Changeset: 53114647
remove SEO entries that carried advertising in their description tag
Closed by woodpeck_repair
Tags
bot | yes |
---|---|
created_by | osmtools/30252 (linux) |
Discussion
-
Comment from muralito
I could provide some nodes that were deleted and were not advertising.
https://www.openstreetmap.org/node/1118036691
https://www.openstreetmap.org/node/5135199111
https://www.openstreetmap.org/node/5057146412/history
https://www.openstreetmap.org/node/5042809017/history
The 4 seems valid elements, and maybe in one of them the description could be considered as biased, but not advertising.
Please review how your algorithm is selecting the cases.
-
Comment from naoliv
It seems the same case from some nodes in Brazil.
I wouldn't call them *wrong* but incomplete and/or with unnecessary information.
(New) users sometimes misunderstand the "description" tag and are too much verbose when using it.
Instead deleting, is it possible to create a list of affected objects, for manual review?
Or maybe simply remove description=* from the objects?
For example, https://www.openstreetmap.org/node/4976338726/history seems to be a case where the new mapper misunderstood description and note (and overly described the object). -
Comment from funnimonk
please never do things like this again. you've basically deleted street names etc from the dataset. some of us rely on these things like names an geometries and now we have to restore them and you know how difficult that is!! to what end, all this deletion? couldn't you just change the descriptions??!! and why didn't you take it up with this mapper that you speak of??
-
Comment from woodpeck
Hello, I'm sorry if I have accidentally deleted things that were not advertising. I'll look into the IDs specified by muralito and replace them. I have only deleted objects that were created by users who had a very small number of edits (usually, only one edit!) and who have added a long description tag that went like "XYZ is the best restaurant in ABC town" or so, i.e. where the description was clearly marketing and not just "descriptive". Tre0xf, it was not possible to "take it up with the user" because all these users have done practically nothing but add one POI and never use their account again! Can you point me to a specific thing that you think should not have been removed?
-
Comment from Prince Kassad
This changeset also deleted objects that no longer had SEO advertising in their newest revision, i. e. https://www.openstreetmap.org/way/266904941/history. This is an extremely harmful changeset and should be undone immediately.
-
Comment from Владимир К
is this a joke? all objects i checked are NOT SEO or even advertising! These descriptions were just descriptions. I think it is totally wrong to delete things, you even don''t understand in another language! You even deleted the hole building, that exists on satimagery!
https://www.openstreetmap.org/way/216570290/history#map=12/56.0350/92.8873
it's vandalism, I'll revert this -
Comment from Владимир К
in Moscow only 2 of 8 deleted objects have description! and another exxisting building was deleted
-
Comment from woodpeck
Hello Владимир К, Prince Kassad, indeed I have removed objects even if they *had* advertising in the past, and other mappers had removed the advertising in the mean time. That was unnecessary. I'll undelete the affected objects. Please refrain from reverting the full changeset because that would make you the "last editor" of thousands of SEO messages in OSM and you don't want that.
-
Comment from woodpeck
Hello Владимир, against my warning you have reverted this and now re-created things like this: http://www.openstreetmap.org/way/384176165 - will you now take over and do this properly, removing advertising from where it doesn't belong, or do you expect others to do this?
-
Comment from muralito
I just wanted to raise the flag that there are some false positives in the algorithm, so i asked for an improvement, I also understand if the algorithm is not disclosed as it can be used by advertisers for "tunning" their ads.
In thousands of elements there are probably a lot of distinct cases to consider.
Maybe using a scoring system, so clearly ads could be deleted, and other cases can be tagged for manual review.For example, this case, https://www.openstreetmap.org/node/5046708259/history , if the element is valid, and the description is considered as advertisement, why not only delete this tag, and not the whole node?
