Dealing with embedded entries

[laurentromary]

It is common practice to embed entries within entries, for instance when compound are described in the entry of the main head. The constructs are (re for related entry):

<entry>
   <re>...</re>
</entry>

or

<entry>
   <entry>...</re>
</entry>

Is there one or several (for different types of relations between the main and sub) for this?

[laurentromary]

Here's an example:

 <entry xml:id="aboi" n="1906-001_unknown">
   <form type="lemma"><orth>ABOI</orth></form>
   <gramGrp>
      <pos expand="nom">n.</pos>
      <gen expand="masculin">m.</gen>
   </gramGrp>
   <etym><pc>(</pc>de <mentioned>aboyer</mentioned><pc>)</pc><pc>.</pc></etym>
   <sense>
      <def>Cri du chien</def><pc>.</pc>
   </sense>
   <re type="gram">
      <gramGrp><pos expand="nom">N.</pos>
            <gen expand="masculin">m.</gen>
            <number expand="pluriel">pl.</number></gramGrp>
       <sense><def>Dernières extrémités où le cerf est réduit</def><pc>.</pc></sense>
       <sense><usg type="style" rend="italic" expand="figuré">Fig.</usg>
            <def>Situation désespérée</def>
           <pc>:</pc>
            <cit type="example">
              <quote>commerçant ruiné et aux abois</quote>
            </cit><pc>.</pc>
      </sense>
   </re>
</entry>

[jmccrae]

This may be quite tricky as you would need to create a lexicog:Entry and relate it to the ontolex:LexicalEntry as described in https://www.w3.org/2019/09/lexicog/

BTW the <re> tag does not seem to be part of the spec

[laurentromary]

Re: <re> I am trying to keep some alignment with mainstream TEI. If we know how to do one we could as well implement the other.

[laurentromary]

Can you outline a snippet in XML on how you would do this?

[laurentromary]

The following should be a TEI Lex 0 compatible structure that you could add to the test samples:

<entry xml:id="aboi" n="1906-001_unknown">
   <form type="lemma"><orth>ABOI</orth></form>
   <gramGrp>
      <pos expand="nom">n.</pos>
      <gen expand="masculin">m.</gen>
   </gramGrp>
   <etym><pc>(</pc>de <mentioned>aboyer</mentioned><pc>)</pc><pc>.</pc></etym>
   <sense>
      <def>Cri du chien</def><pc>.</pc>
   </sense>
   <entry type="gram">
      <gramGrp><pos expand="nom">N.</pos>
            <gen expand="masculin">m.</gen>
            <number expand="pluriel">pl.</number></gramGrp>
       <sense><def>Dernières extrémités où le cerf est réduit</def><pc>.</pc></sense>
       <sense><usg type="style" rend="italic" expand="figuré">Fig.</usg>
            <def>Situation désespérée</def>
           <pc>:</pc>
            <cit type="example">
              <quote>commerçant ruiné et aux abois</quote>
            </cit><pc>.</pc>
      </sense>
   </entry>
</entry>

[jmccrae]

So you need something like this

<lexicog:Entry>
  <lexicog:describes>
     <ontolex:LexicalEntry rdf:ID="aboi">
        <!-- As before -->
     </ontolex:LexicalEntry>
  </lexicog:describes>
  <lexicog:subComponent>
    <lexicog:Entry>
      <lexicog:describes>
        <ontolex:LexicalEntry>
            <!-- Subentry goes here --> 
        </ontolex:LexicalEntry>
      </lexicog:describes>
    </lexicog:Entry>
  </lexicog:subComponent>
</lexicog:Entry>

I think the issue is we want to generate all this structure only when there are subentries.

[laurentromary]

Quick question: can we have this without the subcomponents, i.e.:

<lexicog:Entry>
  <lexicog:describes>
     <ontolex:LexicalEntry rdf:ID="aboi">
        <!-- As before -->
     </ontolex:LexicalEntry>
  </lexicog:describes>
</lexicog:Entry>

Or in this case, the lexicog:Entry should be dropped. I am asking to see whether I need to test the existence of subentries.

[laurentromary]

I have implemented this with a test. Could you have a look at: https://github.com/elexis-eu/tei2ontolex/blob/master/GeneratedOntolex/PLI_1906_20190109%20sample.xml

[jmccrae]

Here is the results of running it through RDF validation

-> % rapper -o turtle GeneratedOntolex/PLI_1906_20190109\ sample.xml >/dev/null
rapper: Parsing URI file:///home/jmccrae/projects/elexis/tei2ontolex/GeneratedOntolex/PLI_1906_20190109%20sample.xml with parser rdfxml
rapper: Serializing with serializer turtle
rapper: Error - URI file:///home/jmccrae/projects/elexis/tei2ontolex/GeneratedOntolex/PLI_1906_20190109%20sample.xml:1588 - Using an attribute 'expand' without a namespace is forbidden.
rapper: Error - URI file:///home/jmccrae/projects/elexis/tei2ontolex/GeneratedOntolex/PLI_1906_20190109%20sample.xml:1588 - Using an attribute 'rend' without a namespace is forbidden.
rapper: Failed to parse file GeneratedOntolex/PLI_1906_20190109 sample.xml rdfxml content
rapper: Parsing returned 721 triples

[laurentromary]

That comes from a case which I had left aside: proverbs. Can you tell we how you would use http://www.lexinfo.net/ontology/3.0/lexinfo#proverb concretely. The example is:

<re type="prov">
   <form type="phrase"><orth>Petite pluie abat grand vent</orth></form>
   <sense>...</sense></re>

would something like:

<ontolex:LexicalEntry>
    <ontolex:canonicalForm>
        <rdf:Description>
            <ontolex:writtenRep xml:lang="fr">Petite pluie abat grand vent</ontolex:writtenRep>
        </rdf:Description>
    </ontolex:canonicalForm>
    <lexinfo:termType rdf:resource="http://www.lexinfo.net/ontology/3.0/lexinfo#proverb"/>
</ontolex:LexicalEntry>

be OK?

[jmccrae]

Yes, proverb is the object of the termType property.