Project

General

Profile

D4.5 OpenAIRE publishing APIs: specification and release plan

Authors: Paolo Manghi, Alessia Bardi, Miriam Baglioni, Claudio Atzori -- ISTI-CNR

Overview of updates - July 2019

  • Added informative note

Overview of updates - February 2019

  • Updated status of the release plan

Overview of updates - December 2018

  • Added option of using Zenodo API in cases when 3rd party services want to delegate the preservation of payloads and metadata
  • Updated links to OpenAIRE guidelines for Literature repository managers, which have been released on November 30th, 2018.

Informative note

The current deliverable describes a design of publishing API for pushing metadata records into the OpenAIRE Research Graph.
A number of discussions followed with CERN and members of the research communities, which highlighted that the Zenodo Deposit API made by CERN was already offering a super set of the needed functionality and that the co-existence of an additional API could have led to duplication of effort, possible sustainability issues, and misunderstandings among the adopters.
For the above reasons, it was agreed to not proceed with the implementation of a new API and to push for the adoption of the already existing Zenodo API. CNR implementation effort has been therefore shifted for a more efficient and maintainable version of the so called "Direct indexing API" which are used by Zenodo to send the deposited records in real time into the OpenAIRE Research Graph. The Direct indexing API is not a public API, but they "close the circle" between Zenodo and the OpenAIRE Research Graph: all the products deposited in Zenodo are available in the OpenAIRE Beta and Production graphs in few seconds after deposition. In addition, if the deposition occurred in one of the Zenodo communities selected as relevant for a community, the relative metadata is also made available in the corresponding Research Community Dashboard.

The remainder of this wiki describes the design of the publishing API as initially conceived.

1. Introduction

This deliverable includes the specification of the software and a release plan for the OpenAIRE publishing APIs that support third-party services at publishing metadata about interlinked and packaged research products into the OpenAIRE Scholarly Communication Graph, in respect of the OpenAIRE interoperability guidelines (https://guidelines.openaire.eu).

The OpenAIRE publishing APIs supports the concept of “continuous publishing” in digital research settings where researchers conduct their activities in digital laboratories using ICT tools and services for processing and analysing research data. By using the OpenAIRE publishing APIs, a service/tool can automatically publish metadata on behalf of the researchers.
The service/tool and its underlying infrastructure is responsible for keeping persistent identifiers, preserving the payload of the objects and the metadata.
If the service does not want to keep the above responsibilities, then using the Zenodo API is preferred. By depositing an object into Zenodo, the object will get a DOI and its metadata and payload will be preserved according to the policies available at http://about.zenodo.org/policies/ . Metadata will also be automatically pushed into the OpenAIRE Scholarly Communication Graph.

When a metadata record is pushed into the OpenAIRE Scholarly Communication Graph, via the Zenodo API or the OpenAIRE Publishing API:

Researchers benefit from a service that uses the OpenAIRE publishing APIs in several ways:

  • The service will support the generation of metadata to improve the FAIRness of the relative research products;
  • Researchers are relieved of the burden of depositing the products they want to publish in a repository external to their digital laboratory;
  • Researchers can choose to publish research products at any step of their research activity.

Section 2 describes the architectural specification of the APIs.
Section 3 describes the formats of records, based on the OpenAIRE guidelines, that can be submitted via the API.
Section 4 presents the release plan of the APIs.

2. Architecture

The high-level architecture of the APIs and how they interact with the OpenAIRE infrastructure is summarised in figure 1.
The external service calls the APIs to push one metadata record into the OpenAIRE Information Graph. The API component will store the metadata in a dedicated metadata store and call the already existing Direct Indexing API in order to make the metadata available in the OpenAIRE portal and the Search API.
Records in the metadata store will be daily processed by the OpenAIRE Aggregation system to perform cleaning and harmonization to the OpenAIRE internal format, so that they can be de-duplicated and enriched thanks to the OpenAIRE mining algorithms.

