'm afraid - IMHO, only a "facette classification" could, i.e. having distinct hierarchies for time, place, topic, etc.
In my dreams we would get rid of categories and have a system of meaningful relations between articles - like the semantic mediawiki project.
-- Daniel
On Thursday 24 August 2006 14:16, Daniel Kinzler wrote:
Rob Church wrote
I don't know if it's quite what we're looking for, but we could inspect the category browser feature (for efficiency and load etc.) and perhaps switch it on...?
Rob Church
I'm working with Tim on making the CategoryTree extension fit for use on Wikipedia. It would help to make navigating categories easier - and would hopefully help to find messy bits. It does not solve the problem in principle, i'm afraid - IMHO, only a "facette classification" could, i.e. having distinct hierarchies for time, place, topic, etc.
Yes, I agree. A clean struturing via orthogonal features that could be combined and browsed easily would really be a huge step forward. The current chaotic categorization seems to be unavoidable since the current system only allows you to view exactly one category at a time. It suggest to make highly specific categories that are useful when viewed in isolation -- quite the opposite of a clean facetted categorization that is most useful when combining facetts.
In my dreams we would get rid of categories and have a system of meaningful relations between articles - like the semantic mediawiki project.
SMW tries to enable the dynamic combination of "facettes" so that it suggests to build complex things out of a few simple ingredients. Of course performance is an issue when doing this dynamically on user reqest! But I am sure there will be solutions for this.
Still I think that categories do have their place as well. Category-like structures ("classes", "concepts", ...) have been proposed and included in many approaches for knowledge representation. Even though Wikipedia-categories often are not classes in a strict sense, it can be practically useful to exploit some of the pragmatic aspects of classes when dealing with categories.
For example, when I load SMW-created RDF into the Longwell browser (a free RDF-based data browser), it immediately presents the categories to me, and I can e.g. click on "Category:Country" to get all countries. This works because SMW exports categories as "classes" and Longwell has a special understanding of "classes", knowing that they are important starting points for browsing data. You first select whether you are interested in "people" or in "countries", and only then start to look for more detailed features (like "birthdate" or "capital"). So I hope that categories will stay, but that their use will be "purified" as soon as other methods are available for expressing certain relations/attributes more naturally.
-- Markus
wikitech-l@lists.wikimedia.org