This is the historical archive of the now-inactive 'grubstreet' list.
Discussion on OpenGuides development has now moved to OpenGuides-Dev. Discussion on The Open Guide to London now takes place on OpenGuides-London.

[grubstreet] When Categories Shift

[prev] [thread] [next] [lurker] [Date index for 2003/3/18]

From: jo walsh
Subject: [grubstreet] When Categories Shift
Date: 19:30 on 18 Mar 2003
in effort to distract myself from impending war and suchlike, i was
thinking about grubstreet categories and meta-categories. 

a while ago i was playing with an ontology bot for spacenamespace; it
keeps track of the space ontology - which is a taxonomic, hierarchical
structure written in OWL[0] which can be extended to feature logical
rules, such as Pub must have the property of ServesAlcohol, etc. people
could use the bot to vote on new classes in the ontology - broadly, new
categories - but i didnt really think hard about how the ontology fitted
together, so the bot has lapsed back into alpha silence. 

discussing hex's experience of the ontology voting bot, he suggested that
it would be nice to integrate the Category hierarchy of grubstreet into
the space ontology. i've been reading the thread about autoindexing:
http://grault.net/cgi-bin/grubstreet.pl?Wiki_Discussion/Categories_And_Autoindexing
one thing that strikes me is i hit the same problem; whenever someone
makes a new type of thing in mudlondon, the ontobot doesnt know where to
put it in the hierarchy, and a human needs to go in and attach it to a
parent class or refactor subclasses. on balance, i think this isn't such a
bad idea, and like earle, it's the sort of thing i enjoy doing myself.  

so i have a few questions, tho appreciate this may be a bad time for
answers, what with earle having buggered off to abu dhabi to get married,
and all. 

- what's the current state of grubstreet thinking re hierarchical,
indented categories of categories? esp. in context of the new version  

- is there a workingness example of autoindexing that i could look at, and
to what extent is that in the CGI::Wiki version?

- might you consider using OWL for this? ;) given the OWL spec is still
subject to flux this is not a practical short-term suggestion, but it
might not be long.

from my POV, i am considering 'forking' the two versions of my
spacenamespace - the london model/bot backend, and the more ambitious,
collaborative UK model based on lots of other peoples data. 

the OWL ontology i made is here: http://space.frot.org/rdf/space.owl - i
derived this from classification codes in the data from  
http://www.nima.mil/gns/html/gis.html , then i have done a bit of by-hand
extension especially in the area of public transport... and there are a
few detached, half-broken bits automatically submitted by the ontology
voting bot. i'm thinking about restarting from scratch with a 'ground-up'
ontology a bit like the grubstreet Category_Category, just covering the
things in mudlondon... and keep the other more formal ontology with its
descriptions of geographical, 'hydrographic' 'hydrotaphic' features and
other arcane things, for the bigger model.

i would borrow-and-repurpose the grubstreet categories rather than reusing
them wholesale... as i would want to say things like 'JapaneseFoodShop'
and 'JapaneseFoodRestaurant' rather than 'JapaneseFood', which in a
logical sense would be referring to nigiri and natto and that.
so automatic updates/ fully machined infosharing wouldnt make sense, but
i'd still like some way of being notified as a human when Categories
shift.

i'm probably just thinking out loud, but curious for your opinions.

zx  
--
"Common sense won't tell you. We have to tell each other." -DNA


-- 
grubstreet mailing list
http://london.openguides.org/old-list-archives/