OpenGraph

The OpenGraph additions to the <meta> header of webpages is based on Microsofts hCard. It seems to be widely used (by 50% of the pages of our test-set), probably because it gets automatically added by various content creation platforms.

There are a few extensions to OpenGraph, but we currently only support structures which are described on https://ogp.me. This looks like the best description. We do not actively monitor discussion about OpenGraph: please informm us when there are changes in the standard.

Writing Tasks

  1. Configure filters
  2. Composing filters
  3. Extracting knowlegde:
    URL normalization
    classic <meta>
    <meta> names
    		 all <meta>
    references
    <link>
    		 OpenGraph
  4. Packaging Results

Output rules

  • We do handle non-standard prefixes, when they are correctly declared via prefix attributes in the HTML anywhere. (But probably no-one else does handle this correctly).
  • When a property MAY have attributes (like og:image) it is called a Structure. We will always return it as a Hash (associative array) even when there are not attributes specified.
  • When a property MAY repeat (be an array, like og:image) it will always either be missing from the returned data-set or an Array. Even when only found once, it will be returned as Array of length one.

Currently not supported

Extensions on OpenGraph may be included, when used by multiple parties. We currently know about the following proposals:

[TODO: include examples for each of the og.type variations. In JSON format]