Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Sep 24 08:00
    ericprud commented #42
  • Sep 24 08:00
    ericprud commented #42
  • Sep 24 07:59
    ericprud commented #42
  • Sep 23 17:58
    gkellogg commented #42
  • Sep 23 15:49
    ericprud commented #42
  • Sep 23 09:43
    ericprud commented #41
  • Sep 23 09:37
    ericprud closed #41
  • Sep 23 09:37

    ericprud on gh-pages

    ~ example of IMPORTS operator t… (compare)

  • Sep 23 09:29

    ericprud on gh-pages

    ~ align with definition http:/… (compare)

  • Sep 23 09:23
    ericprud commented #43
  • Sep 07 19:46
    gkellogg opened #43
  • Sep 03 13:13

    ericprud on gh-pages

    ~ broken fragment ref (compare)

  • Sep 03 13:10

    ericprud on gh-pages

    ~ s/rawgit/shex.io/ (compare)

  • Aug 29 20:09
    gkellogg opened #42
  • Aug 29 20:07
    gkellogg opened #41
  • Aug 22 13:18
    pdehaye opened #8
  • Jul 22 17:26
    ericprud labeled #116
  • Jul 22 17:26
    ericprud opened #116
  • Jul 21 11:57

    tombaker on master

    old agendas/minutes (compare)

  • Jul 12 06:12
    ericprud commented #23
Eric Prud'hommeaux
@ericprud
@pdehaye , i'm not super skilled with advanced calendaring tech, but i recall there being a meeting 13 days ago, suggesting that there will be one tomorrow.
Paul-Olivier Dehaye
@pdehaye
thanks for the response. the q was in relation to holidays, more than the 50% chance of it being the right week :)
Eric Prud'hommeaux
@ericprud:matrix.org
[m]
+1
Eric Prud'hommeaux
@ericprud
hi gang, en route
Eric Prud'hommeaux
@ericprud
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr

During my investigations on the reference quality of Wikidata, I extracted 6 subsets from Wikidata corresponding to 6 Wikidata WikiProjects. The outcomes were used for a paper that was submitted to Wikidata Workshop 2021. The GitHub repo for the paper contains ShEx schemas of each subset for the community reuse if you want to slurp the subsets by ShEx validators.

https://github.com/seyedahbr/Wikidata_Reference_Statistics

Eric Prud'hommeaux
@ericprud
Eric Prud'hommeaux
@ericprud
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr

During my investigations on the reference quality of Wikidata, I extracted 6 subsets from Wikidata corresponding to 6 Wikidata WikiProjects. The outcomes were used for a paper that was submitted to Wikidata Workshop 2021. The GitHub repo for the paper contains ShEx schemas of each subset for the community reuse if you want to slurp the subsets by ShEx validators.

https://github.com/seyedahbr/Wikidata_Reference_Statistics

The extracted subsets (.nt.gz via WDumper) are in https://doi.org/10.5281/zenodo.5117927 BTW.

As I said in the call, I think currently ShEx validators can not deal with large-scale slurping. There are also issues in extracting references. I mean if you want to extract all references for an item you need to know every property used in that item and write the shapes. There is no option as far as I know to extract all references (or qualifiers) of all statements of an item in SheX schemas. I think it would be useful to work on the slurping capabilities of ShEx validators especially slurping from local RDF dumps and some features like extracting references. I don't know it needs changing the syntax or can be done just by an option because the references reification is somehow unique in Wikidata.
SJ
@metasj:matrix.org
[m]
I'd like to understand better an ex. of what you want & don't have from ingestion + ref extraction at scale.
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr

@metasj:matrix.org About subsetting at scale: First of all, just shex-js and PyShex support slurping. If you try to validate+slurp a schema in both over Wikdiata endpoint (without LIMIT) both validators will stop (PyShex will stop and shex-js face with the 'Maximum call stack size exceeded' error). There is no option to validate+slurp data from a local RDF file (instead of an endpoint).

About the reference slurping: Assume you want the references as complete as possible. If you want to extract references in a shape, you need to know the current properties that are used in the item. If another property will be added in the future, you need to update your shape with the newly added property. For example, assume all gene instaces have only P31. You can slurp references using this shape:
<gene> { p:P31 { ps:P31 wd:Q7187 ; prov:wasDerivedFrom { pr:P813 xsd:dateTime ; pr:P686 xsd:string ; prv:P813 IRI ; ... } } ;}
If in the future, a P279 will be added to an instance of genes in Wikidata, you need to write a similar shape for it too. plus that all references don't have similar triples (some have just stated in. some have two, etc.).

I hope I could explain the problem here!

