Thanks a lot for all the recent additions of P31. I was looking for the thank you option on the history pages until finding out it is a bot. Now 10 million items are rooted. Tamawashi (talk) 16:36, 11 July 2014 (UTC)
P31: album
Hello, I noticed that you added a parameters "album" for item P31. Due to filling infoboxes on Wikipedia would be better giving the type of album there: "studio album" / "live album" / "compilation album" / "extended play" (and possibly further). Do you think this could be changed? See there (in Czech) and there. --Marek Koudelka (talk) 13:46, 12 July 2014 (UTC)
I don't have a bot that can parse wikitext. I can "update" some of them if there are appropriate categories on a Wikipedia. --Magnus Manske (talk) 14:44, 12 July 2014 (UTC)
Hi, unfortunately I've had to block your bot because it was adding several incorrect statements over the last several hours: [1][2][3] When there are this many errors that someone has to go through and check, it is inappropriate to use a bot for this task. Please let me know if you are able to fix this, and I can unblock. --Rschen775418:23, 12 July 2014 (UTC)
These are in en.wp categories of "unincorporated areas", that's why the bot added that statement. If you have a clear way of defining the "bad" ones, I can revert them (e.g. item list, or they are all in a Wikipedia category). I won't be running that set of categories again, so you can unblock. Thanks for keeping an eye out! --Magnus Manske (talk) 19:11, 12 July 2014 (UTC)
Unblocked now, and thanks for responding so quickly. --Rschen7754
On Eisenhower Tunnel I don't see a "unincorporated area" category. But your tag of architectural structure was correct. I don't understand why Rschen7754 removed it, and didn't even place an alternative P31? Same for Northern Expressway. So out of three examples of alleged mistagging in the initial post, two were not mistagging. Tamawashi (talk) 20:28, 12 July 2014 (UTC)
While that is an "architectural structure" that is too generic to be useful. Something more specific should have been used. --Rschen775420:57, 12 July 2014 (UTC)
Can you cite any policy that enforces information removal if that information is less specific than some imaginary non-attached information? Furthermore, such statements as "is architectural structure" allow the removal of GND main type = geographic feature. If you cannot cite a policy that enforces such removal of correct information, please seek consensus via an RfC first before doing it again. Tamawashi (talk) 22:36, 12 July 2014 (UTC)
item creator: item creation is possible from Wikisource, but only when we provide the list of link. The "create from category tree" option is apparently restricted to Wikipedia. Is that fixable ? There appears to be many Wikisource pages that still lack an item.
The missing image heat map shows that some items have issues (like they are supposed to be located in Paris but have coordinates in England). Unfortunately, there is no way to find the items the cause the heat. Or do I miss something - is there a way I can see the results of a particular query on a map ? --Zolo (talk) 14:42, 12 July 2014 (UTC)
I added a "project" input to the creator; note that it still only does namespace 0 (article/main). As for the map, in principle yes, but I haven't figured out how to display large numbers of items as "pins" yet. You could patch a query to get some items yourself though; see the API doc, especially the AROUND command. --Magnus Manske (talk) 15:18, 12 July 2014 (UTC)
Thanks for the update.
