Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Quite. The various OS comparison tools are enough to know which roads need checking without removing the name. Tom Sent from my phone. On Jul 15, 2010 12:02 AM, Tom Hughes t...@compton.nu wrote: On 14/07/10 23:44, Ed Avis wrote: I quite agree, there are plenty of errors in the OS data and ev... Nobody is arguing that the names shouldn't be rechecked where there is a conflict - they're simply saying that it is daft to remove the current name before the resurvey has been done. Tom -- Tom Hughes (t...@compton.nu) http://compton.nu/ ___ Talk-GB mailing list Talk-GB@openstreetmap.org htt... ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
OK, I will change my way of working for this task. The name tag will hold the most likely name based on currently available information, even if it needs a resurvey to be certain. -- Ed Avis e...@waniasset.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
On Wed, Jul 14, 2010 at 06:14:42PM +, Ed Avis wrote: ael law_ence@... writes: And I would take a dim view of anyone removing names that I have entered from gps and photos unless there was a very good reason. A different name in the Ordnance Survey maps is such a very good reason. That a good reason for *changing* the name, not removing it. Applications like navit can't be expected to look at FIXME comments. A name matching what is on the street signs is more useful to someone using a sat nav than the official name. Not that I am objecting to correcting it, just removing it from the name tag. I think that you are not considering the whole range of applications of osm. ael ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Thanks for getting the OS Locator tiles updating again. Could I make a feature request? Often when OS and OSM disagree I will tag this in the OSM database with a note such as FIXME=Check name - OSM has Marefield Gardens, OS has Maresfield Gardens Usually I will also delete the name= tag, so that the street shows up in noname checks to be resurveyed (and because the correct name is unknown). It would be useful for these already-looked-at cases to be excluded from the Locator check, since they are being flagged separately by noname checks. I know this was briefly discussed earlier on the list. We could spend all month discussing a suitably elaborate tagging scheme of fixme:name:OS_OpenData_Locator:resurvey=yes;osm_value=x;os_value=y. However, I propose not inventing any new tagging for this. Rather, look to see if the OS name is mentioned as a substring in one of the tag values. That would show that somebody at least is aware of it, and would catch various tagging schemes including the FIXME one I've been using. So, if the street has no name= tag, but does have some tag mentioning the OS street name, then consider it matched against OS for the purposes of the Locator comparison tiles. Would other people also find this useful? -- Ed Avis e...@waniasset.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Often when OS and OSM disagree I will tag this in the OSM database with a note such as FIXME=Check name - OSM has Marefield Gardens, OS has Maresfield Gardens Usually I will also delete the name= tag, so that the street shows up in noname checks to be resurveyed (and because the correct name is unknown). Isn't that last act a little rash? You then create a system where somebody goes out and maps stuff, but then whenever we have another source (of which there are a growing number) which disagrees, data gets cleared, then maybe restored later after somebody goes out there again? How about in any case of disagreement you flag it, but leave the data *exactly as it is* until someone can go and check it? Apart from anything else, there are people *using* OSM data (me for one). If they look at the main site, or a Garmin map generated while the name was removed, what do they get? In this case, you've just removed a useful street name because you had a small doubt that it wasn't perfect. As another suggestion: Is there a special tag we can add to force it to highlight on the noname check without actually destroying the name tag? Tom ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Tom Evans tom.evan...@... writes: Often when OS and OSM disagree I will tag this FIXME=Check name - OSM has Marefield Gardens, OS has Maresfield Gardens Usually I will also delete the name= tag, so that the street shows up in noname checks to be resurveyed (and because the correct name is unknown). Isn't that last act a little rash? You then create a system where somebody goes out and maps stuff, but then whenever we have another source (of which there are a growing number) which disagrees, data gets cleared, then maybe restored later after somebody goes out there again? How about in any case of disagreement you flag it, but leave the data *exactly as it is* until someone can go and check it? I've now resurveyed many street names around London where the OSM name was different from OS. Ten to one, the Ordnance Survey data is correct and OSM was wrong. Only in a tiny number of cases did I find that OS was incorrect (and tagged not:name for them to fix it). I will only tag names for resurveying round my local area where I know that I'll be able to go out and check them before too long. In general, I suggest that the procedure to follow depends on an estimate of the relative quality of the two data sets. I certainly wouldn't remove the OSM name tag just because of some suspicion that it might possibly be wrong; in these cases the right thing is to tag a FIXME and leave the data in place. But if it's more likely than not that OSM is incorrect, is there really a strong reason to leave the probably-wrong name in place? Surely incorrect data on the map is just as bad as missing data? As another suggestion: Is there a special tag we can add to force it to highlight on the noname check without actually destroying the name tag? That would be a good idea. However, if you think it's important that the name be left in place rather than removed, I would suggest the best thing to do is to put the Ordnance Survey name in as 'name' and leave the former OSM name in a note. At least in London, my experience is that the OS data is generally of higher quality, and if you had to pick one or the other, I'd go with OS by default unless it's specifically tagged that OS is wrong. -- Ed Avis e...@waniasset.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
