gmd:purpose to dct:description

[laers]

In the Danish geodata catalog some data providers use the gmd:purpose to describe data.
I miss a transformation of gmd:purpose to GeoDCAT-AP and I have made this suggestion.

dct:description is 1..*

In the current xslt gmd:abstract is transformed to dct:description. The same can be done for gmd:purpose.
I have added this into the iso-19139-to-dcat-ap.xsl

<xsl:param name="ResourcePurpose">
  <xsl:for-each select="gmd:identificationInfo[1]/*/gmd:purpose">
    <dct:description xml:lang="{$MetadataLanguage}">
      <xsl:value-of select="normalize-space(gco:CharacterString)"/>
    </dct:description>
    <xsl:call-template name="LocalisedString">
      <xsl:with-param name="term">dct:description</xsl:with-param>
    </xsl:call-template>
  </xsl:for-each>
</xsl:param>

and

<xsl:copy-of select="$ResourcePurpose"/>

This results in both gmd:abstract and gmd:purpose is transform to dct:description:

[jakubklimek]

The proposed approach has one downside - it is a lossy mapping, meaning that based on the GeoDCAT-AP metadata, it is impossible to distinguish, which of the descriptions was the abstract and which was the purpose.

Since we have seen indications of stakeholders using or planning to use GeoDCAT-AP as the primary representation of geospatial metadata, generating INSPIRE metadata for compliance purposes (which is the reverse direction to the XSLT), it could be beneficial not to introduce new lossy mappings.

The solution could be:

1. introduce a specific property for Resource Purpose, a subproperty of dct:description to keep the distinction from the resource abstract clear, and maybe consider the same for the abstract to have consistent approach
2. If explicit reuse of dct:description interpretable without reasoning is required, then the purpose text can be transformed there as well (introducing some duplicity).

[laers]

From my knowledge of what kind of textual information the data providers has written in gmd:purpose I will be in favor of option 2. The two descriptions (gmd:abstract and gmd:purpose) can be seen as a combined and complete description of the data set.

[jakubklimek]

Well, the idea was to do both options, 1 and optionally 2. Because once you have 2 dct:descriptions, there is no way to tell which is the abstract and which is the purpose, if there was need to distinguish those. And if there is no such need, then the question is why the INSPIRE metadata contains both.