I don't think that Wikidataquery alone can get a list of items with p131 Paris that are implausibly far from the Eiffel Tower, but I could get that using toolscript, it appears to be frightfully powerful... --Zolo (talk) 16:03, 12 July 2014 (UTC)
Oh I see thanks ! While I'm here, I noted that creating items from Wikisource with the item creator does not add a lable. No big deal as other bots can do it later on, just notifying. --Zolo (talk) 10:11, 13 July 2014 (UTC)
There was a bug (unrelated to WDQ) that looked for Wikipedia title matches in Wikidata, no matter what project you specified. Fixed now. --Magnus Manske (talk) 08:57, 16 July 2014 (UTC)
Hi Magnus, it seems that there is now a problem with autolist2 in associating Wikipedia pages from a category with the corresponding Wikidata items. Each time I try I have "Getting corresponding Wikidata items... 0 items found.". — Ayack (talk) 09:22, 16 July 2014 (UTC)
When trying to create items from a category (like 1950年代) with the item creator, I needed to change the hyphen into an underscore (zh_yue). That not totally intuitive but it works. However, I cannot create the item: when alias does not exist it shows: "Problem creating item". --Zolo (talk) 13:37, 16 July 2014 (UTC)
@Zolo: Wir haben uns falsch geeinigt, weil du inception (P571) neu/anders interpretierst? Mit "materiales Buch" meinst vermutlich gedrucktes Buch, das bekanntlich aber auch immateriell als E-Book gelesen und zitiert werden darf ... Ich denke, es hilft nicht weiter, wenn wir jetzt alles verwirren und behaupten, dass die Definitionen der Eigenschaften anders gemeint als sie formuliert sind. @Magnus: Danke für die schnelle Rückmeldung. Ich gebe die Frage an Wikidata talk:Books task force#Buch vs. literarisches Werk weiter. --Kolja21 (talk) 19:43, 18 July 2014 (UTC)
Tippfehler, entschuldigung. Ich meinte book (Q571), nicht P571. Ich interpriere es nicht neu. Q571 = en:Book = de:Buch bedeutet "eine mit einer Bindung und meistens auch mit Bucheinband (Umschlag) versehene Sammlung von bedruckten, beschriebenen, bemalten oder auch leeren Blättern aus Papier oder anderen geeigneten Materialien" (so kein e-book).
Hi Magnus, I'm having a bit of trouble loading lists on Autolist2. I'm using a WDQ "claim[225] AND claim[31] AND noclaim[31:16521]" with "either category or WDQ" selected and I'm getting this message:
Thanks. There is a delay between generating the statement and setting it on Wikidata, which could cause these doubles. The OAuth API code already filters out attempts to create duplicate strings and items "links"; I've now added dates a well. Shouldn't happen again. --Magnus Manske (talk) 11:23, 20 July 2014 (UTC)
Bot should be fixed for this now. Is there a bot that fixes the ~1200? Or at least removes them? I will add the corrected ones soon, but I can't remove specific entries at the moment. --Magnus Manske (talk) 12:11, 20 July 2014 (UTC)
Thanks, I have added the "corrected" ones where I could figure out the correct schema ("n" and "sh" escape me though...). --Magnus Manske (talk) 13:01, 20 July 2014 (UTC)
Hi Magnus, please take a look at this list. The same id "85041920" has been added to multiple items by your bot. I assume this is from mix'n'match. Something went wrong here. Can you look into it? Multichill (talk) 17:35, 22 July 2014 (UTC)
Hallo Magnus, mit TREE[14560][][171] bekomme ich heraus, welche Einträge zu den Kakteengewächsen existieren. TREE[14560][][171] and CLAIM[225]{CLAIM[405]} verrät mir, welche davon einen Qualifier taxon author (P405) für taxon name (P225) besitzen. So weit wunderbar. Nun interessieren mich aber genau die Einträge bei denen taxon author (P405) noch fehlen. TREE[14560][][171] and CLAIM[225]{NOCLAIM[405]} liefert leider kein Ergebnis. Mache ich was falsch? Gruß --Succu (talk) 14:08, 23 July 2014 (UTC)
Das geht leider (im Moment noch!) nicht direkt. Schneller Hack: in autolist2 die Abfrage mit qualifier laufen lassen, auf "Download" klicken; die gleiche Abfrage ohne qualifier (TREE[14560][][171] and CLAIM[225]) laufen lassen, nochmal "Download"; der diff der beiden Listen (oder sort|uniq -u oder so was) gibt Dir die gewünschte Liste. Kannst Du dann auch wieder in autolist2 pasten :-) --Magnus Manske (talk) 16:05, 23 July 2014 (UTC)
Danke. Es ist kein Problem für mich, mir das Ergebnis aus zwei Abfragen zusammenzubasteln. BTW: Hast du schon konkretere Pläne auch das Abfragen von bestimmten Nachweisen zu ermöglichen? Ich hätte da mindestens zwei Anwendungsfälle... Aber es geht auch anders, nur halt nicht so elegant. :) Gruß --Succu (talk) 18:23, 23 July 2014 (UTC)
Hi Magnus, We have been using mix'n'match to match MEP's MEP directory ID (P1186). I just finished the auto matched items. Out of the total of 3,593 pages, we have matched 3,320 (92.5%) items. That leaves us with 273 (7.6%) MEP's that haven't been matched. I've been clicking through them and occasionally I'm able to find a match, but most of the time we just don't have an item yet. Would you be willing to just create these missing items? That's probably easiest. We can just merge the few duplicates this would create. Multichill (talk) 15:22, 26 July 2014 (UTC)
It would be interesting if one could cross autolist with search. Sample: Special:Search/Lucy not in (claim[735:13365715] or claim[734:17379071] or claim[138]) .
I tried just pasting the search results into the manual list, but these can't be crossed with WDQ. --- Jura11:31, 25 July 2014 (UTC)
OK, I have changed the Autolist2 interface a bit to allow any combination of sources. Next, I'll add new sources, such as Wikidata and Wikipedia search. Will take a little while... --Magnus Manske (talk) 16:02, 25 July 2014 (UTC)
NOT is already there; it, actually, the default setting for WDQ. Your example. I tried a query for blank labels, but it does take too many resources. Maybe I'll find a way around that, but for no, no luck. --Magnus Manske (talk) 22:09, 28 July 2014 (UTC)
Your bot made a number of edits like this one. As a result, infoboxes in articles contain contradictory information, for example, a person was born in 2000 and discovered some stars in 1997 and 1998.
I have two questions regarding this: 1) Why is it neccesary to add such a fake date of birth if no article contains a real one? 2) Why the edit and its description contradict each other? January 1, 2000 does not belong to the 20th century, it was the first day of the 21st century. --Michgrig (talk) 08:10, 7 August 2014 (UTC)
The date added is 20th century, not January 1, 2000. Nothing contradictory. It's the way Wikidata stores centuries (see precision indicator). --- Jura08:19, 7 August 2014 (UTC)
So, do you want to say that it's not the problem of Wikidata but the problem of the information processing by infoboxes that do not account precision? --Michgrig (talk) 08:24, 7 August 2014 (UTC)
I don't think it's a problem for infoboxes, but they need take into account precision. Dates with a precision of days, months, years, decades and centuries can have the same ISO timestamp. For a sample of a year, see this. --- Jura08:32, 7 August 2014 (UTC)
It happens quite regularly now that users of the game merge items on the administrative units in Thailand, which just have the name in common, but are totally different - different administrative type (confusingly there are quite a lot pairs of two different administrative units sharing the same name and area, luckily only few have items so far) or even a totally different location. It is quite annoying to revert the changes, especially if incoming links were also changed. Is it possible to add a "white list" to the game, items within that group are known to be unique and should never be shown as candidates for merge? I could easily create such a list for the Thai administrative units as I store them in XML files myself - in fact I only noticed these wrong merges due to that. Apparently many gamers don't check the suggestion enough to recognize that they are different. 12:31, 20 July 2014 (UTC)
There is no "white list" mechanism per se, but if you have an item they all link to (e.g. "administrative unit in Thailand", if that exists), I can auto-"resolve" pairs where at least one of the pair links to that item. --Magnus Manske (talk) 12:33, 20 July 2014 (UTC)
As I wrote a few lines above, there is no mechanism to whitelist a large number of items. However, apparently there is a proposed "homonym" property; the Game will not suggest merging items that link to each other, so once these taxa link to each other as "homonym", this issue will solve itself. --Magnus Manske (talk) 09:27, 23 July 2014 (UTC)
I don't know if I wrong something but Permalink and Download don't "remember" the mode (AND/OR/NOT), the string have always &mode=undefined. It's a my problem or a bug? --ValterVB (talk) 13:50, 13 August 2014 (UTC)
Some of the newer features don't work correctly through the permalink. You need to add them manually to the url.
