citation property in Dataset

13 Jan 2020

Dear all,
During our last meeting we discussed what seems to be contradictory
information on the Google guidelines regarding the Dataset schema.org type (
https://developers.google.com/search/docs/data-types/dataset).
After reading that document again, I think there is no contradiction there.
Let's suppose we have a dataset named "OurDataset" where AA, BB, CC and DD
have contributed and we have a paper about that dataset titled "OurDataset,
integrating omics data" with only AA as author.
What Google recommends is something like:
"OurDataset" object -> citation -> "OurDataset, integrating omics data"
object
{
"@type": "Dataset",
"name": "OurDataset",
"citation": {
"@type": "ScholarlyArticle",
"name": "OurDataset, integrating omics data",
"author": {
"@type": "Person",
"name": "AA"
}
}
}
What Google asks us not to do is something like
"OurDataset" object -> citation -> "OurDataset" object
{
"@type": "Dataset",
"name": "OurDataset",
"citation": {
"@type": "Dataset",
"name": "OurDataset"
}
}
Do you agree this is what is written in that document? Should I still try
to reach Dan Brickley from schema.org?
Kind regards,

  • Adam Shepherd's picture

    Author: Adam Shepherd

    Date: 13 Jan, 2020

    I agree with you, Leyla
    —Adam

  • Mingfang Wu's picture

    Author: Mingfang Wu

    Date: 03 Feb, 2020

    Yes Leyla, this is what I understand as well. The google guidelines is
    consistent with the definition of citation from the schema.org:
    https://schema.org/citation
    Thanks,
    Ming

  • Makx Dekkers's picture

    Author: Makx Dekkers

    Date: 03 Feb, 2020

    If I understand correctly, sdo:citation is being used here in the sense of “is supplementary material for”. If this is the semantic meaning of sdo:citation, it is a bit different from the usual meaning of citation.
    Usually, I would expect citation to be one-way: ArticleA cites ArticleB.
    But in the meaning implied by the Dataset guidelines, both the following statements would be valid:
    ArticleA sdo:citation DatasetD (the usual meaning: the article cites the dataset)
    DatasetD sdo:citation ArticleA (the additional meaning: the dataset requires the article to be cited alongside the dataset)
    I find this confusing.
    Makx.
    From: ***@***.***-groups.org <***@***.***-groups.org>
    Sent: 03 February 2020 12:04
    To: LJ.Garcia <***@***.***>; Research Metadata Schemas WG <***@***.***-groups.org>
    Cc: Adam Shepherd <***@***.***>
    Subject: Re: [rda-research-schemas-wg] citation property in Dataset
    Yes Leyla, this is what I understand as well. The google guidelines is consistent with the definition of citation from the schema.org : https://schema.org/citation
    Thanks,
    Ming
    On Mon, Jan 13, 2020 at 11:43 PM Adam Shepherd <***@***.*** > wrote:
    I agree with you, Leyla
    —Adam
    On Jan 13, 2020, at 3:57 AM, ljgarcia via Research Metadata Schemas WG <***@***.***-groups.org > wrote:
    
    Dear all,
    During our last meeting we discussed what seems to be contradictory information on the Google guidelines regarding the Dataset schema.org type (https://developers.google.com/search/docs/data-types/dataset).
    After reading that document again, I think there is no contradiction there. Let's suppose we have a dataset named "OurDataset" where AA, BB, CC and DD have contributed and we have a paper about that dataset titled "OurDataset, integrating omics data" with only AA as author.
    What Google recommends is something like:
    "OurDataset" object -> citation -> "OurDataset, integrating omics data" object
    {
    "@type": "Dataset",
    "name": "OurDataset",
    "citation": {
    "@type": "ScholarlyArticle",
    "name": "OurDataset, integrating omics data",
    "author": {
    "@type": "Person",
    "name": "AA"
    }
    }
    }
    What Google asks us not to do is something like
    "OurDataset" object -> citation -> "OurDataset" object
    {
    "@type": "Dataset",
    "name": "OurDataset",
    "citation": {
    "@type": "Dataset",
    "name": "OurDataset"
    }
    }
    Do you agree this is what is written in that document? Should I still try to reach Dan Brickley from schema.org ?
    Kind regards,
    --
    Full post: https://www.rd-alliance.org/group/research-metadata-schemas-wg/post/cita...
    Manage my subscriptions: https://www.rd-alliance.org/mailinglist
    Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/67431
    --
    Dr Mingfang Wu
    Australian Research Data Commons
    T: +61 3 9902 4646 | E: ***@***.*** | O:0000-0003-1206-3431
    Physical address: Monash University, Building T, Ground Floor, 100 Sir John Monash Drive, Caulfield East, VIC 3145