Paul-Olivier Dehaye
@pdehaye
@ericprud @ericprud:matrix.org hi! where was your more complex example involving "materialize"? all the links here to the live tool are broken :-( http://shex.io/extensions/Map/
Sameroom
@sameroom-bot
[Eric Prud'hommeaux] @pdehaye, I'll have eyes here in 20 mins...
Eric Prud'hommeaux
@ericprud
@pdehaye , links should be fixed now
1 reply
Paul-Olivier Dehaye
@pdehaye
@ericprud i am curious about how you compare a shex entry that appears in ShexML to one that appears in the RHS of your advanced Shex example. Don't these two serve as "target", once information has already been loaded in memory? cc @labra
Eric Prud'hommeaux
@ericprud
That makes intuitive sense to me but I've not studied ShExML
Gregg Kellogg
@gkellogg
Question on import tests (e.g., https://github.com/shexSpec/shexTest/blob/main/schemas/1dotIMPORT1dot.shex). It references "1dot", which for which no file exists. Presumably, the test instructions should say to append the appropriate file extension if it doesn't respond, but only in a testing scenario. Can't find anything else about it anywhere, and the test README instructions seem out of date.
Eric Prud'hommeaux
@ericprud:matrix.org
[m]
I guess the test drive should say that the harness needs conneg to do IMPORT tests
The idea was that the importing schema wouldn't prescribe a particular format
Eric Prud'hommeaux
@ericprud:matrix.org
[m]
Which is super cool
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Has the start time of the CG meeting changed?
Eric Prud'hommeaux
@ericprud
Hiya Sayed. There's some ambiguity about when the meetings are
I'd assumed it was every other wed, but the prevailing interpretation is that it's the second and fourth Wednesdays
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Ahhh that's the choice! OK will see you on 13th so
Eric Prud'hommeaux
@ericprud
Ohh, maybe 1st and 3rd Wednesdays
Sameroom
@sameroom-bot
[Andra Waagmeester] What does “Error: Structural error: circular negative dependencies “ entail again?
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Hi, may one of the ShEx syntax experts please check a shape I wrote (here) and tell me if it simulates the general reference structure of Wikidata (based on this fig) correctly? I don't want you to check details, I'd just like to know does the way I defined and used shape labels (like p:<property>) work or not? Thanks a lot!
Eric Prud'hommeaux
@ericprud
hi @seyedahbr , i tweaked your syntax to help you get the hang of ShExC (Shape Expression Compnact syntax)
Eric Prud'hommeaux
@ericprud
as for the structure, https://www.wikidata.org/wiki/EntitySchema:E113 is well-polished and illustrates how a property (e.g. p:P279) is captured in statements which have triples with the ps form of the property (e.g. ps:P279) and are connected to provenance by a prov:wasDerivedFrom.
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Thanks @ericprud. E113 and other entity schema are perfect, but they are specific for a few set of properties/refs in a single topic. I was looking for a general shape exp. that works on any property set in any topic and only examines the wikidata RDF model and the correct use of namespaces p:, wdref:,... .
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Thanks a lot indeed, I think your tweak is general enough!
Eric Prud'hommeaux
@ericprud
In ShEx 1, we had some notion of wildcard predicates. It's possible to add that back in the ShEx 2, but we want to really understand the use cases well before committing to some syntax and semantics.
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Not sure what kind of use case you want, but in my case, I'm looking to syntactically assess a given Wikibase-based RDF dump to be sure the refnodes/triples/values/properties completely follow the Wikibase reification model and proper namespaces (not just a specific type of statements/properties/refnode etc.).
Eric Prud'hommeaux
@ericprud
Interesting. That means we also need a way to make one predicates a string function of another
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
Yes exactly!
danbri
@danbri:matrix.org
[m]
Who from here will be at the biohackathon
Seyed Amir Hosseini Beghaeiraveri
@seyedahbr
@danbri:matrix.org Hi, I'll be there!
danbri
@danbri:matrix.org
[m]
Great! I will be virtually there
Eric Prud'hommeaux
@ericprud
Andra, Jose Labra and I will also be there
jaxoncreed
@jaxoncreed
Hey, I’ve been out of it for a while. What’s the most up-to-date shex javascript implementation now? I see https://www.npmjs.com/package/shex hasn’t had a contribution for 4 years.
I also see libraries like @shexjs/validator but can’t find the documentation on how to use them
Eric Prud'hommeaux
@ericprud:matrix.org
[m]
@jaxoncreed, look for@shexjs
jaxoncreed
@jaxoncreed
Got it. Do you know where the documentation for those are? It doesn’t seem like any of the @shexjs libraries have ReadMes
Eric Prud'hommeaux
@ericprud:matrix.org
[m]
Maybe like 1? Somewhere?
I'll be back at keyboard in 1hr?
Eric Prud'hommeaux
@ericprud
i've honestly been really lazy about this
(well, distracted, but the effect is the same)
validate gives a good idea of the requires