Datasets may need to be deaccessioned or deleted, depending on institutional policy, when:
- Errors are discovered in a published dataset that renders it unusable for research.
- If the content is unlawful or the dataset is fraudulent from a scientific point of view.
- If one or more of the authors or rights holders did not give permission for publication.
- If personal data appears to have been deposited without a legal ground such as permission of research subjects.
- If the content consists of personal data, and a research subject rightfully objects to the preservation of a digital object with an appeal to the GDPR (e.g. right to be forgotten; or revocation of informed consent).
- If there is a legally binding maximum preservation period for the content.
If there is sufficient ground to decide to deaccession a dataset, the deaccessioning is done by a Data Manager who will conduct an appraisal of the request.
In all cases where access to a published dataset is terminated, a notice will be added to the landing page of the Persistent Identifier (PID) associated with the dataset to indicate that the dataset is no longer available: this is called a 'tombstone page'. This is done to ensure that anyone that has citations to resources that have subsequently been removed are retained, and furthermore that the provenance of the deaccession or removal is recorded within all associated metadata.
Purpose, Context and Content
Within the SEADDA consortium, workflows for dealing with deaccessioning of datasets were examined from three long standing digital repositories for archaeological data: the Archaeology Data Station of Data Archiving and Networked Services (DANS), the Swedish National Data Service (SND), and the Archaeology Data Service (ADS).