There is no way to convey that the Microdata should be ignored.
In the ideal case, you would give the Microdata and the JSON-LD items that are about the same thing the same URI (itemid
in Microdata, @id
in JSON-LD).
<div itemscope itemtype="http://schema.org/Article" itemid="#the-article">
</div>
<script type="application/ld+json">
{
"@context": "http://schema.org",
"@type": "Article",
"@id": "#the-article"
}
</script>
That way, supporting consumers can learn that these items describe the same thing, i.e., there are not two articles, only one, and properties added to one item are also relevant for the other item.
If that’s not possible, you could try to "destroy" the Microdata without making the document invalid. You could do this with a script, after each release of a new version of the theme. Simply remove every itemtype
attribute. Your document will still keep the Microdata, but it’s no longer using a vocabulary, so the structured data will likely not be re-used.