My problem is that I use Autolist for cross search in Wikipedia category and Wikidata statements but I need the results with Q number to use with my bot or with Quick Statements. Download can be useful but don't work correctly, maybe can be useful a flag to show only Q number in results. --ValterVB (talk) 19:13, 13 August 2014 (UTC)
tool featured on new Main page
Hello,
As you may already know, I've recently been working on an update to the Main page; I am hoping to launch this page in the next two weeks. I wanted to give you a heads up that your Wikidata tempo-spatial display tool is slated to be the first featured content in the new "Discover" section (note: this will only be temporary as this section's content is to be updated regularly as per discussions found here). Can you let me know if you have any objections or revisions to having your tool included? I'm really excited about including it as an example of some of the interesting applications and work coming out of the community! Cheers. -Thepwnco (talk) 18:24, 13 August 2014 (UTC)
I am using the Wikidata useful tool to update tons of Hebrew Wikidata items without male / female indication. Great tool.
However, I encounter 2 issues which are taking out the fun...
1. Working with Chrome 10, the links do not show on the right or left side as in the picture but rather in the bottom. So instead of having it handy, you need to scrol all the way down to get it, which adds up when you update bunch of them
2. Each time I select and update, a new window pops up. So after updating many of them, the browser is crowded with many useless open screens and it is slow due to memory usage. I suggest to open a new screen on the first use, and than each additional use will just reopen the same open screen and populate it with new data.
Hi, I'm afraid I can't check this, as the current Chrome version is 37, and I wouldn't know where to get 10, let alone if it runs on my machine. Chrome 10 was released over three years ago. I suggest you upgrade to a current version, or try Firefox, and tell me is the problems persist. --Magnus Manske (talk) 17:54, 16 August 2014 (UTC)
Thanks for looking in on User:Micru's talk page a couple of days ago.
The question was how to check how many commons links that ought to go to Category pages (because they belong to items that satisfy CLAIM[31:4167836] AND LINK[commonswiki]) actually in reality are not pointing to Commons Category pages, but are pointing to Commons galleries.
And also, to check how many commons links that ought not to go to Category pages (because they belong to items that satisfy LINK[commonswiki] AND NOCLAIM[31:4167836]) actually in reality are not pointing to Commons gallery pages, but are pointing to Commons galleries. (Even though the items are supposed to be article-like, not category-like).
I noticed some edits went "missing" from the perpetual update cycle. That shouldn't happen, but... I'll have to look at the code, but currently juggling too many things at once. The next Wikidata dump should be in a week or so, when I'll re-seed the database. Meanwhile, "reported" lag is here. --Magnus Manske (talk) 17:56, 18 August 2014 (UTC)
Ok, so it wasn't me. Wikidata was rather unstable around that time. I had quite a few api errors, that might explain it.
I know the feeling. I'll just wait for the next dump and do a rerun to find the missing items. Any suggestion on how to make a list that have the same collection (collection (P195)) and the same inventory id (inventory number (P217))? I'll probably need that to hunt down the duplicates I created. Multichill (talk) 18:54, 18 August 2014 (UTC)
Until I have time to introduce variables a la SQL to WDQ (so, next millennium ;-) you'll have to do a script. Add "&props=195,217" for a convenient JSON list. One can probably do a jq one-liner. I so need to check that one out in detail! --Magnus Manske (talk) 19:21, 18 August 2014 (UTC)
Hi Magnus, just in case you were thinking what awesome tool to do next, I will explain the difficulties I found while working on this tree. I am trying to enter values for a minimal set of properties (tonality, date of creation, date of publication, IMSL, musicbrainz), but I don't know which ones are already in some items, and which ones are missing from others. I can do queries but it is cumbersome to work with.
