Tomer Chachamu wrote:
On 29/07/05, Lars Aronsson lars@aronsson.se wrote:
And what if you discover such complex relations as the project develops, what is the UW strategy for modifying the schema over time? Right now you seem to be designing an "ultimate" schema that will then be frozen and kept static for all time. The very name of the UW project suggests this kind of thinking, and to me that is about as foreign as marxism.
As was done with WM 1.4 to 1.5, some very complex queries - and a lot of processing time :) - would be needed to change the database design. Therefore it would be undertaken with care and, well, it can't really be planned for.
Before MediaWiki 1.5 the same database schema had been used for a long time with only minor additions that didn't require conversion. But with UW it seems that bigger changes will be required more often, because so much detail about the structure of every human language needs to be encoded in the database schema. Oops, Hungarian has two kinds of plural, let's change the schema. Oops, Thai has genders for verbs, let's change the schema. Oops...
The axiom that it "can't be planned for" sounds like a recipy for failure.
Are there indeed any non-free (commercial or research) projects that have attempted anything like this? Wikipedia has many precursors such as Britannica, Brockhaus, etc. And so does Wiktionary, of course. But which precursors does UW have? What kind of data model or database schema do they apply? Or is UW a piece of original research in computational linguistics?