  • Leyla Garcia's picture

    Author: Leyla Garcia

    Date: 03 Feb, 2020

    Hi all,
    It is indeed confusing. I have added a comment to an open schemaorg issue
    in GitHub: https://github.com/schemaorg/schemaorg/issues/1031
    Please have a look to contribute to the discussion as you see it fits.
    Hopefully @danbri will reply to it.
    Regards,
    On Mon, Feb 3, 2020 at 1:29 PM makxdekkers via Research Metadata Schemas WG
    <***@***.***-groups.org> wrote:

  • Robert Casey's picture

    Author: Robert Casey

    Date: 03 Feb, 2020

    Interesting discussion. This sounds like the issue relating to Schema.org's current recommendations for Metadata, where there are two different terms used depending on the direction of reference, but the reference is nonetheless bidirectional:
    From: https://github.com/ESIPFed/science-on-schema.org/blob/master/decisions/4...
    Linking Metadata docs to Datasets: Use schema:subjectOf or schema:about
    schema:subjectOf
    or inversely
    schema:about
    Could we consider using the terms 'Citations' and 'References' as directionally opposite in the same manner? Or perhaps there's a better term.
    -Rob

  • Makx Dekkers's picture

    Author: Makx Dekkers

    Date: 03 Feb, 2020

    To me the issue is not that there are two properties that are each other’s inverse, like schema:subjectOf and schema:about, but that there is one property schema:citation that is used in two different ways, not even with an inverse relationship. One meaning is to point from one thing to another (in line with the definition ), the other one, from the Dataset example, to say something like “use this in combination with that” which is completely different.
    In my mind, if “A schema:citation B” means “A cites or references B”, then the statement “DatasetD schema:citation ArticleX” can only mean “DatasetD cites or references ArticleX” and nothing else. Otherwise, one would allow the meaning of the property to change based on the type of thing it is applied to.
    By the way, I cannot find the property ‘references’ in schema.org. It is in Dublin Core, which also has its inverse ‘isReferencedBy’.
    Makxz.
    From: Robert Casey <***@***.***>
    Sent: 03 February 2020 19:11
    To: ***@***.***
    Cc: Makx Dekkers <***@***.***>; Research Metadata Schemas WG <***@***.***-groups.org>
    Subject: Re: [rda-research-schemas-wg] citation property in Dataset
    Interesting discussion. This sounds like the issue relating to Schema.org 's current recommendations for Metadata, where there are two different terms used depending on the direction of reference, but the reference is nonetheless bidirectional:
    From: https://github.com/ESIPFed/science-on-schema.org/blob/master/decisions/4...
    Linking Metadata docs to Datasets: Use schema:subjectOf or schema:about
    schema:subjectOf
    or inversely
    schema:about
    Could we consider using the terms 'Citations' and 'References' as directionally opposite in the same manner? Or perhaps there's a better term.
    -Rob
    On Feb 3, 2020, at 6:40 AM, ljgarcia via Research Metadata Schemas WG <***@***.***-groups.org > wrote:
    Hi all,
    It is indeed confusing. I have added a comment to an open schemaorg issue in GitHub: https://github.com/schemaorg/schemaorg/issues/1031
    Please have a look to contribute to the discussion as you see it fits. Hopefully @danbri will reply to it.
    Regards,
    On Mon, Feb 3, 2020 at 1:29 PM makxdekkers via Research Metadata Schemas WG <***@***.***-groups.org > wrote:
    If I understand correctly, sdo:citation is being used here in the sense of “is supplementary material for”. If this is the semantic meaning of sdo:citation, it is a bit different from the usual meaning of citation.
    Usually, I would expect citation to be one-way: ArticleA cites ArticleB.
    But in the meaning implied by the Dataset guidelines, both the following statements would be valid:
    ArticleA sdo:citation DatasetD (the usual meaning: the article cites the dataset)
    DatasetD sdo:citation ArticleA (the additional meaning: the dataset requires the ar ticle to be cited alongside the dataset)
    I find this confusing.
    Makx.
    From: ***@***.***-groups.org <***@***.***-groups.org >
    Sent: 03 February 2020 12:04
    To: LJ.Garcia <***@***.*** >; Research Metadata Schemas WG <***@***.***-groups.org >
    Cc: Adam Shepherd <***@***.*** >
    Subject: Re: [rda-research-schemas-wg] citation property in Dataset
    Yes Leyla, this is what I understand as well. The google guidelines is consistent with the definition of citation from the schema.org : https://schema.org/citation
    Thanks,
    Ming
    On Mon, Jan 13, 2020 at 11:43 PM Adam Shepherd <***@***.*** > wrote:
    I agree with you, Leyla
    —Adam
    On Jan 13, 2020, at 3:57 AM, ljgarcia via Research Metadata Schemas WG <***@***.***-groups.org > wrote:
    
