[SCHEMA] Temporal Relationships

[adambuttrick]

Version
v2

Describe the problem you would like to solve
Currently, ROR only captures the current state of relationships between organizations. This limitation impacts the ability to accurately track research outputs over time, especially when parent relationships change. For a concrete example, the research unit Laboratoire Magma et Volcans was a child organization of Université Jean Monnet - https://ror.org/04yznqr36 from 2005 until 2021. When services like OpenAlex use relationship data in ROR, however, they only have access to the current state of relationships. This means that when this relationship is removed, research outputs from the period when Laboratoire Magma et Volcans was under https://ror.org/04yznqr36 will not be correctly associated with https://ror.org/04yznqr36 in OpenAlex, as they only associate (or plan to associate) with the current parents.

Describe the schema change that you would like in order to solve the problem
We could expand relationships object in the schema to allows for the representation of time-bound relationships. This could include a temporal relationships array within each organization's record. Each entry would have the existing relationship values and a start date and end date for the relationship, e.g.

"relationships": [
    {
      "label": "Université Jean Monnet",
      "type": "parent",
      "id": "https://ror.org/04yznqr36",
      "start": "2005-01-01",
      "end": "2021-01-01"
    }
]

This change would allow us to maintain the current representation of active relationships, while also providing a way to capture and reference relationships that were valid during specific time periods.

Who would benefit from this change?

• Institutions: Would be able to correctly associate research outputs with the organizational structures that existed at the time of publication.
• Services like OpenAlex: Would have access to more accurate historical affiliation data for publications, improving bibliometric analyses.

Additional information
RNSR already captures this information, indicating a clear use case. Whether it is available in other sources is unknown. We should assess relative to our ability curate more generally vs. relative to a single source.