High-level architecture of the OpenAIRE publishing APIs
Figure 1 High-level architecture of the OpenAIRE publishing APIs

Details on the API components are shown in figure 2. The APIs provide two main entry points:

  • The API key deliverer: a GUI for requesting the API key needed to call the publishing endpoint;
  • The publishing endpoint that clients, provided an API key, use to publish metadata into OpenAIRE.

Publishing APIs and dependencies with existing OpenAIRE components
Figure 2 High-level architecture of the OpenAIRE publishing APIs

Figure 3 illustrates the procedure to obtain an API key. The manager of the external service willing to use the OpenAIRE publishing endpoint goes to the Web GUI for token requests and fills in the request form. The form asks the service manager the following information:

  • Contact email (mandatory): it will be used by the OpenAIRE Technical team to communicate with the service manager
  • Service name (mandatory): it will be used to generate a new OpenAIRE Content Provider. This is needed to track provenance information of the metadata published via the APIs.
  • Organization (mandatory): the organization that is responsible for running the service. The service manager will be able to select one of the organization already in OpenAIRE or to add a new one. In case a new organization must be added, the service manager must provide the following information:
    • official/legal name (mandatory)
    • english name (optional)
    • jurisdiction/country (optional, EU will be used by default)
    • web site URL (mandatory)

The API key deliverer component interacts with the Content Provider Manager Service of OpenAIRE to create a new content provider and associate it to the selected (or just created) organisation. An API key associated to the content provider is generated and returned to the service manager.

Process to request an API key for the OpenAIRE publishing APIs
Figure 3 Process to request an API key for the OpenAIRE publishing APIs

Once the API key is obtained, the service manager can configure its service to perform calls to the publishing endpoint providing the API key in the HTTP header.

The publishing endpoint exposes methods to push metadata records in XML format, complaint to the OpenAIRE guidelines. If a record is not compliant to the guidelines, the request will be rejected with an HTTP 4xx Client Error status code.

The sequence diagram in figure 4 describes how the various OpenAIRE components collaborate in case of a successful push request.

The publishing endpoint receives a push request from an authorised service. The request is forwarded to the Publisher component, which validates the input XML record via the XML validator. If the validation succeeds, as in the case depicted in figure 4, the record is stored in the metadata store and transformed into the format expected by the OpenAIRE Direct Indexing API. The Direct Indexing API transforms again the record into a Solr document and sends it to the OpenAIRE Solr server serving the OpenAIRE production portal and the public search API.

Push request to the OpenAIRE publishing endpoint
Figure 4 Push request to the OpenAIRE publishing endpoint

2.1 Existing OpenAIRE services used by the publishing APIs

As described in the previous section, the OpenAIRE publishing APIs use existing OpenAIRE internal services to accomplish their task: the Content Provider Manager Service, the Direct Indexing API and the Metadata Store Service. In the following, a brief description of the mentioned services is provided. Note that all these services are not for public use and are accessible only by other OpenAIRE services.

Content Provider Manager Service
The Content Provider Manager Service is a service for the management of data sources providing content that is aggregated into the OpenAIRE Information Graph by the OpenAIRE Aggregation infrastructure.
The service supports typical CRUD (Create, Retrieve, Update, Delete) operations via a RESTful interface on the data sources integrated by the OpenAIRE infrastructure. The service is currently used by OpenAIRE Aggregation Infrastructure and the Content Provider Dashboard.

Direct Indexing API
The Direct Indexing API has been developed to enable selected components of the OpenAIRE infrastructure (namely Zenodo and the Claim functionality of the OpenAIRE portal) to push metadata records into the Solr index used by the OpenAIRE portal and Search API without waiting for the normal aggregation and content provision workflows to take place. The normal workflow execution that generates updated content for the Solr index takes 1 or 2 weeks and includes several steps of harmonization and enrichment, as explained in details in https://www.openaire.eu/aggregation-and-content-provision-workflows and http://doi.org/10.5281/zenodo.996006. Using the direct indexing API, the harmonization and enrichment steps are not executed, but end-users can view the metadata record on the OpenAIRE portal and the search API just some seconds after their action on Zenodo or on the portal. Records added via the Direct Indexing API that have not been harmonised and enriched are marked as “Records in preview” in the OpenAIRE portal.

