Search engines use the http://schema.org vocab to analyse content that they crawl. To make search engines understand iso19139 a mapping to schema.org should be made available. For those iso19139 aspects currently not available in schema.org, we can suggest an extension of schema.org.

This mapping can be implemented in GeoNetwork in 2 ways:

• create a schema.org schema-profile, which can be set up to store metadata using the schema.org model and transform documents to other schema's. The output format (xml,rdf,json) is arbitrary
  
• add a new schema.org output format (json-ld), which can be adopted by all schema-profiles (this is how the transformation to DCAT is currently implemented)

Option 1 may over time be the best option, however may have too much impact in the scope of the current testbed.

There is a number of ways to expose schema.org so it can be ingested by search engines.

• as RDFa/microdata, where the model is embedded in html
• as json-ld embedded in

If we are thinking on crawlers we will probably need some kind of list (rss?) for schema.org formatted data, right?

And link that list from the robots.txt

@delawen, You mean a startpoint of the crawl (bootstrap)? The geonetwork sitemap will do, make sure it supports pagination though

Please mind this initiative: http://www.w3.org/wiki/WebSchemas/Datasets and http://blog.schema.org/2012/07/describing-datasets-with-schemaorg.html to see if we can use or contribute to some of their work. Discussion moved to github, for example schemaorg/schemaorg#713, schemaorg/schemaorg#688, schemaorg/schemaorg#583, schemaorg/schemaorg#113

Links of interest:

http://schema.org/docs/full.html
http://schema.org/DataCatalog: not sure if useful to define the GeoNetwork instance or just a very big dataset.
http://schema.org/Dataset
http://schema.org/DataDownload
http://schema.org/Map