Ensuring users of data have confidence in the supply and maintenance of the data is a critical element that affects the value of the data.
Once datasets are released data publishers should consider open data maturity of the dataset. This can be done by releasing new open formats for users. A review of the delivery method should also be considered if automation could improve the usability and reduce the manual resources required to maintain the data. The data quality should be considered, including improvements to the data collection and timeliness of the release of the data.
Datasets must be refreshed according to their update frequency. This ensures user confidence and reliability in our data. If the Dataset is manually maintained, processes to update the data should be in place. The person named as the data maintainer in CanWIN is the person who has responsibility for performing the refresh.
If a new collection of data is to be published for an existing dataset, it is recommended that
- additional data files be added to the existing dataset, (e.g.) new figures for 2017 are added to a time series data file
- a new file covering the complete refreshed dataset from inception to current can be published. e.g. Extract containing 2008-2012 data can be replaced with a new file
containing 2008-2017 data.
This is not an amendment but additional refreshed data.
Once a dataset has been published, agencies need to be mindful that members of the public may be using the dataset, so the impact of any changes needs to be considered. Once a dataset is published it cannot be removed. There is significant value to accessing historical data even if it is no longer maintained. Amendments to datasets will be treated as below:
- Minor changes occur to data content if errors are fixed or metadata has changed. Data can be updated anytime if it improves the users ability to understand the data
- Where a dataset has major structural changes, the old dataset needs to remain but can be marked as inactive. The new dataset can be added as a new dataset record.
Dataset updates can be emailed to email@example.com or uploaded by the metadata maintainer.