Metadata Store
A metadata store (mdstore) is a storage unit capable of storing metadata objects in XML format. Metadata stores are managed by the D-Net MDStore Service and are typically fed via metadata aggregation workflows. The OpenAIRE publishing APIs will feed a dedicated mdstore with the validated XML records received from external services. Records will be subject to the OpenAIRE harmonization and enrichment processes like any record collected from “traditional” data sources.

3. Publishing research products into the OpenAIRE Scholarly Communication Graph

Requests to the OpenAIRE publishing API must include the XML metadata record to publish. Its schema varies based on the type of the described resource, according to the OpenAIRE guidelines (http://guidelines.openaire.eu).
The OpenAIRE guidelines are an integrated suite of guidelines for data sources, whose purpose is to set a common and interoperable approach for exporting metadata about research products.

In the following, the guidelines for each type of research product whose metadata can be published via OpenAIRE are briefly described and the following support material is linked to support adopters of the OpenAIRE publishing APIs:

  • Guideline documentation;
  • XML Schema used by the XML validator component;
  • Examples of valid metadata records.

3.1 Publishing literature products

Fast links:

Metadata about literature products (e.g. scientific articles, technical reports) can be pushed into the OpenAIRE Scholarly Communication Graph provided they are compliant to the application profile specified by the OpenAIRE guidelines for Literature Repository Managers v4 (https://guidelines.openaire.eu/en/latest/literature_v4/index.html).

The application profile is based on the Dublin Core and DataCite Metadata Schema v4 and COAR vocabularies for access rights (http://vocabularies.coar-repositories.org/documentation/access_rights/) and resource types (http://vocabularies.coar-repositories.org/documentation/resource_types/).

Table 1 lists the fields that are mandatory. Examples of complete and minimal metadata records compliant to the OpenAIRE guidelines for Literature Repository Managers v4 (https://guidelines.openaire.eu/en/latest/literature_v4/index.html) are available at https://github.com/openaire/guidelines-literature-repositories/tree/master/samples.
The XML schema adopted by the XML Validator component is available at https://www.openaire.eu/schema/repo-lit/4.0/. If the submitted record cannot be validated via the schema, the request will be rejected.

Table 1. Mandatory fields in the application profile for literature products

OpenAIRE field
Metadata Element
Note


Title
dc:title



Creator
datacite:creator
Possibly with identifier, like ORCID


FundingReference
datacite:fundingReference
Mandatory if the literature product is funded by a project


Embargo Period Date
datacite:date (with dateType="Available")
Mandatory if the embargo period hasn’t ended yet


Publisher
dc:publisher
Mandatory if available.


Publication date
dc:date



Publication type
dc:type
Values from the COAR vocabulary of resource types


Description
dc:description



Resource identifier
dc:identifier
In the form of PID or stable URL. The value will be used by OpenAIRE to uniquely identify the product.


Access rights
dc:rights
Values from the COAR vocabulary of access rights


Full-text location
aire:file
Specify where the full-text of the product can be directly downloaded

3.2 Publishing dataset products

Fast links:

The current OpenAIRE guidelines for Data Archive Managers 2.0 ( https://guidelines.openaire.eu/en/latest/data/index.html) are based on the DataCite Metadata Schema v3.1 by making some of the otherwise optional DataCite properties mandatory, as well as enforcing specific encoding schemes on the values of some DataCite properties. A new application profile based on DataCite Metadata Schema v4 is expected to be proposed by the OpenAIRE consortium in the context of the OpenAIRE-Advance project.

