The NZRIS data warehouse has been constructed to link entities by a primary key, which must be unique. However, the primary key is not explicitly labelled in the NZRIS Data Specifications. Also, there may be legitimate situations where the primary key is not unique, for example when a funding organisation makes one payment to a research organisation for a number of awards (Resource Distributed) – the amount for each contract may be different, the award IDs are different but the primary key for the entity (Local Resource Distributed ID) would be the same for this set of awards.
This issue is to find out whether the NZRIS system should:
retain the primary key model, explicitly label these in the Data Specifications and provide rules to standardise responses in cases where IDs are not unique, such as the situation referred to above, or
move the data warehouse to use a composite key model – where the values in any column may not be unique, but all values taken together in a row are unique.
The NZRIS data warehouse has been constructed to link entities by a primary key, which must be unique. However, the primary key is not explicitly labelled in the NZRIS Data Specifications. Also, there may be legitimate situations where the primary key is not unique, for example when a funding organisation makes one payment to a research organisation for a number of awards (Resource Distributed) – the amount for each contract may be different, the award IDs are different but the primary key for the entity (Local Resource Distributed ID) would be the same for this set of awards.
This issue is to find out whether the NZRIS system should: