Bot edit proposal: fixing wikipedia tags pointing at redirects in USA (where it can be done reliably)

Communities United States
1

How would you feel about automatically updating wikipedia tag redirect in USA in cases where for given element:

  • wikipedia tag exists
  • wikidata tag exists
  • Wikipedia article redirects to a new article
  • wikipedia and wikidata are mismatching
  • Redirect target matches existing wikidata on an element
    ?

Note that the last requirements handles cases where wikidata element points to Wikipedia redirect, not the target page.

It is useful as sometimes useful redirect are deleted at Wikipedia and then you need to clean that manually and it gets more complex[1]. And in general direct links are preferred over redirects.

See Mechanical Edits/Mateusz Konieczny - bot account/fixing wikipedia tags pointing at redirects in USA - OpenStreetMap Wiki for wiki docs.

It is useful to do this automatically as it is not worth human time to do the same manually. Sample of example edits is listed at https://matkoniecz.github.io/OSM-wikipedia-tag-validator-reports/California%20-%20obvious.html#wikipedia%20wikidata%20mismatch%20-%20follow%20wikipedia%20redirect (bot checks before edit, so substantially modified entries would be skipped)

I am running such bot edit for Poland - see Mechanical Edits/Mateusz Konieczny - bot account/fixing wikipedia tags pointing at redirects in Poland - OpenStreetMap Wiki so it would be just matter of enabling it for USA

See Changeset: 138714235 | OpenStreetMap Changeset: 138714812 | OpenStreetMap for an example edits (reviewed manually)

See Changeset: 138008736 | OpenStreetMap for an example edit in a bot mode (note different changeset tags).

Bot edit would be recurring and new entries also would be fixed.

(if you feel that such edit would be unwelcome in USA also let me know - in such case I can list such edit as requiring human review or stop reporting it at all, even as something potentially bottable)

See https://matkoniecz.github.io/OSM-wikipedia-tag-validator-reports/USA%20-%20obvious.html#wikipedia%20wikidata%20mismatch%20-%20follow%20wikipedia%20redirect for general USA match page but it may load slowly due to its size.

note: not all states are being processed right now and new may be added in future

I am experienced with bot edits, including single case where I needed to revert some - and then I did it on my own.

https://matkoniecz.github.io/OSM-wikipedia-tag-validator-reports/ lists many reports where human help and review is welcome

MapRoulette was recently created for some subset and may export more such cases if people would be interested in using such lists

[1] see this for just nodes in some selected areas where wikipedia points to nonexisting article - some of them point to now deleted redirects. BTW, people fixed all cases in USA: let me know if anyone would want more reports from that area and would be interested in fixing what would be found

6 Likes
2

Perhaps it makes sense to group edits by wikipedia particle?
For example, rather than each road way gets its own separate changeset, all ways associated with the same wikipedia article, getting the same edit, get grouped into one changeset.

1 Like
3

Maybe, though it would require reworking how edits are made… Would it be a significant improvement?

4

So I realize this was previously discussed, but this edit just made data worse in one specific example:

In this case, the wikidata item was wrong - it was the item associated with a wikipedia article about a list of road routes rather than one specific route.

The bot edit dutifully changed the wikipedia tag to the one that corresponds to the wikdata tag on that object.

Before, we had a wikidata tag and wikipedia tag that didn’t agree - a great case for a human to look at it.

After, we had a wikidata and wikipedia tag that agreed, but both were wrong. In my opinion, this makes the data worse.

I fixed this by deleting both the wikidata and wikipedia tags since there didn’t seem to be an entry in either for this specific route.

I’m opposed to further automated edits to the wikipedia tag in the United States. In my opinion, if anything they should be removed. If they disagree with wikidata tags, a human needs to look at it.

1 Like
5

Part of the problem is that the original Wikidata matching bot wanted some item to link to, so it assumed that a redirect is always semantically equivalent to its target. As we can see in this example, that isn’t necessarily the case.

I flagged this issue in a similar bot edit proposal thread not long ago:

1 Like
6

For now I stopped this bot edit. Will read this comments and reply when able.