Open dietervu opened 3 years ago
After discussion with Twan, can you confirm this (https://github.com/clarin-eric/VLO/issues/209) is the VLO functionality you are referring to?
yes!
References:
First proposal for embedded schema.org metadata on collection detail pages:
<script type="application/ld+json">
/*<![CDATA[*/
{
"url": "http://localhost:8080/service/virtualcollections/1079",
"name": "test query",
"description": "test collection; test collection; test collection; test collection",
"identifier": [
"https://doi.org/10.17907/test1079",
"dummy:identifier-1079"
],
"includedInDataCatalog": {
"url": "http://localhost:8080",
"@type": "DataCatalog"
},
"creator": [
{
"name": "admin1, ",
"@type": "Person"
}
],
"hasPart": [
{
"url": "http://www.clarin.eu/",
"name": "CLARIN website",
"description": "CLARIN main website",
"@type": "CreativeWork"
},
{
"url": "http://www.google.nl/",
"name": "Google search engine",
"description": "Google search engine main page",
"@type": "CreativeWork"
},
{
"url": "http://www.bing.nl/",
"name": "Bing search engine",
"description": "Microsoft bing search engine.",
"@type": "CreativeWork"
}
],
"@context": "https://schema.org",
"@type": "DataSet"
}/*]]>*/
</script>
Question(s):
hasPart
, any better suggestion?Would be good to include a publication date (at very least year), as to enable citation managers to create (creator, year) strings.
Updated with dateCreated
, dateModified
and datePublished
:
<script type="application/ld+json">
--
| /*<![CDATA[*/
| {
| "url": "http://localhost:8080/service/virtualcollections/1079",
| "name": "test query",
| "description": "test collection; test collection; test collection; test collection",
| "identifier": [
| "https://doi.org/10.17907/test1079",
| "dummy:identifier-1079"
| ],
| "includedInDataCatalog": {
| "url": "http://localhost:8080",
| "@type": "DataCatalog"
| },
| "creator": [
| {
| "name": "admin1, ",
| "@type": "Person"
| }
| ],
| "hasPart": [
| {
| "url": "http://www.clarin.eu/",
| "name": "CLARIN website",
| "description": "CLARIN main website",
| "@type": "CreativeWork"
| },
| {
| "url": "http://www.google.nl/",
| "name": "Google search engine",
| "description": "Google search engine main page",
| "@type": "CreativeWork"
| },
| {
| "url": "http://www.bing.nl/",
| "name": "Bing search engine",
| "description": "Microsoft bing search engine.",
| "@type": "CreativeWork"
| }
| ],
| "dateCreated": "2021-06-18T02:00:00.000+02:00",
| "dateModified": "2021-06-28T12:39:14.000+02:00",
| "datePublished": "2021-06-26T12:39:14.000+02:00",
| "@context": "https://schema.org",
| "@type": "DataSet"
| }/*]]>*/
Not sure if this is important to consider, but Google doesn't seem to know/accept {"@type": "CreativeWork"}
in a hasPart
property for ingesting data sets - it can only be one of a few specific subtypes. Unfortunately this information is not always available. Not sure if we can solve this in a proper and/or easy way but it's probably good to keep it in mind.
Edit: see spec
Actually, I think hasPart
may technically not be the right property to use except for resources in the collection that can in fact somehow be considered data sets - in which case they should be typed as such. I guess we cannot know to which resources this applies unless we explicitly ask the user. But there are not properties available that are a better fit as far as I can tell.
e.g. Zotero now recognizes the VCs with DOIs but not yet the ones with only a handle. Include some simple metadata headers (see VLO for an example) so that reference mangers can automatically detect the author, title, and URL