What it would help is to be able to perform a query and to show the results in a table with some chosen properties as columns (one value per property column is enough), and then be able to edit this spreadsheet cell by cell. I have found some components, I don't know if they could be used for this purpose (SlickGrid, EditableGrid, EditableTable, etc).--Micru (talk) 08:10, 23 August 2014 (UTC)
Thanks, that works really well, it is exactly what I had in mind. If some day you manage to make the cells editable, that will be my editing interface of choice for this kind of tasks.--Micru (talk) 23:00, 23 August 2014 (UTC)
Hi Magnus, I see you started importing before you had some people clean up. The number of constraint violations is now really massive. Are you going to work on this after the initial import? It wouldn't be good to leave it in the state it's currently in. Multichill (talk) 10:49, 31 August 2014 (UTC)
Hi Multichill. First, I am but one guy; I'm hard-pressed to auto-filter and create items, and I now need to create the Welsh ones as well. And the NI ones. By tomorrow. Second, most of the "violations" I've seen are bogus. Having a church with a National Heritage List for England number (P1216) is a violation? Ridiculous. Also, some other bot ingeniously added National Heritage List for England number (P1216) to list items, which is where most of the duplicate "violations" come from. Third, yes, in time I hope I will be able to fix some of the actual violations. My own database contains over 160K entries right now, most of which should/will be Wikidata items. Expecting me to keep these in perfect state on Wikidata would be hilarious if that were my full-time job; as it is, I'm not even discussing this. If there is some big problem with what I do, please talk to WMUK, and if they tell me to stop, I will do so immediately. Otherwise, more items. And more bogus "violations". Today. --Magnus Manske (talk) 11:47, 31 August 2014 (UTC)
Hi Magnus, like my bot did by mistake, yours is also "ingeniously" (sorry ;) ) adding English Heritage ID to lists, even when they are marked as such. See for example [13]. — Ayack (talk) 17:29, 1 September 2014 (UTC)
Wikidata item creator
Hi, it was nice talking to you at Wikimania – like I said there, I'm in love with your tools. :-)
I have a request about this tool. When I create articles from nowiki, it automatically adds labels in language "no", but the language should be "nb". Would you be able to make it so the labels are added in "nb" instead? "nb" is the content language of nowiki. If you would like a more generalised solution, the $wgContentLanguage of nowiki is "nb", so maybe the script could fetch that, since this may be an issue for other languages as well (als, zh-yue and be-x-old come to mind). Jon Harald Søby (talk) 00:48, 1 September 2014 (UTC)
this edit is an error. The item is indeed a book but is a dead-serious rabbinical commentary on an earlier rabbinical book, nothing close to a book written for " aesthetic or recreative purposes". DGtal (talk) 06:58, 1 September 2014 (UTC)
You may wish to look up the definition of "error". literary work (Q7725634) is the highest class of written word on Wikidata, AFAIK. The next higher class is creative work (Q17537576), which encompasses other forms of art. Therefore, if it's written down, it's literary work (Q7725634), which means the edit was correct. There may be a more precise subclass like "angry rabbinical commentary" somewhere. Feel free to change the instance of the item to that specific subclass; in fact, it would have been quicker to do that rather than complaining on my talk page. Of course, my bot could have read the article and found the correct subclass on its own; that is, right after it achieves sentience. I'll get back to you on that one later. --Magnus Manske (talk) 07:53, 1 September 2014 (UTC)
Dear Magnus, User:Jon Harald Søby seems to have got the point: literary work is described in English as "creative work by a writer created with aesthetic or recreative purposes" which to the best of my understanding does not fit non-fiction written works. Therefore I commented to you in hope that you will be able to further perfect your work, not to attack you personally in any way. It is quite possible that the answer lies in rewording. DGtal (talk) 11:27, 1 September 2014 (UTC)
WiDaR and blank values
Hi, 'set_label' and 'set_description' currently requires non-blank values, but it would be nice if you could accept blank values as well. At least for deprecated languages, there is a need for removing labels and descriptions. Danmichaelo (talk) 20:12, 7 September 2014 (UTC)
Hi Magnus, is there a way to remove coordinates (coordinate location (P625)) with autolist2? When I put -P625 in the "Add/remove statements" box, I've got an error message saying "Put something in the box above!". Thanks. — Ayack (talk) 19:05, 10 September 2014 (UTC)
No, it only works with a specific item as "target". Giving it the ability to mass-delete statements indiscriminately seemed ... risky. --Magnus Manske (talk) 20:40, 10 September 2014 (UTC)
Wikidate - The Game - Thanks and some suggestions :)
This is a splendid tool to find Persons in all items without P31... :)
One question : is there a place, a list, where items tagged as "not a person" can be explored to find other P31 values ?
Knowing that a whole list of items is NOT human (Q5) is already a great start - filtering it with language links would be even better ;)
No-item is also really a great tool, and your recent input, that allows to add directly to an item is MAJOR improvement... - could you please display instance of (P31) next to every item name (when existing) ? :)
I also use image game to add images, using human (Q5) filtering... it is very efficient this way, much more than without filtering :)
Do you think it would be possible to note items like Renzo Revoredo (Q919683), where the image is not situated on Commons, so that it would be possible to search later if the image can be imported to commons ? or remove such items from the Game ?
