User talk:Magnus Manske

Previous discussion was archived at User talk:Magnus Manske/Archive 9 on 2015-08-10.

Hi Magnus,

Could you elaborate why you thing clinical trials data is not applicable to WD? (https://tools.wmflabs.org/mix-n-match/#/list/1984/na)

We are working on a data model (Wikidata:WikiProject Medicine/Data models/Trials) to be able to import this very important publicly available information. (@Bluerasberry: you might find this relevant)

That appears to be a table header row that you accidentally imported. Please, to what Wikidata item would you match that?

The other 318 entries you imported look OK.

Thanks Magnus.

Could you please change the URL from https://clinicaltrials.gov/$ID to https://clinicaltrials.gov/show/$ID ?

Done.

I have added all the items in this catalogue to WD (using quickstatements) but they are still not matched in Mix'nmatch! I tried refreshing but no lock! Any idea what is the issue and how I can solve it?

Solved!

"Manually Sync" under action synced them all!

Things seem okay here so I will take no action. I am bringing someone with clinical trials structured data to WikiCite and hope to get some more of these in.

I was trying to delete this catalog but couldn't find how to do it. The "edit catalog" did not work as I expected. Could you please let me know how I can delete it or kindly delete it yourself. Thanks.

We're playing around with IIIF manifests (IIIF manifest (P6108)). Could you update reasonator to use the third-party formatter URL for the IIIF property so that https://tools.wmflabs.org/reasonator/?q=Q254923&lang=en (Alba Madonna (Q254923)) points to http://media.getty.edu/viewers/mirador/?manifest=https://www.nga.gov/api/v1/iiif/presentation/manifest.json?cultObj:id=26 in reasonator?

Like so?

Exactly! Thanks for changing that. Is the url hardcoded or using the third-party formatter URL? I might want to change it in the future (phab:T209697).

It uses third-party formatter URL (P3303) for all string-based values (including URL, in the case of IIIF manifest (P6108)).

Would it mess up Reasonator if I also added a third-party formatter URL (P3303) for a different viewer, namely http://universalviewer.io/uv.html?manifest=$1 ?

In this discussion on the IIIF list, there was a desire that Wikidata should express that multiple viewers are available, rather than anointing any particular single one.

There's also been a suggestion to create a kind of jump-station, offering links to different viewers, similar to the way GeoHack offers a portal to different map views, but it remains to be seen whether anything will come of this.

Good afternoon! I have an excel file containing all archive-actors (creators) of the National archive of Norway with liks to their Archives. Can you have a look on that file and see if it is possible to make a Mix'n'Match for this?

PS itss an Excel file and i can send it to you If you want to look at it.

Hi, you can use https://tools.wmflabs.org/mix-n-match/import.php to import a tab-separated text/file (like you get from copy&pasting Excel) as a new catalog.

Thank you, but where should the Excel file be stored then? Can i just take the Excel file or should it be changed to a CSV file and paste it into the fiel below the row;

Text needs to be UTF-8. You can either paste the tab-delmited text into the box (for shorter catalogs), or upload it as as tab-delimited file (max. 20MB).

what does UTF-8 means

Maybe outside Mixn'Match but I have this error when using the Mix'n'Match importer. Do you have any suggestion?

Hi Magnus - will you be attending the Wikicite meeting in San Francisco the week after next? The final day is a "hack day" and I had some thoughts about potential improvements to SourceMD. Whether or not you're there, what would be a helpful process to implement improvements? In particular I was looking at the "new_resolve_authors" piece, which I'm not really sure of the status of, but it looks like it is close to doing something significant for author disambiguation that I've been wanting to work on for ages...

A Gutenberg copy that is a translation of a work is not the same as the original work, and needs a separate data item. So edits like [https://www.wikidata.org/w/index.php?title=Q663886&curid=624947&diff=787973013&oldid=787292908 this one] are incorrect.

That was synced from Mix'n'match. My bot can't know that it's "wrong".

