Page MenuHomePhabricator

Property Alvin P6821 created and need to be matched
Open, Needs TriagePublic

Description

About Alvin


Objects list of 1000 connected objects https://w.wiki/4r9


In the picture above we can see the problem with open data and occupations. I did some cleaning who is a "fornforskare " and now we have the picture below


The theory is that WIkidata/WIkipedia should be based on authorities BUT

Swedish related properties today in WD

https://www.wikidata.org/wiki/Template:Sweden_properties

Queries

and not

Event Timeline

There are a very large number of changes, so older changes are hidden. Show Older Changes
Salgo60 renamed this task from Property Alvin P6821 created and need to be matches to Property Alvin P6821 created and need to be matched.Jun 11 2019, 3:11 PM
Salgo60 updated the task description. (Show Details)
Salgo60 updated the task description. (Show Details)Jun 11 2019, 3:16 PM

Skickat

SVAR: Alvin-supporten tar semester och är tillbaka måndagen den 1 juli.

Salgo60 updated the task description. (Show Details)Jun 11 2019, 7:30 PM
Restricted Application added a subscriber: Liuxinyu970226. · View Herald TranscriptJun 11 2019, 7:30 PM
Salgo60 updated the task description. (Show Details)Jun 11 2019, 8:04 PM
Salgo60 updated the task description. (Show Details)Jun 12 2019, 8:06 AM

alvin-place

Salgo60 updated the task description. (Show Details)Jun 12 2019, 12:28 PM

alvin-record

Looks like this also could be an exhibition like Swedish Excavations at Sinda alvin-record:146876

I guess this could maybe also with other databases. Odd that no one tries to coordinate things like this or is Wikidata the best place

Salgo60 added a comment.EditedJun 12 2019, 2:28 PM

Had a phone call with Per Cullhed about Wikidata and Alvin...

He gets an email to this Task and can either update this task or contact me on 0705937579 / salgo60@msn.com

Things to discuss

  • get a load file and upload to Wikidata MixandMatch
    • I guess we need
      • alvin-person
        • Name
        • Birth/death
        • VIAF
        • SBL
        • see also what properties we have on the small test load link
        • Wikipedia links
      • alvin-record
          • ????
        • alvin-place
          • name
          • country
          • coordinates
          • ????

More:

Next step: action Per Cullhed

Salgo60 updated the task description. (Show Details)Jun 12 2019, 2:31 PM

Runestones feels not grouped together Search Runsten U 1048 shows object same as

Salgo60 updated the task description. (Show Details)Jun 12 2019, 8:56 PM
Salgo60 updated the task description. (Show Details)Jun 12 2019, 10:04 PM
Salgo60 updated the task description. (Show Details)Jun 13 2019, 1:18 AM

Thanks for your tweet alerting me to this, Magnus!

The majority of the photos of runestones in Alvin at the moment have been digitised as part of the Evighetsrunor project. They have been tagged with their runic signa (e.g. "U 1048") per the latest version of Samnordisk runtextdatabas. This is the main human-readable identifier we're using.

As part of Evighetsrunor, both Alvin and SRDB will become data providers to K-samsök.

We already have mappings from runic signum to existing K-samsök objects (and to RAÄ-numbers, in the case of runestones). As a consequence, once those two new datasets are in K-samsök with the images linked against the inscriptions, it will be very easy to create additional identities with the inscriptions' corresponding entries in Wikidata. In fact, that's something we intend to do – and if you look at the UGC links for existing runestones in K-samsök today you'll see that there already are such identities. This in turn will make it trivial to create direct "Alvin image –> depicts -> Wikidata runestone"- style statements, if that's something you're interested in doing (although by then such links should be implicit because of the aforementioned K-samsök-Wikidata sameAs identities).

If you want to create such "depicts" links from Wikidata to Alvin images now, you can do. But I would recommend waiting until Alvin and SRDB are in K-samsök, because then all of the mappings between the different identifiers (signum, RAÄ-number, Wikidata object, etc) will be provided for you, as well as links connecting all the images (not just from Alvin) to the inscriptions they depict. This will apply not just to existing Swedish inscriptions in K-samsök, but all inscriptions in SRDB.

So if you're prepared to wait another few months for the data to get into K-samsök, the process of linking against Wikidata will be made a lot easier, and a lot of it will probably have already been done!

(Evighetsrunor is a collaboration between RAÄ and UU, so if you're also in contact with Alvin please don't request changes to the runic material that may result in mixed messages or duplication of effort.)

Thanks feels like maybe we should group them at both places and try to have a consistency.... My unskilled guess is that its easy add identifiers in Alvin.....

