Scidudebot

Maxlag and botflag

Hey User:SCIdude, in violation of the Wikidata:Bots policy, your bot seems to be ignoring the maxlag parameter as it edits while maxlag>5. It also seems that its edits are not marked as bot edits; can you please investigate? Thanks, MisterSynergy (talk) 20:57, 20 May 2020 (UTC)

I'm using the wikibase-cli library. The documentation says (https://github.com/maxlath/wikibase-edit/blob/master/docs/how_to.md#maxlag):

If the Wikibase server returns a maxlag error, the request will automatically be re-executed after
the amount of seconds recommended by the Wikibase server via the Retry-After header. This automatic
retry can be disabled by setting autoRetry to false in the general config or the request config.

If you want me to put this to the developer it would be helpful to have logs that show this actually happen. At the moment I have just your word on this.

As to the bot flag I also found docs and I will set the flag right now. --SCIdude (talk) 06:56, 21 May 2020 (UTC)

Retrying is okay, but the retry should not be successful unless there is again maxlag<5. Does this actually happen, or is the retry successful regardless of the maxlag value?
I do not have official logs, but a script which monitors maxlag and edit rates per account real time and I see that your bot account is one of very few which remain active while maxlag>5. The others are using QuickStatments, which currently has problems with proper throttling as well. —MisterSynergy (talk) 08:25, 21 May 2020 (UTC)

Two questions: is the bot flag visibly set now, with current edits? And where do I see the same that you see with maxlag, is there a tool? --SCIdude (talk) 08:28, 21 May 2020 (UTC)

Also is the server always returning a maxlag error, or only with bot edits? If the latter that might be an explanation, and it might be fixed now the bot flag is set. --SCIdude (talk) 08:33, 21 May 2020 (UTC)

The bot flag is now attached to its edits, thanks.
I am using a self-written Python script to monitor edit rates, as there are pretty often users who violate the bot policy. The script is not yet published, as is crashes sometimes and I need to polish it a bit more. Based on the output, your bot currently goes at ~22/min while maxlag~9. Usually a complete stop is expected for automated processes, but I would not complain if it just stayed below 10/min (i.e. manual editing speed).
Usually, a bot sends an edit to the server, together with a maxlag parameter (which contains the value which you configure in your tool); if the actual maxlag value found on the server is larger than the parameter your bot has provided, the edit will not be done and the server tells you when you can retry; if you just omit the maxlag parameter when making an edit, or report "0", the edit will be done no matter what the actual maxlag value is. —MisterSynergy (talk) 08:40, 21 May 2020 (UTC)

So it appears I can't set maxlag in the wikibase-cli configuration, although it seems supported in the underlying wikibase-edit library. I have opened https://github.com/maxlath/wikibase-cli/issues/109. --SCIdude (talk) 08:58, 21 May 2020 (UTC)

The quickest fix here is to insert a 'sleep 6s' if maxlag>5 which I'll monitor myself now. --SCIdude (talk) 09:07, 21 May 2020 (UTC)

A very dirty fix :-) Make sure that you do not start multiple instances of the script at the same time, tough. Other than that, it should be okay for now. —MisterSynergy (talk) 09:25, 21 May 2020 (UTC)

The bot seems to misbehave and remove data

@SCIdude: Your bot removed GO:0005623 from cell (Q7868) and added entity obsoleted in Gene Ontology (Q93740491). For entity obsoleted in Gene Ontology (Q93740491) to make any sense it would have to be on items that actually have an GO id. Instead of removing the statement the GO:0005623 statement should be deprecated with reason for deprecated rank (P2241) deprecated identifier value (Q67125514). ChristianKl ❪✉❫ 17:24, 4 December 2020 (UTC)

I might do that but I don't agree that it removes real data. There is no use for that information. And there is real danger that it is used despite the deprecated flag. Weren't there such cases already? --SCIdude (talk) 18:33, 4 December 2020 (UTC)

• There were cases where we discussed how to do with deprecated data in source databases and our community consenus is quite clear that we deprecate it on our end as well. The bot is not approved to remove IDs instead of deprecation.

Knowing what deprecated IDs mean is useful. Just because the IDs are deprecated doesn't mean that there aren't still external databases that use them. ChristianKl ❪✉❫ 23:48, 4 December 2020 (UTC)

Show me one case where this is useful. Anyway, I'll comply. I will not, however, reimport any deprecated information. That's gone. --SCIdude (talk) 05:04, 5 December 2020 (UTC)

In the particular case of cell it wasted some unnecceary time on my end trying to figure out why the concept of cell might be depricated even through it's a central concept in biology. Reading that it was deprecated because GO doesn't want to replicate CO was listed as the answer but that wasn't directly accessible and not clear from the reference to the instance of (P31) statement either.

Deprecation notes are generally valuable in cases where our structure somehow copies the structure of Gene Ontology. ChristianKl ❪✉❫ 11:59, 5 December 2020 (UTC)

Bot is creating subclass loops

I have removed a few subclass relations your bot had added that were obviously wrong; however, I'm confused by this one - you have Sucrose alpha-glucosidase (Q10891385) and beta-fructofuranosidase (Q419189) subclasses of one another. Please fix: subclass relations should be in only one direction, from the more specific to the more general term. ArthurPSmith (talk) 14:37, 3 June 2021 (UTC)

@ArthurPSmith: Thanks for your note. The subclass statements are inferred from molecular function statements, and in this case the confusing statement comes from InterPro, see http://www.ebi.ac.uk/interpro/entry/IPR021792. I will issue a bug report. What were your other corrections? --SCIdude (talk) 15:54, 3 June 2021 (UTC)

See the history of Prephenate dehydrogenase superfamily (Q82671060) and Glycogen debranching enzyme (Q24777945). ArthurPSmith (talk) 16:02, 3 June 2021 (UTC)

Merging error in publications

This edit introduces a merging error. — Finn Årup Nielsen (fnielsen) (talk) 16:59, 9 May 2022 (UTC)

I think there might be citation errors now. I am looking at https://www.wikidata.org/wiki/Special:WhatLinksHere/Q29395041 — Finn Årup Nielsen (fnielsen) (talk) 17:01, 9 May 2022 (UTC)

Old (bot) edit

I've noticed today your bot edit from December 2021, I don't know if you are aware of it now or if I reported it to you before: [1] (CID added by your bot was already present in the item, however it was deprecated). Regards, Wostr (talk) 18:51, 15 May 2022 (UTC)

Thanks, I'll look into it when I'm doing chem bots again. SCIdude (talk) 06:02, 16 May 2022 (UTC)

https://editgroups.toolforge.org/b/wikibase-cli/dd07779a6626/

Hi SciDude,

I'm interested in biochemistry. What can be done with the new relations?-- Schnorrantenthum (talk) 09:57, 11 September 2022 (UTC)

Go from small molecule ligand to protein. You need to appreciate how GO is used for annotating proteins, e.g. in UniProt. SCIdude (talk) 10:08, 11 September 2022 (UTC)