16 KiB

Raw Blame History

Datacite

This section describes the aggregation workflow used to gather the bibliographic material from Datacite and the relative mapping.

Datacite datasource

Datacite is a leading global non-profit organisation that provides persistent identifiers (DOIs) for research data and other research outputs.

Datacite API

The DataCite REST API allows users to retrieve, query, and browse Datacite metadata records. In particular, it exposes a method for harvesting new records incrementally.

https://api.datacite.org/dois?page[cursor]=$CURSOR&page[size]=$NUMBER_OF_ITEM_PER_PAGE&query=updated:[$FROM_DATE_TIMESAMP TO $TO_DATE_TIMESAMP]

On this API Request, we introduce some variables:

CURSOR: The value of the cursor to iterate the pages; the cursor is extracted from each API response and used in the next request.
NUMBER_OF_ITEM_PER_PAGE: (max 1000) defines how many records must be returned within each API response.
FROM_DATE_TIMESAMP, TO_DATE_TIMESAMP interval timestamp of the updated record.

Each record contains two pieces of information needed for incremental harvesting:

isActive: tells if the record is deleted (isActive:false)
updated: timestamp of last update

Collection Workflow

The collection workflow is responsible for aggregating new records. Each record is stored locally on a table with the following schema:

DOI: The DOI of the Datacite record (it is the primary key)
update_timestamp: the last update date timestamp
json: the native record JSON

The metadata collection process identifies the most recent record date available locally and uses such date to requests the records to the Datacite API, populating the FROM_DATE_TIMESAMP variable. The records in the API response are included in the local storage in upsert mode.

Datacite Mapping

Entity Mapping

The table below describes the mapping from the XML baseline records to the OpenAIRE Graph dump format.

OpenAIRE Research Product field path	Datacite record JSON path	# Notes
`id`	`\attributes\doi`	id in the form `doi_________::md5(doi)`
`instance` `instance.type`	`\attributes\types\resourceType` `\attributes\types\resourceTypeGeneral` `attributes\types\schemaOrg`	Use the vocabulary *dnet:publication_resource* to find a synonym to one of these terms and get the `instance.type`.
`type`	`\attributes\types\resourceType` `\attributes\types\resourceTypeGeneral` `attributes\types\schemaOrg`	Using the *dnet:result_typologies* vocabulary, we look up the `instance.type` synonym to generate one of the following main entities: `publication` `dataset` `software` `otherresearchproduct`
`pid`	`\attributes\doi`	`scheme = doi`
`originalid`	`\attributes\doi`
`dateofcollection`	`attributes\updated`	the timestamp is defined in milliseconds we convert to "yyyy-MM-dd'T'HH:mm:ssZ" format
`author`	`\attributes\creators`	Each creator field will be mapped in the author entity below the subfield. If the record has no Creator it will be skipped
`author.fullname`	`\attributes\creators\name`	if name is not defined, we construct from given and family name
`author.rank`		Incremental index starting from 1
`author.name`	`\attributes\creators\givenName`
`author.surname`	`\attributes\creators\familyName`
`author.pid`	`\attributes\creators\nameIdentifiers`	this is a list of pids associated to the creator
`author.pid.scheme`	`\attributes\creators\nameIdentifiers`	mapping with vocabulary dnet:pid_types
`author.pid.value`	`\attributes\creators\nameIdentifiers/nameIdentifier`	the pid value
`maintitle`	`\attributes\titles`	Titles whose title type is null or title type is Main
`subtitle`	`\attributes\titles`	Titles whose title type is Subtitle since the title type vocabulary in OpenAIRE use the datacite title type vocabulary
date section		for each date in particular for DOI starting with 10.14457 we Apply a fix thai date convert a date to ThaiBuddhistDate and reformat to local one see ticket #6791
`publicationdate`	`\attributes\dates`	where `dateType` is issued
`publicationdate`	`\attributes\publicationYear`	we create this date format `01-01-publicationYear`
`embargoenddate`	`\attributes\dates`	where `dateType` is available
`subjects`	`\attributes\subject`	`scheme=keywords`
`description`	`\attributes\descriptions`
`publisher`	`\attributes\publisher`
`language`	`\attributes\language`	cleaned by using vocabulary `dnet:languages`
`publisher`	`\attributes\publisher`
`instance.license`	`\attributes\rightsList`	if the rights value starts with http and matches a particular regex
`instance.accessright`	`\attributes\rightsList`	if not present :`unknown` if datasource is Figshare:`open` If `embargo_date < today()`: OPEN

Relation Mapping

OpenAIRE Relation Semantic and inverse	Datacite record JSON path	Source/Target type	#Notes
`isProducedBy/produces`	`attributes\fundingReferences`	`ResearchProduct/Project`	only when the fundingReferences matches the pattern `(info:eu-repo/grantagreement/ec/h2020/)(\d{6})(.*)`
`IsProvidedBy/provides`		`ResearchProduct/Datasource`	Datasource is always set to `Datacite`
`isHostedBy/host`	`\attributes\relationships\client\id`	`ResearchProduct/Datasource`	we defined a curated map clientId/Datasource if we found a match we create an hostedBy Relation
`isRelatedTo`	`\attribute\relatedIdentifiers`	`ResearchProduct/ResearchProduct`	we create relationships whenever the pid of the target is resolved on the Research Graph

16 KiB Raw Blame History