Loading ...
Sorry, an error occurred while loading the content.

4648Re: [TaxoCoP] Converting Data to SKOS assistance

Expand Messages
  • Gabriel Tanase
    Oct 30, 2013
    • 0 Attachment
      Indeed, that is the intended meaning.

      IMHO it might have been more obvious if the source database had been normalized e.g. in three tables: Concept(Code, ...) , ConceptAttribute(Code, Name, Value) and ConceptAttributeSource(Code, Name, Value).
      Maybe Nathan does not have control over how data is made available in the database.

      My follow-up question is: since the implementation would be based on customizing the underlying OWL ontology, rather than on plain annotations, how simply would tools that handle annotations out-of-the-box be able to show values from the added property as annotations, or in a manner very similar to how they display annotations?
      Perhaps this is not possible at all, and - hopefully - perhaps this does not matter for the end-users / downstream systems' perspective.
      All they might need is to access the annotative data in a meaningful manner.


      Thank you,
      Gabriel
      ----------
      http://ie.linkedin.com/in/gabrieltanase



      On 29 October 2013 22:14, Bob DuCharme <bducharme@...> wrote:


      Doing that, you could assign three Q10 values, three Q12 values, and three Q47 values to concept C113266, but I believe he wants to record the fact that first Q10, Q12, and Q47 values go together (as Annotation set 123), the second Q10, Q12, and Q47 values go together (as Annotation set 456), etc., which would require an extra data structure. 

      Bob


      On Tue, Oct 29, 2013 at 5:46 PM, Jim Tivy <jimt@...> wrote:
       

      Why not just add a few more properties to skos:Concept that contain your extra data.

       

      From: TaxoCoP@yahoogroups.com [mailto:TaxoCoP@yahoogroups.com] On Behalf Of Bob DuCharme
      Sent: October-29-13 2:11 PM
      To: TaxoCoP@yahoogroups.com
      Subject: Re: [TaxoCoP] Converting Data to SKOS assistance

       

       

      SKOS has no way to group annotations together into packages with their own identity (e.g. http://some/path/123, http://some/path/456, http://some/path/789) separate from the identity of Concepts and, in SKOS-XL, from Labels. However, because  SKOS is an OWL ontology it's extensible, so you could declare an xyz:Annotation class in your own namespace that has the properties you need. Then, to attach each annotation package to a Concept, I would declare a new property, perhaps as a subproperty of skos:notation, that has skos:Concept as its domain (to show that it's a property of concepts) and xyz:Annotation as its range to show that that's where the values come from. 

       

      The flexibility of SKOS-XL is great in theory but makes it more difficult to implement an easy-to-use tool for manipulating the data, and if you make the model even more complex by adding a customized xyz:Annotation class, then specialized SKOS tools will not be able to handle something like this out of the box, so you'd have to use a more generalized RDF tool like TopBraid Composer or Protege. 

       

      (Full disclosure: I work for TopQuadrant, which makes the SKOS-based EVN vocabulary manager EVN as well as TopBraid Composer.)

       

      Bob

       

      On Tue, Oct 29, 2013 at 4:07 PM, Nathan Wilson <wilsns01@...> wrote:

       

      All,

       

      I am working on a project which involves converting data from a database into a SKOS-XL format and I have an issue involving annotative data.

       

      For each term there are three pieces of annotative data:

        1. Source (Q10)

        2. Source_Domain (Q47)

        3. Source_Local_Code (Q12)

       

      When I convert the data to SKOS I get the following output:

        <rdf:Description rdf:about="#L113266">
          <rdf:type rdf:resource="http://www.w3.org/2008/05/skos-xl#Label"/>
          <skosxl:literalForm rdf:datatype="http://www.w3.org/2001/XMLSchema#string">Black (Not of Hispanic Origin)</skosxl:literalForm>
          <Q10 rdf:datatype="http://www.w3.org/2001/XMLSchema#string">PROTRACK-CC</Q10>
          <Q10 rdf:datatype="http://www.w3.org/2001/XMLSchema#string">NIAAA</Q10>
          <Q10 rdf:datatype="http://www.w3.org/2001/XMLSchema#string">MIS-CC</Q10>
          <Q12 rdf:datatype="http://www.w3.org/2001/XMLSchema#string">Black (Not of Hispanic Origin)</Q12>
          <Q12 rdf:datatype="http://www.w3.org/2001/XMLSchema#string">B</Q12>
          <Q47 rdf:datatype="http://www.w3.org/2001/XMLSchema#string">Race and Ethnicity Combined</Q47>
        </rdf:Description>

       

      In the data base there is an ID which is used to group annotative data together:

      ID    Concept Code    Attribute Code    Attribute Name    Attribute Value    Annotation Code    Annotation Name    Annotation Value
      123    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q10    Source    PROTRAK-CC
      123    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q12    Source_Local_Code    Black (Not of Hispanic Origin)
      123    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q47    Source_Domain    Race and Ethnicity Combined
      456    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q10    Source    NIAAA
      456    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q12    Source_Local_Code    B
      456    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q47    Source_Domain    Race and Ethnicity Combined
      789    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q10    Source    MIS-CC
      789    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q12    Source_Local_Code    B
      789    C113266    P90    SYN    Black (Not of Hispanic Origin)    Q47    Source_Domain    Race and Ethnicity Combined

       

      Is there a way to include this grouping in SKOS-XL or is this grouping unique to the database and not implementable in the vocabulary?

       

      Thanks

       

      Nathan

       



    • Show all 9 messages in this topic