Hi Li I'm convinced that you will find that it is practically impossible to mass import (or even mass merge) a new road dataset into existing data when there is a large amount of existing data. The only practical way would be to delete all of the Victorian data and start again with the mass import.
This would destroy OSM Victoria. Using high quality external data to complete the existing data is a realistic and not to difficult job. Firstly any existing roads that lost their names in the licence change (or had only ever been traced rather than surveyed in the first place) will appear highlighted in OSMI and can easily be named using vicmap data by any number of armchair mappers. This is an easy and enjoyable task. Also the vicmapdata will probably have a lot of roads that are new or that have never been traced or surveyed in OSM. These can easily be identified by overlaying the vicmap data in one JOSM layer and the existing data in another. Using correct transparency and colouring the new roads stand out like.....oops. Then the vicmap data (on a road by road basis) can be click selected and merged into the existing layer with all its tags. If a whole neighbourhood is new then it may be merged as a unit. You will have to manually connect the new road/s to existing ones but that is also easy. Bing imagery (which is very good in Australia) can aslo be used, at the same time, to verify or tweak the vicmap data as it is manually merged. All this results in excellent road data which will be better than any other prroviders, by a wide margin. I do this every year that new TIGER data comes out and I make sure a few cities in the USA are updated in this fashion. Every time vicmap produce new datasets this process can easily be repeated, without having to do a whole new mass import. Of course, ideally, local mappers will have surveyed any new roads way before vicmap have produced a new data set. This certainly happens in Canberra - which reminds me, there may be three of four new roads open today - I'll go have a look in a minute or two. Given the current amount of good quality existing Victorian road data in OSM, the task of getting it perfect using vicmapdata will not take too long. Of even more use would be if there is cadastral data available. This is absolutely essential if OSM is to ever succeed and this data is not easily surveyed. Walking around with a pad an pencil peering into people's letterboxes or doors is not safe or enjoyable. Therefore this data is ideal to be mass imported. (especially as it would probably be high quality data). Even then you would have to be careful not to "step on peoples toes" who had already done a street or two. Maybe you could import address info on a street by street basis and if your import program found any existing address data, it could ignore that street and create a list somewhere that people could use to do a partial manual import from vicmap data. How you programmatically compare streets in OSM and Vicmapdata is the difficult bit. No one in OSM has ever been brave (or maybe clever) enough to achieve this yet and I certainly couldn't. Mass imports only can work if there is no (or almost no) existing data. USA found this out when the mass import of TIGER data, effectively destroyed the USA OSM community and it has taken nearly ten years to recover. Unfortunately the original TIGER data was/is very low quality in terms of geometry and this still plagues the USA data to this day (and probably for the next 5 years or so). I'm sure that people in OSM-US can help in converting vicmap data into an imagery layer (and may even host it for you as well :-) Cheers Nick
_______________________________________________ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
