spacenamespace
news from spacenamespace
2003-07-25
karosta
i spent some time at the locative media workshop in karosta, latvia hosted by the obliging RIXC, a kind of latvian mute with oopmh.
opportunity so see some ideas in new light, to see some take hold and others appear impenetrable. time tied to the keyboad, designing locative packets for spatially annotated media. wandering around the space, not dreaming up stories or planning scenes or imbibing audio strangeness, but modelling and re-modelling the space in RDF...
2003-07-11
libresoftwaremeeting
mudlondon was featured on the free software for geographic applications track at libresoftwaremeeting in metz, france.
it was an interetsing session in a slightly chaotic context;i learnt the strange military history of GRASS GIS, and was treated to a walkthrough of a solid reference implementation of Web Map Server / Web Feature Server - deegree, by the very cogent lat/lon
the semantic-web-n-bot work felt tangential to the direction of the track, but it was a fascinating day just the same, and i hope not to have lost the audience too completely. presentation slides available here are a slightly earlier version than the final ones, which were lost.
2003-06-17
a visit to the ornance survey
i paid a visit to the ordnance survey , to talk semantic web and spatial modelling...
2003-06-08
OSCOM redux, cartographic congress barreling along
talked about collaborative mapping on the semantic web at oscom2003; oscom slides (pdf, gzipped); basically the same as xmleu, with the noderunner in RDF material i ran past the london perlmongers putting the final spin on it. my oscom notes convey some of what i found out there; a very good one.
the talking's over for a while now i hope, and i can get back to doing things. the Cartographic Congress has been fantastic, action-packed, inspiring, tiring. the IAA residency providing much thoughtful food, and glimpses of lovely visual map rendering/annotation interfaces that could plug into a selection of RDF data sources rather beautifully.
'collaborative mapping' now seems to have memic self-sufficiency; the crisps list cooking up plans for our own london data garden, chris heathcote with thinking hat on, Kake cf grubstreet coding up a whirlwind. also connecting together properly the ideas in mapping contemporary capitalism with mapping space and time.
winding my head around new ways of modelling that allow for more interestingness, more context in the connections between places... starting to make more concrete my thoughts on remodelling space, now i actually have time and space to stop talking and start coding again. other than that, planning to spend a good chunk of the next week on my bike gathering gps traceroutes of london
2003-05-07
xmleurope talk, upcoming cartographic congress
i did a re-run of the same talk, except with a little more rambling detail on the subject of OWL ontologies, at xmleurope2003, entitled 'collaborative mapping with RDF. there are the slides, for completion sake
next week in london there will be a series of collaborative mapping workshops and events, which should be fun, interesting collections of people, and a game of wireless noderunner. the wireless ontology angle as it connects to spacenamespace, should be interesting...
2003-04-24
etcon talk slides
here is a gzipped PDF of slides for 'gonzo collaborative mapping on the semantic web at etcon
i had much cheerful positive feedback about it, and almost enjoyed doing it.
2003-03-29
in the grauniad
spacenamespace appeared in a recent article in the guardian online: get caught mapping, along with headmap, stef's conversations, edward's webmapper, et al.
that was peculiarly validating; name in lights, all that. i'm still playing with getting together an improved version of the england model, with a better API, hosted on the headmap box... talking to chris at landmap about ver offering a RDF interface to the wireless terrain plotting tools.
2003-03-16
mapinterest meetup that was
i meant to mention this before now; a week or two ago, a small clique of people with shared interest in open geodata and london mapping met up in a dingy pub backroom in covent garden. ben diligently took notes, and earle wrote a blog entry about it. i've been to busy panicking over [http://space.frot.org/talks/xml_paper.html|papers and presentations to do much about it yet.
i need a pen and tablet of some kind.
2003-02-27
virtual denmark!
morten frederiksen, one of the FOAF cabal, is making a virtual denmark - dk.space.frot.org. a node on the spacenamespace network! he's had to deal with a lot of issues i havent dared or thought to rigourously encounter yet, like internationalisation encoding in URIs, resolving clashes between spaces with the same name, etc.
i am so pleased and impressed with it, and feel oddly validated. i've been talking to a few people about putting together an informal RDF vocabulary for spatial annotation, building on FOAF and the developing work on the w3 geopositioning namespace. now i know what i'm looking for, i see more and more traces of collaborative mapping efforts on the web. interesting times...