Search Runsten U

Exactly. At present it's unclear exactly at which "end" (Alvin or SRDB) the links will live, but the most likely scenario is this: Based on the runic signa that are already in Alvin, we will then add to Alvin identifiers (URIs) for the inscriptions the photos depict. This will form part of the data that is then delivered to K-samsök. So it is likely that the identifiers will be in Alvin itself – no need to worry! :)
(The other alternative is that we would add links to the Alvin images to SRDB, which would then be provided to K-samsök. But putting the links there is less desirable.)

As to your example query… yes, if you search fot just "Runsten U", you should expect a lot of results! There are 1181 inscriptions in Upplands runinskrifter, and many more added after the last volume was published.

Salgo60 added a comment.EditedJun 13 2019, 8:01 AM

The have

  • alvin-place
  • alvin-person
  • alvin-record

Why not add

  • alvin-runestone or a more generic grouping....

Feels like Johannes Bureus needs to be part of the Alvin project and start get it to focus on our roots again ;)

Possibilities in Alvin

Looks they support some identifiers but has no depict or possibility to add depict in part of a picture

alvin-record:92895 Medeltida stiglucka vid Gothems kyrka Säve, Pehr Arvid, 1 has

  • subjects as Strings
  • no depict
  • no coordinates
  • description mostly with strings....

Salgo60 updated the task description. (Show Details)Jun 13 2019, 9:26 AM
Salgo60 updated the task description. (Show Details)Jun 13 2019, 12:11 PM
Salgo60 updated the task description. (Show Details)Jun 13 2019, 5:30 PM
Salgo60 updated the task description. (Show Details)Jun 13 2019, 6:35 PM
Salgo60 updated the task description. (Show Details)Jun 13 2019, 6:56 PM
Salgo60 updated the task description. (Show Details)
Salgo60 updated the task description. (Show Details)Jun 14 2019, 7:10 AM
Salgo60 updated the task description. (Show Details)Jun 14 2019, 7:13 AM

I'm not sufficiently familiar with Alvin to know exactly what you mean here, but I think I can guess. By adding "alvin-runestone" I presume you mean adding a new attribute where you could put the signum of the inscription depicted in the photo: <photo> -> alvin-runestone -> "U 1048" or something like that? If that's what you mean, then it's pretty much what we intend to do anyway, except we would want the relation to be some kind of "depicts": "alvin-runestone" is a bit ambiguous, since we want to make it clear that the photo shows an inscription, not that it is the inscription - especially as Alvin also includes two actual inscriptions (the objects, not just photos) from the Museum Gustavianum collection. And instead of the string "U 1048" we would of course have the URI for the inscription. (As you're so fond of saying, Magnus, "Things not strings!" 😉)

So on that point, if I've understood correctly, we're in full agreement - we already intended to do that! But... like I said, I'm not as familiar with Alvin as you clearly are, and I wasn't aware that they currently lack a "depicts"-style attribute. In that case, it would be good if they could add one so that we can do as I describe above; or failing that, do as you suggest and have an attribute for the signum (string) of the inscription depicted. K-samsök/Evighetsrunor have a meeting booked with Alvin after the summer, and I will raise this then; we need it in order to be able to harvest the data to K-samsök. Otherwise we will have to create such links outside of Alvin to be provided to K-samsök, which I'd very much like to avoid if possible.

I do not think it would be a good idea to model any new "alvin-runstone" attribute on the existing "alvin-person" such that Alvin would suddenly have separate, mostly empty, posts of their own for every runestone. That kind of duplication is best avoided; use existing identifiers, preferably URIs or failing that runic signum.

(If you're curious about the two actual inscriptions in Alvin, they are type-C bracteates, SE IK365,1 [urn:nbn:se:alvin:portal:record-16261] & SE IK365,7 [urn:nbn:se:alvin:portal:record-16263].)

Salgo60 added a subscriber: Abbe98.EditedJun 14 2019, 11:05 AM

For me just grouping adding a container that I easy can link "Give me all pictures of object type xxx".... not add all the description properties again.,... I spoke 2 seconds with Abbe that we should have like you have in programming "global" concepts that you tell other institutions "if you would like to interact with us we "expose" the following concepts

  1. persons using alvin-person
  2. locations using alvin-place
  3. concepts like runestones with "Category" Runestone and as a subcategory a specific runestone
  4. individual records with alvin-record

by exposing things like that I can see that we in a future we can easier get rid of Silos and easier find things.... best would be if refering same as something like Wikidata Q number

The pattern I tweetet/discussed with Olof Karswall, Riksarkivet last month was that we should start having "grunddata" also for culture things. E.g we have an official or inofficial authority for e.g. Runestones.....

