ontology INDEX MUDLONDON MUDENGLAND TECHDETAILS CONTACT NEWS LINKS TALKS PLANS
i screenscraped a list of station names from the search facility at rail.co.uk, and cross-referenced this with the set of the things typed 'Railway Station' in GIS, from my GIS-based model of england, and gave each station approximate co-ordinates of the nearest town, where i could infer it (i'm sure a few such inferences have gone wrong, but this model is growing, and wiki-like, mapping errors can be easily refactored).
i am mapping the rail network slowly, armed with this information. i don't find canonical versions of lines, or even scrapable web-based versions of timetables; all the rail system metainformation online sits behind html forms-based ticketing systems and 'journey planning' systems.
in the UK the railways system is split between the line operator, (formerly bankrupt railtrack, now network rail) and various train operating companies. there must be centralised timetabling of some sort for the journey planners, but you can never be sure you're getting the right information or the cheapest ticket.
the individual TOCs provide visual maps though, and in odd moments i have been filling in the routes in the england model. if you feel like helping with that, send a file in the format 'TOC Route Name:Station Name,Station Name,Station Name' - check with me first to make sure it's not already done. Alternately by hammering and scraping web-based journey planning systems i could work out connections and infer routes, but i'd rather avoid this; perhaps i'll crack.
Rail network mapping is partially motivated by a desire to semi-automate an understanding of psychological distance between places. this doesnt really speak to the car-oriented world, modelling of road networks - on the level of the london model. the england model itself is a cross-reference of GIS data from nima.mil, and information about the structure of england from knowhere.
that moves away from the simple idea of cross-referencing as applied to finding things on the web and putting them together - whether in a structured format like GIS or unstructured HTML - and putting the connections you have found back on the network. like the cross-referencing on the web being available in harder, more digestible form on the semantic web.
the shared modelling aspect of the mudlondon bot, the 'collaborative annotation' and use of obseved or articulated routes is another aspect of cross-referencing. in a future, one could set out with a default common or random view of the space, and filter it or add detail according to people you meet, or around areas you travel to.