Step 4: Notify validation request

Context

Pre-conditions: repository manager accepts notifications about mentions, using the CORE repository dashboard.

The trigger: A mention was identified with or without an url to a code repository. With this trigger the validation process will begin to get the author’s approval. The approval is about the connection between the article and identified software mentioned.

End result: article id, software url or name sent to OA.

Action: Integration of notifications with repository platforms (T6.3)

The notification sent to the repository will contain the following information about a paper that has been analysed as containing software mentions;

From the paper metadata:

  • paper_id
  • paper_title

From the Softcite metadata:

  • software_mention_context
  • software_mention_attributes
  • software_repository_link

CORE Dashboard - new notifications tab / module (T6.1)

For the management of notifications, CORE will develop a new module for its existing repository dashboard. The repository manager will need to enable notifications and provide the endpoint URL to which notifications should be sent.

OA MUST have credentials to connect to the dashboard and enable notifications.

CORE Dashboard Module

Figure 1: CORE Dashboard module

Once the software mentions are generated, there will be two modes of operation: - automatic, all the software mentions (over a certain confidence threshold) automatically detected will be sent to the authors for validation. - manual, the repository manager can review and select which mentions are valid or ask for which mentions to be sent to the authors for validation.

The mentions that require validation will be propagated to an endpoint hosted internally by the Institutional Repository and configured using the CORE Dashboard settings.

The endpoint will have a standard authentication system (to be agreed between CORE and the Institutional Repository) and the format will be compliant with the COAR-Notify protocol.

The payload of the message will be as follow:

{
 "@context": [
   "https://www.w3.org/ns/activitystreams",
   "https://purl.org/coar/notify"
 ],
 "actor": {
   "id": "mailto:library@repo.com",
   "name": "Repository manager",
   "type": "Person"
 },
 "id": "urn:uuid:0370c0fb-bb78-4a9b-87f5-bed307a509dd",
 "object": {
   "id": "https://research-organisation.org/repository/record/201203/421/", //identifier of paper
   "ietf:cite-as": "https://doi.org/10.5555/12345680", //PID of paper
   "sorg:citation":{
       "@context": "https://doi.org/10.5063/schema/codemeta-2.0",
       "type": "SoftwareSourceCode",
       "name": "SoFAIR",
       "referencePublication": "https://doi.org/10.1016/j.procs.2012.04.202"
   },
   "mentionConfidence":99.0,
   "mentionType":"used",
   "mentionContext": "In this paper, we present the software X vY (http://sw/link)"
 },
 "origin": {
   "id": "https://research-organisation.org/repository",
   "inbox": "https://research-organisation.org/inbox/",
   "type": "Service"
 },
 "target": {
   "id": "https://review-service.com/system", // CORE
   "inbox": "https://review-service.com/inbox/", // CORE edit/input form
   "type": "Service"
 },
 "type": [
   "Offer",
   "coar-notify:ReviewAction"
 ]
}

In particular: - actor: it's the entity that started the validation process, in this case the repository manager that requested the software mention validation. - id: a unique id identifying the activity - object: will contain the software mention validation request, in practice, it will include the link between the paper and the software mention. - id the identifier or the record in the repository for which the mention has been discovered - ietf:cite-as additional identifiers for the research papers, such as DOI, ARK or handle identifiers. - sorg:citation a description of the software using the codemeta standard - mentionConfidence the score for which the machine learning tool detected the citation. A higher score implies an higher degree of confidence that the mention is correctly identified. - mentionType the type of usage of the software detected in the paper. It could be, used, created, cited. - mentionContext the snippet of text where the mention was detected. - origin the details of the repository endpoint for which the mention has been directed - target the details of where the results of the activity should be completed. In practice this should include the details to where to send the validation results. Most likely a CORE endpoint that would support both the editing and the confirmation.