Just keep hunting ads, but instead of a nuke use a sniper. :-)
-
Comment from woodpeck
Yes, it was not an algorithm that I was planning to run regularly, I just had identified long-description objects made by users who have contributed very little, and then went through the list manually, using a translating web site for anything other than English and German. Quite possible that I accidentally removed a few good ones. My idea was to not reward spammers by leaving their shop or office on the map at all. I'm now waiting to see if Владимир К has more to offer than a knee-jerk revert.
-
Comment from SimonvdP
Did you even check your example?, removing part of a likely valid secondary road seems worse than leaving it there with the useless description on it,
-
Comment from Chris Fleming
This feels excessive to me - generally when I see these added, provided the information is accurate then I don't mind the "spammy description"; but there is also a case for removing. My worry with this changeset is that by completely removing the object we remove information that had been added prior to the "spammy" additions. Is there a standard policy on what is considered spammy and what isn't - for example should we be removing all the hilton hotels that would get hit by this?
-
Comment from woodpeck
I think that any contribution made by a spammer should be entirely removed because it has not been made with OSM's interests at heart; most likely the coordinates are from a proprietary geocoder and the object will usually be placed with little regard to what's around it. So if the spammer adds a hotel I would remove it. However if the spammer only adds a description tag to an already existing object then, as you say, it is excessive to delete the full object. That was the problem with this changeset - it even deleted a couple of roads that had been the target of spammers.
-
Comment from Chris Fleming
Thanks - I think we are in Broad agreeement :)
-
Comment from GRUBERND
i think deleting things by algorithm is the most wrong thing one can to do to the OSM database. especially at this scale. it's the same as an import: automated changes of HUGE amounts of data.
things like these should always run through individual inspection on a case-by-case basis. cue map-roulette etc.
just sayin'.
-
Comment from muralito
@GRUBERND, I would agree if you write that sentence considering that using an algorithm does not necessarily imply automation.
-
Comment from muralito
@woodpeck (regarding your last comment)
Often a spammer (or any other user) deletes an existing object (street, POI, etc.) and adds a new one with her data. Please consider those cases when deleting or redacting objects. -
Comment from woodpeck
I have started carefully removing the spam objects here https://www.openstreetmap.org/changeset/53216381 : Where a spammy description and contact information were added to a building but there was no shop/office/amenity tag (just name/phone/payment methods etc), I reduced the object to a mere building and did NOT help the spammers by finding the right tags for the business.
Where a spammy description was added to a properly tagged business, I removed the spammy description only. Where a spammy description and address were added to a node or way but the address did not add any info because another nearby object already had that address, I d
eleted the whole thing. I also deleted the object where it was clear from aerial imagery that the node had not been properly placed (middle of road etc). -
Comment from SafwatHalaby
I think the algorithm is horrible.
-
Comment from SafwatHalaby
For instance, this has a slightly "positive tone" but seems to be a 100% valid node: https://www.openstreetmap.org/node/5109075827
.
I think your definition of a "spammy description" need to be narrowed. -
Comment from SafwatHalaby
Oh, I was reviewing via OSMCHA and I have a long backlog, and it seems that particular node was added back. Please ignore my comment if all the non spammy nodes were reverted.
473706412, v2189891274, v398979977, v6211396577, v4139015084, v8266074808, v3484085662, v2529863394, v2530206896, v3530236983, v3112743094, v7357468318, v4303464326, v4217896511, v4378534950, v3378534951, v3370733946, v4110140303, v11342036741, v486851111, v4
Relations (6)
95486230, v395486239, v395486240, v395486242, v34677699574, v24677699575, v24677699576, v24677699577, v22005004344, v22494205290, v32494205447, v22494205449, v22494205483, v2105508966, v3105508968, v3105508970, v496706863, v396706864, v396706867, v396706868, v3
Welcome to OpenStreetMap!
OpenStreetMap is a map of the world, created by people like you and free to use under an open license.
Hosting is supported by Fastly, OSMF corporate members, and other partners.
https://openstreetmap.org/copyright | https://openstreetmap.org |
Copyright OpenStreetMap and contributors, under an open license |