Relationship of encodings, folder structure

pfefferniels commented 3 years ago

At the moment, the relationship of the different encodings to each other is (more or less) defined by the folder structure. It would be good to instead explicitly encode this information in the headers, e.g. in the case of an exemplary realization something like

<manifestation>
  <titleStmt>
    <title>Realization of Probstück 4 by XY</title>
  </titleStmt>
  <relationList>
    <relation rel="isRealizationOf" target="../original-text/4/score.xml">
  </relationList>
</manifestation>

Using the isPartOf/hasPart relation we could relate a global oberclasse.xml (only consisting of a header) to the single parts of it (Probstuecke).

To encode relations between the textes (isRevisionOf/hasRevision for first and second edition, isTranslationOf for the translations) we would need to find a parallel way in TEI ...

Once all the relations are defined, the folder structure and filenames don't matter as much anymore, we could re-arrange it into a more intuitive structure, e.g.

encodings 
  + guidelines
  + indices
  + original-text
    + 1
      score.xml
      text-1st.xml
      text-2nd.xml
  + translations
    + en
      probstueck1.xml
      probstueck2.xml
  - lessons
  - recordings

Any thoughts, @rettinghaus ?

rettinghaus commented 3 years ago

This sounds very good and will be a big step into the right direction if one doesn't want to rely upon an underlying database. But, as you pointed out, getting the same in TEI is a bit tricky and I don't have a good idea right away. Maybe @peterstadler could be of any help?

peterstadler commented 3 years ago

I don't know of any canonical way to do this since TEI does not employ these FRBR relationships directly. But maybe the (new) <standOff> comes in handy with some nested <link> elements?

pfefferniels commented 3 years ago

Very interesting, thanks, @peterstadler ! It seems that <standOff> containers are defined in the <body> instead of <teiHead>, which feels a bit counterintuitive to me …

This suggests <relatedItem>, which also might be an alternative worth considering.

pfefferniels commented 9 months ago

So here are my suggestions: We introduce one single MEI file describing abstract works, something like:

<?xml version="1.0" encoding="UTF-8"?>
<mei xmlns="http://www.music-encoding.org/ns/mei" xml:id="works">
    <meiHead>
        <workList>
            <work xml:id="work001">
                <title>Probstück 1</title>
                <key mode="minor" pname="d" />
                <incip>
                    <incipCode>@clef:F-4\n@keysig:bB\n@timesig:C\n@data:,{8D'8Dx6C6D8E}{8F8D8,A8G}{6F6G8A6F6G8A}{8D,8A8F8D}</incipCode>
                </incip>
            </work>
            <work xml:id="work002">
                <title>Probstück 2</title>
                <key mode="minor" pname="g" />
                <incip>
                    <incipCode>@clef:F-4\n@keysig:bB\n@timesig:C\n@data:3-{',3D3Ebx3F3G3Ab3B3G}{6Ax6F6D6A}b8B8-x8F8-/3-{3G3A3B'3C3Db3E3C6}{D'6,B6G'6D}b8E8-,8B-/</incipCode>
                </incip>
            </work>
        </workList>
    </meiHead>
</mei>

(we could easily introduce some vertical hasPart/isPartOf relations here, such as Probstück 1 → is part of → Oberclasse → is part of → Große GB-Schule, but for now that should be fine).

The score of each Probstück becomes part of the TEI encoding, just like any other figure containing notated music. So each TEI encoding will start with something like this:

      <div n="0" type="chapter">
        <pb n="301" facs="#facs-p385"/>
        <head>Erſtes Prob-Stueck.</head><lb/>

        <figure type="notatedMusic">
          <notatedMusic xml:id="score">
            <ptr target="score.xml"/>
          </notatedMusic>
        </figure>

        <pb n="305" facs="#facs-p389"/>
        <div n="1">
          <head>Erlaͤuterung.</head><lb/>
          ...

Now this TEI encoding can be considered to be a realisation of the abstract work ("Probstück 1") by using <standOff>:

<TEI xmlns="http://www.tei-c.org/ns/1.0" xml:id="expression002">
  <!-- header -->
  <standOff>
    <listRelation>
      <relation name="realises" active="#expression002" passive="#work001" />
    </listRelation>
  </standOff>
  <!-- ... ->
</TEI>

In order to give more information about what this expression actually represents – edition, arrangement, translation or whatsoever, we could use <profileDesc>, e.g. in case of an English translation:

      <profileDesc>
        <langUsage>
          <language ident="eng">English</language>
        </langUsage>
        <textDesc>
          <derivation type="translation" />
        </textDesc>
      </profileDesc>

Any opinions on this?

rettinghaus commented 9 months ago

Let's ask @torstenroeder what he thinks.

peterstadler commented 8 months ago

Without knowing the details (of sources etc) here I have no objections regarding the encoding. The only thing that I want to comment on (or discuss) is whether it might be better(?) to move the definitions of relations into the MEI file within the work sections. So, rather then pointing from the expression to the work, reverse the direction and point from the work to the expression. That has the advantage (for me) to employ the established FRBR relations within MEI rather than adding these to TEI.

rettinghaus commented 8 months ago

Having a workList with all the incipits encoded is great also for the content overview. Let's start with this and see what we can improve later.

pfefferniels / probstuecke-data

Relationship of encodings, folder structure #20