Piggybacking on this, I have a question about the configuration for various languages. For example, for the date game, dates in Espreanto aren't marked properly (the days are missing). I'd love to be help with this and other issues (such as gender pronouns for Norwegian), but am not sure where/how. Is the code in a git repo or something? If not, maybe an idea could be to have such configuration in e.g. User:Magnus Manske/wikidata-game.js so that it can be edited on-wiki? Jon Harald Søby (talk) 10:02, 13 September 2014 (UTC)
enwiki:"Vid Cencic" Item(s) exists with this label/alias: Q9368671
enwiki:"Santiago Cepeda" Item(s) exists with this label/alias: Q6120860
enwiki:"Charo Soriano" Item(s) exists with this label/alias: Q5765150
enwiki:"Brent Wright" Item(s) exists with this label/alias: Q3644275
enwiki:"Cheng Changjie" Item(s) exists with this label/alias: Q11963508
enwiki:"Chi-chi Nwanoku" Item(s) exists with this label/alias: Q5095025
enwiki:"Chiang Chen" Item(s) exists with this label/alias: Q9182115
enwiki:"Luis M. Chiappe" Item(s) exists with this label/alias: Q15990383
enwiki:"Chris Manchester" Item(s) exists with this label/alias: Q11690196
enwiki:"Chucky Jeffery" Item(s) exists with this label/alias: Q16025060
When creating items, at some point I end up with lists like the above. Generally, the article should be added to the existing item so creating additional items for these wouldn't be worth it.
When there are just a few items, it fairly easy to open article and item, check both and add the article.
For longer lists, it might be worth feeding them into an interface like the one of the Wikidata merge "game". --- Jura04:03, 5 September 2014 (UTC)
Hi Jura, automatic addition of the article to the item would be a bad idea, as there is no guarantee that the article and the item actually describe the same topic. As for an "adding game", I actually have already most of the backend code up and running; however, the interface may become so complicated that it's not worth it. That said, "un-itemized" articles are becoming an issue IMO, so we should do something. --Magnus Manske (talk) 07:52, 5 September 2014 (UTC)
Yeah, automatic addition would be a bad idea. As I usually add "P31:Q5" given the categories I use to select articles, it might be worth creating the items, even if there are already items with the same label. Eventually the "merge game" would pick them up.
Once in a while I come across a category in a Wikipedia with a fairly large number of old "un-itemized" articles. I'm wondering what is the extent of these. We could try to work these more systematically. --- Jura09:38, 6 September 2014 (UTC)
may I suggest you add the P31 value (in game prefered language / English / language of the selected wp ) to help identify existing items ? --Hsarrazin (talk) 22:59, 6 September 2014 (UTC)
It might a heavy dose. I tend to work from categories in a given language. This avoids me tons of pages on "Eurovision prelimary for the 15th participating country in 200x". --- Jura16:29, 7 September 2014 (UTC)
@Jura1, considering your original question, it is possible to add links using WiDaR the following way: Create new items with Item Creator, and then merge using Quick Statements. It would be even more convenient if one could add language links by Quick Statements, but this is okay. Kind regards, Lymantria (talk) 09:58, 21 September 2014 (UTC)
Autolist error when trying to get items in a frwiki category
Hello Magnus when I get this message "Getting Wikipedia category tree... There was an error running the query "Table 'frwiki_p.page' doesn't exist" when I try to get items in a fr.wikipedia category. I have no issue for en or de. --Zolo (talk) 09:00, 21 September 2014 (UTC)
There are quite many pages on Wikiquotes which has not been associated with Wikidata items, for example in Russian Wikiquote: >4000 pages. The task is quite simple but tedious. Many pages should be connected to pages in Wikipedia which titled similarly, but sometimes parentheses should be omitted or added. I suppose it can be successfully gamified. May be it worked for all other languages too. Thanks in advance. Infovarius (talk) 09:41, 21 September 2014 (UTC)
added after edit conflict