Cycling Ontologies

As cycling is one my of favourite hobbies during the summer I have long considered whether a GPS system for navigation would be worthwhile (probably in combination with a PDA). So far I have resisted as these systems (PDA+GPS) are about as expensive as a new bike and it is still not clear to me whether the maps cover the small cycling roads.

GPS is obviously the high-tech solution to the navigation problem and printed maps are the low-tech solution. As may be obvious cycling and looking at a printed map do not get along very well. For the time being I have settled on a medium tech solution that is simple but very effective. The steps are: (1) define an unambiguous vocabulary for navigation; (2) write down the route using this vocabulary; (3) print it in two-column format; (4) fold it twice; (5) put it in your breast pocket and, finally, (6) on your bike.

Of these steps the first two are obviously the most critical. In order to develop a "cycling navigation vocabulary" I had a look at: Fietsroutes vanuit Lonneker. This site contains many cycling routes in Twente. Yesterday I cycled the shortest route ("Door Drienerlo") to test whether the vocabulary used for navigation was appropriate. There was one discrepancy between the directions and the route I took. Today, a very nice evening for cycling, I took a longer route ("De hoge Lutte"), and more importantly, including some roads I did not already know. Amazingly, I followed the entire route as described without any problem. Clearly, Paul Swaters, the author of these routes, has at least some tacit knowledge on writing down cycling routes.

I would suspect that the vocabulary used by GPS systems is restricted to navigation (go left, go right, etc.) and street names. This is probably sufficient for main roads. However, many cycling routes go over roads that don't have names, so the cues have to be more subtle. Two interesting cues used by Swaters are: describing the type of road and using the notion of "named" versus "unnamed" roads. For example, "end of cycling path, turn left, sandpath" and "at the first named road, turn left".

A full ontology of the vocabulary Swaters uses will take a little more thinking. Given that the instructions work so well, it would be interesting to study them further and compare with other descriptions of cycling routes. By way of some groundwork I loaded the routes into my text analysis tool, and it considers the following terms to be the most important (Dutch followed by English translation): weg (road), weg volgen (follow road), volgen (follow), voorrangsweg oversteken (cross main road), kruising (crossing), pas op (watch out), verharde weg (solid road), fietspad (cycling road), etc. From a linguistic point of view it is notable that most of the instructions are not proper sentences at all (for example: "first road left").

Comments

[Nick Hyndman]

Bill, using GPS has transformed my approach to cycling. I use it all the time and cannot understand why so few other cyclists do. You do not need the latest expensive GPS with map displays, my basic Garmin GPS12 does very well. I would recommend getting the cheapest available that is capable of storing routes and connecting to a pc. The display maps on the more sophisticated ones are too small to read comfortably on a gps on the 'bars of a moving cycle. You do not need a pda either, unless you are doing some very long tours and need more memory than the GPS has built in. My GPS will navigate me around 300 or 400 km on small rural lanes using the 20 routes and 500 waypoints in its internal memory. If I need to store more, I back them up in my psion and download them at the side of the road. I use the Garmin mapsource software on my pc to compile and archive my routes. This does not cover the off road bridleways that I often use, but it is easy to enter waypoints taken from a large scale paper map. Have a look at the Garmin Geko 201. This is more modern than my model, but not too expensive. Hope this helps. Regards, Nick Hyndman, Sheffield, UK.