Columns in a dataset have description fields in Tamr. These descriptions are used to improve the results of schema mapping predictions. Certain types of data sources have attribute descriptions, which can be added as the dataset is created in Tamr. (For information on importing datasets, also see our public docs on uploading a dataset to a project).
Descriptions can also be added to datasets already in Tamr in two ways, through the UI and through the API.
In the UI, go to the Schema Mapping page of a project using the dataset. The left panel shows input attributes, while the right shows unified attributes. Descriptions can be added to either type by expanding information on the attribute and clicking the "Edit Description" text.
Descriptions can also be added through API endpoint for updating an attribute, /datasets//attributes/. For information on using the API, visit the dedicated section on the Tamr Core API in our public docs.
To use the endpoint, you will need the dataset id and attribute id, both of which can be found using the internal API GET endpoint
<name> is the name of your dataset. Note that the internal API is subject to change.
Missing description fields will weaken the algorithm's confidence when suggesting a mapping for that field. The algorithm will still use the name and data, but with the description missing, the recommendation confidence will be lower. In general, it is better to always try to have a description in order to achieve higher confidence recommendation algorithm suggestions, though it is not completely necessary.
Updated about 1 year ago