For this publishing APIs, it has been agreed not to adopt the current application profile for research data based on version 3.1 Datacite Metadata Schema, but to enhance it to adopt the new features of the Datacite Metadata Schema v4, in a similar way it has already been done for the new OpenAIRE guidelines for Literature Repository Managers v4. Namely, the new features that have been included to generate the application profile and the relative XML schema for the publishing APIs are:

Table 2 lists the fields that are mandatory. An example of complete XML record compliant to the application profile is available at http://data.d4science.org/workspace-explorer-app?folderId=SXpRSHJsclI0VmtSZysyM0hGYVdHUmRIYzQ2K1lYYm45UFJNZkQ2ejlDaFR2NVlORWRJRFIwZitkTUZnQWNVVg. The XML schema adopted by the XML Validator component is available at http://data.d4science.org/T25TTTRkZC9jeDRQc3NicTZ1SkpGdVl5YVlnbDRNWFZHbWJQNStIS0N6Yz0. If the submitted record cannot be validated via the schema, the request will be rejected.

Table 2. Mandatory fields in the application profile for dataset products

OpenAIRE field
Metadata Element
Note


Resource identifier
datacite:identifier
In the form of PID or stable URL. The value will be used by OpenAIRE to uniquely identify the product.


Creator
datacite:creator
Possibly with identifier, like ORCID


Title
datacite:title



Publisher
dc:publisher
Mandatory if available.


Publication year
datacite:publicationYear
In the form YYYY


Date
datacite:date
The type of date must be specified


Access rights
datacite:rights
Values from the COAR vocabulary of access rights


Description
datacite:description

3.3 Publishing software products

Fast links:

Metadata about research software can be pushed into the OpenAIRE Scholarly Communication Graph provided they are compliant to the OpenAIRE guidelines for Software Repository Managers (https://guidelines.openaire.eu/en/latest/software/index.html).
The guidelines are, at the time of writing this deliverable, still in a draft status and open for consultation and feedback from the software community. We invite developers of external services using the OpenAIRE publishing APIs to provide their feedback in order to improve the application profile and guidelines and contribute to best practices on research software citation.

The application profile is based on the DataCite metadata schema v4, the OpenMinTed SHARE-OMTD software guidelines, DOE CODE and Codemeta initiatives.

Table 3 summarises the mandatory field defined by the application profile.

Examples of XML records compliant to the application profile is available at http://data.d4science.org/workspace-explorer-app?folderId=WDVNdi82TFZnb2RoR2gyQlFiSitEVjVUcDNoeTIrWVg2VVVxVGllQkhjRVh0OFRtdkhXOTZnWTJXVjRpa3U4dQ.
The XML schema adopted by the XML Validator component is available at http://data.d4science.org/eVZlWGRmR2ErcEFQc3NicTZ1SkpGbVRqdXBscFlGTUpHbWJQNStIS0N6Yz0.
If the submitted record cannot be validated via the schema, the request will be rejected.

Table 3. Mandatory fields in the application profile for software products

OpenAIRE field
Metadata Element
Note


Resource identifier
datacite:identifier
In the form of PID or stable URL. The value will be used by OpenAIRE to uniquely identify the product.


Creator
datacite:creator
Possibly with identifier, like ORCID


Name
datacite:title



Software Type
datacite:resourceType
The attribute resourceTypeGeneral must be “Software”. The text value of the element can contain the specific type of software (e.g. Data mining, Statistical analysis)


Access rights
datacite:rights
Values from the COAR vocabulary of access rights

3.4 Publishing “other products”

Fast links:

Metadata about other research products can be pushed into the OpenAIRE Scholarly Communication Graph provided they are compliant to the OpenAIRE guidelines for Other Research Products (ORP) Repository Managers (http://guidelines-other-products.readthedocs.io/en/latest/index.html).
The guidelines are, at the time of writing this deliverable, still in a draft status and under review by partners of the OpenAIRE-Connect project. We invite developers of external services using the OpenAIRE publishing APIs to also provide their feedback in order to improve the application profile and guidelines and contribute to best practices on sharing, citation and re-use of research products that are not literature, data, nor software.