On Wed, 14 Jul 2010, Tom Evans wrote: Often when OS and OSM disagree I will tag this in the OSM database with a note such as FIXME=Check name - OSM has Marefield Gardens, OS has Maresfield Gardens Usually I will also delete the name= tag, so that the street shows up in noname checks to be resurveyed (and because the correct name is unknown). Isn't that last act a little rash? You then create a system where somebody goes out and maps stuff, but then whenever we have another source (of which there are a growing number) which disagrees, data gets cleared, then maybe restored later after somebody goes out there again? How about in any case of disagreement you flag it, but leave the data *exactly as it is* until someone can go and check it? That sounds like a much better idea to me too. I thought some people already used not:name as well? I would find it a bit rude of people removed data that I put in; especially when it's surveyed and OS is just wrong. And it really doesn't matter too much if there is an ' missing or a slightly mistyped name. The map is still usable with that. regards, Derick -- http://derickrethans.nl | http://xdebug.org Like Xdebug? Consider a donation: http://xdebug.org/donate.php twitter: @derickr and @xdebug ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Tom Evans wrote on 14/07/2010 09:19: As another suggestion: Is there a special tag we can add to force it to highlight on the noname check without actually destroying the name tag? Tom Perhaps what is required is a new hierarchy, like: query so you might then have query:tag_name Not sure how tags work with multiple instances, but I guess you might have: query:name;surface and perhaps sometimes you might want query:path or query:route where you are not happy that the map properly reflects what is on the ground. Spenny ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
I will only tag names for resurveying round my local area where I know that I'll be able to go out and check them before too long. Fair enough. If you're always fixing it soon afterwards it doesn't really matter much. But in that case why not just go do it - why change the database first? If you tag the discrepancy you're only likely to attract somebody else's attention (which you don't actually want if you're definitely going to check it yourself soon - no point in both of you resurveying the same place straight away). In general, I suggest that the procedure to follow depends on an estimate of the relative quality of the two data sets. I certainly wouldn't remove the OSM name tag just because of some suspicion that it might possibly be wrong; in these cases the right thing is to tag a FIXME and leave the data in place. But if it's more likely than not that OSM is incorrect, is there really a strong reason to leave the probably-wrong name in place? Surely incorrect data on the map is just as bad as missing data? Regarding the last point: Are most of the incorrect entries not just a little bit wrong? i.e. misspelt, but still better than missing data. I see your argument about the relative likelihood of the two answers. I think replacing with OS would be better than temporarily removing the name entirely. Something rankles about overriding the guy/gal who did a ground survey without going to the spot to check first though. Tom ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
I'd go for fixme:name=OSSV has xyz or (if you really think OSSV is superior) fixme:name=was xyz but I've assumed OSSV is correct pending survey Richard On Wed, Jul 14, 2010 at 10:19 AM, Ian Spencer ianmspen...@gmail.com wrote: Tom Evans wrote on 14/07/2010 09:19: As another suggestion: Is there a special tag we can add to force it to highlight on the noname check without actually destroying the name tag? Tom Perhaps what is required is a new hierarchy, like: query so you might then have query:tag_name Not sure how tags work with multiple instances, but I guess you might have: query:name;surface and perhaps sometimes you might want query:path or query:route where you are not happy that the map properly reflects what is on the ground. Spenny ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
On Wednesday 14 July 2010, Ed Avis wrote: Thanks for getting the OS Locator tiles updating again. Could I make a feature request? Often when OS and OSM disagree I will tag this in the OSM database with a note such as FIXME=Check name - OSM has Marefield Gardens, OS has Maresfield Gardens Usually I will also delete the name= tag, so that the street shows up in noname checks to be resurveyed (and because the correct name is unknown). It would be useful for these already-looked-at cases to be excluded from the Locator check, since they are being flagged separately by noname checks. I know this was briefly discussed earlier on the list. We could spend all month discussing a suitably elaborate tagging scheme of fixme:name:OS_OpenData_Locator:resurvey=yes;osm_value=x;os_value=y. However, I propose not inventing any new tagging for this. Rather, look to see if the OS name is mentioned as a substring in one of the tag values. That would show that somebody at least is aware of it, and would catch various tagging schemes including the FIXME one I've been using. I really think this is exactly the sort of thing that does not belong in the OSM database, which is why I am working on a separate but connected database of manually overridable match states. Development isn't as fast as I'd like it to be due to work constraints and my home processing power being limited (testing can take a while). robert. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