Yes, it can. The language of the translation is different from the original; the translation has a translator listed, and it's a Gutenberg edition, which would never be added to a "Work" in the first place. If none of those safeguards or checks are in place, then you need to improve the coding of your bot.

You misunderstand. The bot knows nothing about the items. It just synchronizes Mix'n'match catalogs and their metadata.

So, your bot adds whatever nonsense is in Mix'n'Match no matter how daft it is, without performing even basic checks? I've already corrected that particular problem twice.

If you have fully functional AI code to do your "basic checks", I'll be happy to accept a pull request.

So, even though we KNOW this edit is wrong, your bot will continue to make it over and over because you do not have control over your own bot?

If an ID was once in Wikidata and then was removed, the ID should not be re-added without manual verification. Since you already have a database of matches, it shouldn't be difficult to keep track of whether the ID was already in Wikidata or not.

Regarding your response on Topic:Uolbihtl8b8ut2un, please see Wikidata:Bots. The bot operator is responsible for its contributions. It doesn't matter what happened in Mix'n'match, it's your bot doing the syncing, so it's your responsibility to stop it from overriding decisions made on Wikidata.

EncycloPetey is not the only person frustrated by the behaviour of your bot. If you're going to keep running the bot knowing that it's causing problems, then you're going to leave us with no choice but to block it.

Currently, Mix'n'match contains ~50M entries in 1615 catalogs.

Expecting me to manually validate every match made by another user is ridiculous.

Expecting me to write an AI chat can validate matches is ridiculous.

As for the "re-addition", I already have code to prevent that. Turns out that the switch to Multi-Content Revisions changes the table structure of Wikidata, and the check would silently fail. This has now been corrected.

Nikki, please check and review your overreacting threats.

In [https://www.wikidata.org/w/index.php?title=Q7723256&type=revision&diff=790937305&oldid=712882656 this edit] a lithographic print was added to a data item about a stage play.

The bot is only importing what another human determined. I suggest you take your issues there.

Hi! I imagine this bot is doing other users' edits from mix n' match, but it's now twice added an NNDB people ID to a fictional character https://www.wikidata.org/w/index.php?title=Q18710699&action=history . Is there a way to make the NNDB catalog only match with instanceOf:human on mix n' match? In the meantime I just created an item for the human in question (Tom Buchanan (Q58650775)) to avoid this happening again. I can't tell if this was an automatic match or a manual match done by a user; if it was automatic then that should be changed.

Cheers,

Hardwigg

It was "automatched via auxiliary data" , but I'm not sure where it came from. I have changed the entry in Mix'n'match to point to your new item. I will try to add a "Q5" filter to the sync code.

The only shared information between the NNDB page and this character are their name and their gender, so not sure how it could have automatched them together. But yeah, adding that filter should do the trick.

This id is for vascular cryptogams, not for all cryptogams. I'm not sure that Wikidata has a data item for this, because the concept is no longer used in classification. I couldn't find anything by searching for likely English terms.

I won't bother to look up what you mean. Presumably an edit. Presumably by a bot of mine, or through a tool of mine. If it's wrong, just fix it.

[https://www.wikidata.org/w/index.php?title=Q333458&curid=319599&diff=790433883&oldid=787722476 Same edit] made again. It's still a mismatch and shouldn't be added here. I've corrected it twice now. Can you please prevent your bot from doing it again?

Hi Magnus, I have noticed that some items are not showing up as expected in catalog lists made from Listeria when they have more than one catalog number and the missing number is not listed as the first catalog number in the item's P528 statement (at least for the ones missing that I checked). See e.g. Q31158443 for Wikidata:WikiProject sum of all paintings/Catalog/Adam Pynacker catalog raisonné, 1926. I admit I use a pretty weird query to enable sorting (catalog numbers are not always whole integers) but it used to work as expected. Thx in advance.

Looks like one of the WDQS servers is out of date. If you keep re-running the query (add "?x1", "?x2" etc in the results to bypass caching), you will see that sometimes, you get 22 results, sometimes 24. Nothing I can do about it. Fell free to file a Phabricator ticket.