  • In a loosely coupled community like Wikidata we define Shape Expressions how we will implement the concept and try to agree internally in WD
  • The "owner" of the "culture grunddata" does the same
    • in Wikidata we then try to agree what parts we implement and how to handle differences (could be that a RAÄ church is to complex for a "Wiki church") example ShEx in Wikidata E37 Human Genes

I can see when start playing with Alvin that e.g. alvin-place is rather difficult to understand and rather vague defined what it is (feels just like a coordinate Property 625) and I they use it sometimes same as a Swedish Church parish.... I guess Alvin lacks good query tools to produce maps and are never doing federation with other databases..... but I guess it makes sense to have a more distinct definition what an alvin-place is and also how they will interact with Riksarkivet Tora "Bebyggelsepunkter" / Wikidata / Riksarkivet SBL....

IF they defined it as a Swedish Church parish it would be easier for e.g. to link to Alvin on a Swedish Parish page saying please give me all objects related to this parish..... now we just guess if its a dot or all Alvin material in the parish.....
Vague defined concepts make it more difficult to reuse/link.... we need to start think linked data

What I feel is needed

  1. Explain if you are an Authority for some data and the status of your data do you have a QA procedure / taking care of errors
  2. Define the structure of your data and the quality of your data. Best is using a standard as ShEx
  3. Make a statement that you care about your data and will maintain it (not always the fact)
  4. Have tools like Phabricator so you can ask questions/ subscribe on changes / make it easier to discuss and coordinate between more organisations involved, or use WIkimedia Phabricator
  5. have some kind of feedback possibility per record and possibility to see the version history...
    1. Problem I see with Alvin after 1 hour
      1. looks like they have more alvin-person for the same person but I cant report a problem (alvin-person:4796 looks same as alvin-person:37500 link
        1. Query finds duplicates we have added in WD
      2. As always we need to say that it is maybe same as. I test a new pattern in WIkidata see list
      3. I prefer Wikidata way of adding a reference/source to a statement. Alvin feels not so focused giving sources
      4. search is rather simple.... I need to do some webscraping
    2. Good things
      1. They link VIAF and VIAF has Wikidata/Libris -- much easier find
      2. they also link not just the Swedish Wikipedia as same as
    3. To be improved
      1. they have sources but I am not 100% friend if its good or bad. Most sources is a text string telling a book title instead of linking an electronic source or LIBRIS XL
      2. more grouping I feel we need to have a common concept not just for Runestnes but also for Exhibitions done etc.....

SPARQL feels like a god candidate

Yesterday I showed @Abbe98 a SPARQL query that I feel makes SPARQL interface the prefered way of maintain/compare/access "grunddata"

  • my first try comparing Wikidata and @Nobelprize is rough but works for finding vandalization in Wikidata and also the problem not having sources
  • SPARQL federation comparing locations query. Feels like an excellent way to compare location of Runestones with an authority.....

How to depict

I guess we will learn much more when Wikicommons fully supporting SPARQL etc...

is a bit ambiguous, since we want to make it clear that the photo shows an inscription, not that it is the inscription

  1. ok in WIkidata I think I have seen using qualifiers and/or with the coming possibility with Wikicommon and SPARQL I guess you can ask for
    • all runestones from sweden that depicts a runetext (or maybe have text transcribed)

Example from Q12418

Salgo60 updated the task description. (Show Details)Jun 15 2019, 5:25 AM
Salgo60 updated the task description. (Show Details)Jun 15 2019, 10:29 AM
Salgo60 removed a subscriber: Salgo60.Jun 16 2019, 7:30 AM
Salgo60 updated the task description. (Show Details)Jun 16 2019, 7:33 AM
Salgo60 added a comment.EditedJun 16 2019, 7:27 PM