On Wednesday 14 July 2010, Sam Larsen wrote: With all this talk of changing street names, can i just remind you to make sure that if you are changing street names that there are no addresses liked to that street. I have added many addresses linked to streets using Karlsruhe schema (without relations) - i guess this is where relations would help. I just did it the way the germans did it - they seem to know what they are doing. If there are, either change them also, or add a fixme tag or something. I must admit I hadn't thought about this. This would complicate the situation slightly. Perhaps this is the job of some other analysis tool that simply checks a nearby same-named highway exists for every stated addr:street. This would prompt on situations where the highway was fixed but not the addr:street. Does keepright do something like this? robert. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Robert Scott li...@... writes: With all this talk of changing street names, can i just remind you to make sure that if you are changing street names that there are no addresses liked to that street. Perhaps this is the job of some other analysis tool that simply checks a nearby same-named highway exists for every stated addr:street. Yes, this would be a good idea. I'm not aware of any tool that does it; Geofabrik's OSM Inspector has a 'street not found' check for addresses but I believe this just flags where no addr:street tag exists. In practice, where differences arise between OS and OSM data and it is necessary to change OSM, this is in areas without addresses. The more thoroughly-mapped parts of OSM have higher data quality and tend to reconcile cleanly with OS. It's also unlikely that anybody would manage to get the wrong street name repeatedly, both for the street and for buildings along it; and if I did happen to see such a case, I'd assume that the extra address data gives more chance that OSM is correct and OS is wrong. (The above generalization is based on central and north London - other parts of the country may have different data quality in both OS and OSM data sets.) -- Ed Avis e...@waniasset.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
On Wed, Jul 14, 2010 at 09:19:31AM +0100, Tom Evans wrote: Usually I will also delete the name= tag, so that the street shows up in noname checks to be resurveyed (and because the correct name is unknown). Isn't that last act a little rash? Apart from anything else, there are people *using* OSM data (me for one). If they look at the main site, or a Garmin map generated I agree: I use navit and gosmore on a satnav. Suddenly having names vanish unless they are manifestly wrong is seriously antisocial. And I would take a dim view of anyone removing names that I have entered from gps and photos unless there was a very good reason. ael ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
ael law_ence@... writes: And I would take a dim view of anyone removing names that I have entered from gps and photos unless there was a very good reason. A different name in the Ordnance Survey maps is such a very good reason. I know we all like to imagine that our own data quality from many rain-soaked trudges through the streets with GPS receivers must automatically be superior to all outside sources, but sadly it just isn't the case. From the streets I've rechecked in London, it is far more likely that OSM is wrong than that OS is wrong. This also includes streets that I have mapped myself from surveys and written notes. It can be something as simple as clicking on a road to add the name to it, not realizing that the road has two names along its length, and you only saw the sign at one end. This is a fairly common occurrence in OSM data; yes, the name was surveyed from photographs or eyeballing, but yet it is not entirely correct. Of course, the original work of surveying is not lost. Both the original name entered into OSM and the name found in OS are preserved in the FIXME tag. But if the name is most likely incorrect, it surely shouldn't be in the 'name' tag. -- Ed Avis e...@waniasset.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
Ed Avis e...@... writes: A different name in the Ordnance Survey maps is such a very good reason. I have found a significant number of mistakes in OS Streetview/Locator. Examples I have surveyed personally are Hanway Street in the West End where SV wrongly calls the Oxford Street end Hanway Place, Queens Road in Kingston which is spelt with an apostrophe in Locator but not the street signs and Ventnor Street in Manchester that is misspelt differently by Streetview and Locator. All three of these were mapped correctly at the time of the OpenData release. -- Andrew ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
I quite agree, there are plenty of errors in the OS data and even cases where Locator and Street View disagree. However that doesn't affect the general rule that, where OS and OSM disagree, it is more likely than not that OS is correct and OSM is wrong. At least based on the London streets I have resurveyed or otherwise checked so far. If OS were infallible there would be no need to mark things for rechecking - we could simply blast out all the conflicting names from OSM and use OS ones instead. This is not the case. -- Ed Avis e...@waniasset.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Map layer with OS Locator comparison from ITO - handling already-tagged fixmes
On 14/07/10 23:44, Ed Avis wrote: I quite agree, there are plenty of errors in the OS data and even cases where Locator and Street View disagree. However that doesn't affect the general rule that, where OS and OSM disagree, it is more likely than not that OS is correct and OSM is wrong. At least based on the London streets I have resurveyed or otherwise checked so far. If OS were infallible there would be no need to mark things for rechecking - we could simply blast out all the conflicting names from OSM and use OS ones instead. This is not the case. Nobody is arguing that the names shouldn't be rechecked where there is a conflict - they're simply saying that it is daft to remove the current name before the resurvey has been done. Tom -- Tom Hughes (t...@compton.nu) http://compton.nu/ ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb