[Rebuild] Redaction bot is almost done

Harry Wood mail at harrywood.co.uk
Thu Jun 14 11:35:00 BST 2012


Correction. Turns out I'm wrong about the "Supercalifragilisticexpialidocious Stret" example. That's a spelling correction within a word, which *does* get correctly identified as trivial change. See trivial change tests related to spelling here: https://github.com/zerebubuth/openstreetmap-license-change/blob/master/test_tags.rb#L32

----- Original Message -----
From: Harry Wood <mail at harrywood.co.uk>
To: Ed Loach <ed at loach.me.uk>; "rebuild at openstreetmap.org" <rebuild at openstreetmap.org>
Cc: 
Sent: Thursday, 14 June 2012, 10:57
Subject: Re: [Rebuild] Redaction bot is almost done

Yes. Or to spell it out some more:

We have a happy outcome:
* Agreer adds a street name tag "Воздвиженка улица"* Disagreer swaps it around to "улица Воздвиженка"
* Fixing bot puts it back again "Воздвиженка улица"
* redaction bot doesn't need to do anything. Win!

But we still have an outcome which may be bad
* Disagreer adds a street name tag "улица Воздвиженка"  (Should be treated as copyrightable work)* Agreer (or fixing bot) swaps it around to "Воздвиженка улица" (A trivial change)
* The redaction bot should perhaps recognise the last edit as trivial and redact the data, but currently does not.

Is this a big problem worth worrying about? Possibly. We don't currently have a test for this, but when you arrange the logic this way around there's plenty of other cases we might worry about. For example:

* Disagreer adds a street name tag "Supercalifragilisticexpialidocious Stret" (Copyrightable work)
* Agreer changes it to "Supercalifragilisticexpialidocious Street" (trivial spelling fix)* The redaction bot should perhaps recognise the last edit as trivial and redact the data, but currently does not

How clever do we expect the bot to be about this kind of stuff?


Harry

________________________________
From: Ed Loach <ed at loach.me.uk>
To: 'Ilya Zverev' <zverik at textual.ru>; rebuild at openstreetmap.org 
Sent: Thursday, 14 June 2012, 10:53
Subject: Re: [Rebuild] Redaction bot is almost done

Ilya wrote:

> We have a bot that monitors the whole country and fixes word
> order and
> abbreviations in street names (and in addr:street tags). So any
such
> edits will be reverted to the "correct" state. I doubt the
redation
> bot
> should have some special processing for Russian street names,
> since the
> street bot will have its way regardless :)

As Gnothgol pointed out in response to my previous query here, the
main reason the redaction bot needs to have rules for Russian street
names is for where a non-agreer added a street name, and only
trivial changes have been made since, so the name needs completely
removing. In the case of Russia "trivial" includes changing the
order of the words (as well as expanding abbreviations), such as the
street bot does, but it could also have been done manually in some
cases. So the rules are required to ensure street names aren't kept
where they shouldn't be.

Ed

_______________________________________________
Rebuild mailing list
Rebuild at openstreetmap.org
http://lists.openstreetmap.org/listinfo/rebuild




More information about the Rebuild mailing list