New report for constraint violations Alvin <> Wikidata

  • Status today how to do change management with Wikdata and external institution is bad formalized see EPIC T202530 and a suggestion to start have workshops with good patterns T219418#5207385
    • I have seen no institution that Wikidata has a formalized change process with that is trackable i.e. change management like Phabricator and tasks IDs as we have in Phabricator like T225522 is not used
        • Swedish Film institute use Wikidata / en:Wikipedia for updating death dates see blogpost I guess that changes at Swedish Film institute are not written back to Wikidata
      • We miss formalized commitments that our data should have no duplicates in Wikidata or a procedure what data is mapped etc... I hope we can formalize some part with tools like Shape expression
      • We miss standardized flows from most data vendors like
        • recent changes in WIkidata
        • a commitment to technologies like SPARQL to make integration easier between "SILOS"
          • the Swedish National Archives we have had some contact with but its email, twitter and add hoc and they told us 2018 feb tweet that Jens Kling is investigating how to work with Open Data, We are still waiting.... I am coming from a background working with international bank transactions not having traceability is just asking for problems and make the change cycle unpredictable and much more expensive... I guess culture data institutions are very immature and have no requirements to be efficient when working with linked data... but that is not an excuse doing a bad work. Linked data needs another skill matrix compared to before is my lesson learned playing with Wikipedia/Wikidata and much more dialogue is my guess link
Salgo60 updated the task description. (Show Details)Jun 17 2019, 7:23 AM
Salgo60 updated the task description. (Show Details)
Salgo60 updated the task description. (Show Details)Jun 17 2019, 4:07 PM
Salgo60 updated the task description. (Show Details)
Salgo60 updated the task description. (Show Details)Jun 17 2019, 10:22 PM
Salgo60 updated the task description. (Show Details)Jun 17 2019, 11:02 PM
Salgo60 updated the task description. (Show Details)Jun 18 2019, 12:30 PM
Salgo60 added a comment.EditedJun 18 2019, 3:08 PM

Created template sv:WIkipedia

Feedback was a better Landing page

Salgo60 updated the task description. (Show Details)Jun 23 2019, 1:39 PM
Salgo60 added a comment.EditedJun 24 2019, 9:04 AM

Phone call nr 2 with Per Cullhed and explained that we now have connected about 1000 items and I suggested they could have a look into it and see if this will add value for them... plus that we have found some duplicates..... feels like they have no Linked data strategy

Salgo60 added a comment.EditedJun 24 2019, 1:03 PM

Alvin mallen avskaffad från sv:Wikimedia kanske bättre landningssidor gör Wikipedia redigerare gladare...

Compare different landingpages

Salgo60 updated the task description. (Show Details)Jun 24 2019, 1:50 PM
Salgo60 added a comment.EditedJun 24 2019, 2:27 PM

Interesting comment if we can have all Alvin people locations in Wikidata

Salgo60 updated the task description. (Show Details)Jun 25 2019, 9:46 AM
Salgo60 updated the task description. (Show Details)Jun 25 2019, 12:57 PM
Salgo60 added a comment.EditedJun 25 2019, 7:37 PM

sent a question to the TORA project how they see themself same as or ???

As "same as" means that the two things share all the same properties and so every property of one thing can be inferred to be a property of the other
My (Magnus) feels that a TORA point will never be same as the Ksamsök object as Ksamsök points on the objects found today and will have another set of properties is my guess...

Lists

list when just 128 locations has been added

Salgo60 updated the task description. (Show Details)Jun 27 2019, 12:50 PM
Salgo60 updated the task description. (Show Details)Jul 2 2019, 6:27 AM
Salgo60 updated the task description. (Show Details)Jul 3 2019, 8:23 AM
Salgo60 updated the task description. (Show Details)Jul 3 2019, 12:22 PM
Salgo60 updated the task description. (Show Details)Jul 4 2019, 12:23 PM
Salgo60 updated the task description. (Show Details)Jul 4 2019, 6:47 PM
Salgo60 updated the task description. (Show Details)Jul 5 2019, 11:58 AM
Salgo60 updated the task description. (Show Details)Jul 5 2019, 2:37 PM
Salgo60 updated the task description. (Show Details)Jul 7 2019, 1:41 PM
Salgo60 updated the task description. (Show Details)Thu, Aug 1, 12:24 AM
Salgo60 updated the task description. (Show Details)Fri, Aug 2, 7:46 AM
Salgo60 updated the task description. (Show Details)Sun, Aug 4, 5:13 AM
Salgo60 updated the task description. (Show Details)Mon, Aug 5, 4:02 PM
Salgo60 updated the task description. (Show Details)Wed, Aug 7, 2:32 PM
Salgo60 updated the task description. (Show Details)Wed, Aug 7, 7:26 PM
Salgo60 updated the task description. (Show Details)Fri, Aug 9, 4:53 AM
Salgo60 updated the task description. (Show Details)Fri, Aug 9, 5:28 AM
Salgo60 updated the task description. (Show Details)Sun, Aug 11, 10:48 AM
Salgo60 updated the task description. (Show Details)
Salgo60 updated the task description. (Show Details)Sun, Aug 11, 11:18 AM
Salgo60 updated the task description. (Show Details)Sun, Aug 11, 11:27 AM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 9:27 AM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 9:29 AM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 9:36 AM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 12:01 PM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 1:05 PM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 1:07 PM
Salgo60 updated the task description. (Show Details)Mon, Aug 12, 1:22 PM