    Dear all,
    During our last meeting we discussed what seems to be contradictory information on the Google guidelines regarding the Dataset schema.org type (https://developers.google.com/search/docs/data-types/dataset).
    After reading that doc ument again, I think there is no contradiction there. Let's suppose we have a dataset named "OurDataset" where AA, BB, CC and DD have contributed and we have a paper about that dataset titled "OurDataset, integrating omics data" with only AA as author.
    What Google recommends is something like:
    "OurDataset" object -> citation -> "OurDataset, integrating omics data" object
    {
    "@type": "Dataset",
    "name": "OurDataset",
    "citation": {
    "@type": "ScholarlyArticle",
    "name": "OurDataset, integrating omics data",
    "author": {
    "@type": "Person",
    "name": "AA"
    }
    }
    }
    What Google asks us not to do is something like
    "OurDataset" object -> citation -> "OurDataset" object
    {
    "@type": "Dataset",
    "name": "OurDataset",
    "citation": {
    "@type": "Dataset",
    "name": "OurDataset"
    }
    }
    Do you agree this is what is written in that document? Should I still try to reach Dan Brickley from schema.org ?
    Kind regards,
    --
    Full post: https://www.rd-alliance.org/group/research-metadata-schemas-wg/post/cita...
    Manage my subscriptions: https://www.rd-alliance.org/mailinglist
    Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/67431
    --
    Dr Mingfang Wu
    Australian Research Data Commons
    T: +61 3 9902 4646 | E: ***@***.*** | O:0000-0003-1206-3431
    Physical address: Monash University, Building T, Ground Floor, 100 Sir John Monash Drive, Caulfield East, VIC 3145
    --
    Full post: https://www.rd-alliance.org/group/research-metadata-schemas-wg/post/cita...
    Manage my subscriptions: https://www.rd-alliance.org/mailinglist
    Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/67431
    --
    Full post: https://www.rd-alliance.org/group/research-metadata-schemas-wg/post/cita...
    Manage my subscriptions: https://www.rd-alliance.org/mailinglist
    Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/67431

  • Robert Casey's picture

    Author: Robert Casey

    Date: 03 Feb, 2020

    Good arguments, Makx. What came to mind from Seismology, for data sets and data products, is that the Data Set will by cited in Papers that use it for study (Paper schema:citation DataSet) but the investigator that created the Data Set will insist that it is accompanied by a landing page that References a document describing the experiment or study that created the data (DataSet schema:references Paper). The DataSet must exist prior to studies that cite its use, so there's some directionality there, but the origin of the DataSet points to the foundational Paper written by the PI, so that is also directional.
    The fundamental issue we've identified is that schema:citation can accidentally be cyclic, and I think we want the referential graph to be acyclic so we can know what is prior art.
    -Rob

  • Robert Casey's picture

    Author: Robert Casey

    Date: 03 Feb, 2020

    Now that I read what I typed a moment ago, I think that my second example was incorrect and should actually be (DataSet schema:citation Paper). Going with that, I like Makx's suggestion of 'isReferencedBy', since the reverse reference in the case of a PI's experiment notes would be (Paper schema:isReferencedBy DataSet). In this case the arrow points from B to A. I think that better illustrates where opposing terms apply.
    -Rob

submit a comment