The application profile is based on the DataCite metadata schema v4, and the OpenMinTed SHARE-OMTD software guidelines.

Table 4 summarises the mandatory field defined by the application profile.
Examples of complete XML records compliant to the application profile are available at http://data.d4science.org/workspace-explorer-app?folderId=THlEcW1MZjFYOWU0SFlzSW5KZmtBRmhmMGhXbHFrT3VjUU9td3Y4aUhCOHp1MFQxNHE3SXNhVEo4QVd5czRHNw
The XML schema adopted by the XML Validator component is available at http://data.d4science.org/YStjcU9WL2l5V2tQc3NicTZ1SkpGdVFlWVgyTnA5ODBHbWJQNStIS0N6Yz0.
If the submitted record cannot be validated via the schema, the request will be rejected.

Table 4. Mandatory fields in the application profile for other research products

OpenAIRE field
Metadata Element
Note


Resource identifier
datacite:identifier
In the form of PID or stable URL. The value will be used by OpenAIRE to uniquely identify the product.


Creator
datacite:creator
Possibly with identifier, like ORCID


Name
datacite:title



Type
datacite:resourceType
The attribute resourceTypeGeneral must be one of the Datacite terms: “Other”, "Workflow", "Service", "Model". The text value of the element can contain the specific type (e.g. protocol, trial, web service)


Access rights
datacite:rights
Values from the COAR vocabulary of access rights

4. Release Plan

The release plan of the OpenAIRE publishing APIs is articulated in three steps:

1) MS18 - M16 (April 2018) : 1st release of the APIs as TRL6 service.

This initial version is planned to include:

  • The publishing endpoint, capable of validating, and storing input XML records
  • The XML validator for all types of entities based on the application profiles as described in this deliverable

The milestone is postponed to M18 (June 2018). The task has been re-assigned to CNR in due course because of a misunderstanding during the proposal submission. Initially assigned to the task was ARC, which, in the context of the task, adapted the OpenAIRE Search API to the new types of research products and to the model changes realised in OpenAIRE-Connect. Once realised that the goal of the task was different, CNR took over as better suited to perform it, being in charge of the back-ends.
This change of the plan devised in the DoA is not affecting subsequent milestones and releases of the APIs.

Status: API released on the development infrastructure.

2) MS19 - M23 (November 2018): 2nd release of the APIs as TRL6 service. This version is planned to include:

  • The publishing endpoint, capable of validating, storing and real-time indexing the input XML records
  • The XML validator for all types of entities based on the application profiles modified according to the updates of the OpenAIRE guidelines
  • API versioning to support the smooth transitions of clients to the new application profiles
  • Updates based on the feedback of the pilot/test services using version 1

Status: API available on the development infrastructure. Tests have been done internally by CNR. No third-party pilot/test services could be used as the only Research Initiative interested in the functionality is EPOS-IT, which plans to use the Zenodo deposition API for the publishing of research objects composed of EPOSAR experiments.

3) MS20 - M27 (March 2019): TRL8 release of the APIs. This version is planned to include:

  • The API key deliverer GUI
  • The XML validator for all types of entities based on the application profiles modified according to the updates of the OpenAIRE guidelines
  • API versioning to support the smooth transitions of clients to the new application profiles
  • Updates based on the feedback of the services using version 2

Status: The milestone is postponed to the end of the project, M30 (June 2019). The implementation of the API is being performed in the context of EOSC and the specific feature of authentication and authorization is crucial. The idea of implementing a dedicated API key deliverer has been "freezed" in favour of a centralised solution, more coherent with EOSC, to be realised with the support of ARC and GRNET (who supported the implementation of OpenAIRE AAI).