2003-02-23
from DAML to OWL
i've moved the spacenamespace supporting ontology over from being expressed in DAML to be in OWL, the w3's Web Ontology Language which is nearing recommendation status.
i don't pretend to fully understand the deeper issues behind the general transition; OWL is a very similar specification, with some rethinking in the direction of implementation simplicity. It allows for three nested versions of implementation of inference rules - 'Lite', DL (Description Logic) and 'Full' - rather than DAML's one-shot, possibly overcomplex approach. bijan's working on some OWL reasoning code for swi-prolog, which i'm looking forward to immensely.
the transition was easy as my space ontology doesn't go far beyond a pure taxonomy, and didn't use any of the now-deprecated DAML properties. it's backed up by a consensus voting bot, which i'm still working to render actually accessible...
2003-02-06
routefinding and prolog
experimenting with routefinding in prolog, the interestingness and weirdness in the results, convinces me that i have the model of the tube system - wrong? unuseful? mismodelled, anyway. so i am thinking reified connections, with comments on them. [Bethnal Green] [connects] [Liverpool Street], and the central line is a property of that connection - or used to be, anyway.
until the prolog feels right, algorithmically and search-wise, i'll keep refining the model... i should also think about putting interfaces to various cities in the same bot ... e.g. danbri wanting his New Model Bristol with a monorail in it. What else is wishlisted? scriptable objects, tied into the ontology. more bots that talk to you and to each other, more of a conversational model. better use of FOAF to attribute comments on comments to different people.
2003-02-06
writing the web
giles turnbull asked me for a writeup of spacenamespace for his relaunched collaborative web editorial site write the web. so i wrote something.
2003-01-28
spatial logics, multimodelling
everything connects to everything else, in a brave new world of data homogeneity. but i've been thinking about remodelling the connections, now the map is growing and everything is at least nominally in a daml:Class, to reflect the connections in the world; and to allow them to be usefully commented. e.g., Fortis Green connects to Muswell Hill Road in a different sense than Fortis Green connects to The Copse. or Leytonstone station connects to a road, but only through one exit. how to represent such things in a machine-useful, as well as human-useful way?
adding new cities, new spaces; a brief trip to amsterdam this weekend for another foaf f2f seems a perfect excuse to start modelling mudamsterdam (mudadam?). so i've expand the space.frot.org namespace to allow declarations like http://space.frot.org/amsterdam/a_space/Leidseplein and http://space.frot.org/london/a_space/Leicester_Square . for backwards compat reasons, the original http://space.frot.org/a_space/ uri space will always point to london by default; as london is, after all, the centre of the world, this seems a reasonable asseveration.
i've been dipping my toes in the world of formal spatial logic and inference; traditional RCC logics for space; looking at the work of oliver lemon in manchester; breaking my brain with various papers that bijan sends me refs to. deeply fascinated, but my formal logic goes no further yet than hofstadter's TNT
mu
2003-01-16
svg map
i made a simple svg map of the model. it just shows points with co-ordinates, many of which are tube stations, from a distance.
2003-01-14
bot interface changes, thoughts on routes
some changes to the jabber interface after a lot of testing/playing with the bot by the helpful ian malpass.
on consideration, being prejudiced in favour of quiet bots by irc, i decided on his suggestion to make the bot a lot 'noisier', making a response, positive or negative to every message sent to it, cheerfully admitting that it don't know jack. not-understood messages get logged to a file, which i will look at, and hopefully the bot will be able to help with some day soon.
i also got thinking about reification of triples, and about routes. a route is an abstraction of a person's path, or instructions for following that path, like a sequence of beads on a string. my experimental model for routes is the same as the story sequence in the narrative engine behind the porn0graph.
this way, routes can be articulated, observed, or inferred, and represented with the same serialisaion. people seem to want route-planning. articulated routes - like cycle routes perhaps, or recomended walks - can be commented on. and reification should allow for comments on connections - to connect Old Street to Old Street roundabout via exit 3, for example. This is perhaps too loose a way to deal with the idea of exits and entrances which featured in danbri's old sketch of MOO RDF but don't really fit with the simple model of connections i'm using; an exit may be an interstitial space between two rooms, or a building and a street, but there's no exit or interstitial space between roads, one would say they were directly connected. typing connections soon introduces an amount of complexity in the model i'm keen to avoid.
i wonder whether to re-model train lines as routes, or rather to make routes co-existent with them. i've been starting to cross-reference map the train network.
2003-01-12
W3C geopositioning namespace
spacenamespace was involved in the creation of a w3 xml namespace for geo positioning. it is very simple and includes just a predicate fr latitude, longitude and latitude in WGS84 format.
i adjusted the spacenamespace schema and API so everything uses the same geo co-ordinates and also just talks to the one development schema document