Lens.org Patents

This section describes the ingestion process and the mapping implemented to integrate patent records and their relationships from Lens.org into the OpenAIRE Graph.

Ingestion Process

The ingestion of patent records from Lens.org follows a three-step pipeline:

1. Import and Mapping: Patent records are imported from Lens.org and mapped into the OpenAIRE internal representation (Publication entities with instnacetype Patent) along with Citations to other records (referred to by DOI or other PIDs).
2. Processing and Mention Identification: The pipeline processes publication texts to identify mentions and citations pointing to patent records.
3. Filtering: A post-processing filter is applied. Because the graph contains a vast number of patents, the pipeline only retains patent records that are linked to at least one non-patent publication (via IsRelatedTo or Cites relationships). Patents that are not put into context by such references are discarded.

Entity Mapping

The table below describes the mapping from the Lens.org JSON payload format to the OpenAIRE Graph Publication format.

OpenAIRE Research Product field path	Lens.org JSON Field / Source	Notes
Publication Mapping
id	lens_id	Mapped as lens________::md5(lens_id) because Lens.org is a PID authority for the lens PID type.
pid	lens_id, alias_lens	Structured properties with class ID/classname lens and scheme dnet:pid_types.
originalId	lens_id, alias_lens, doc_key, doc_number	A list of all unique identifiers associated with the patent.
resourcetype	(fixed value)	Set to publication.
collectedfrom / hostedby	(fixed value)	Set to the Lens datasource identifier: fairsharing_::3b2d8f129ae2f408f2153cd9ce663043 ("Lens").
dateofcollection	(generated)	Current ingestion timestamp.
maintitle	biblio.invention_title	Filters for English titles (lang == "en") to populate title (qualifier: main title).
author	inventors	Inventor names extracted and mapped to author structure (assigning sequential rank starting from 1).
dateofacceptance	date_published	Mapped as the publication date of the patent.
description	desription.text	Mapped to description. (Note: source key has typo desription in mapping code).
subject	kind	Mapped to subject with value kind, scheme/class patent kind, and vocabulary dnet:subject_typologies.
Instance Mapping
instance.pid	pid	Matches the record's main pid (qualifier lens).
instance.url	lens_id	Formatted as https://www.lens.org/lens/patent/{lens_id}.
instance.dateofacceptance	date_published	Normalized date of publication.
instance.instancetype	(fixed value)	Patent as defined in the dnet:publication_resoruce vocabulary.
instance.instanceTypeMapping	publication_type	Captures original type mapping.
instance.alternateIdentifier	doc_key, doc_number	Alternate identifiers that do not have the qualifier lens.
instance.accessright	(fixed value)	Hardcoded to unknown.
instance.collectedfrom / hostedby	(fixed value)	Set to the Lens datasource identifier.

Relationship Mapping

Relations are extracted from literature references cited by the patents:

OpenAIRE Relation Semantic	Source / Target Type	Source Field / Logic
Cites	Patent $\rightarrow$ Publication	Extracted from references_cited.citations. Targets must be DOIs or PMC (PubMed Central) identifiers. Mapped with relation class Cites and sub-relation type citation.