Attribute Recommendations

Schema Mapping using Tamr Machine Learning

Tamr can accelerate the schema mapping process by learning which source attributes are likely candidates to be mapped to unified attributes. The process for this is as follows:

1. Create unified attributes and partially populate them (done by you).
2. Update the unified dataset (done by Tamr).
3. Learn from mappings (done by Tamr).
4. Generate mapping suggestions (done by Tamr).
5. Review and accept, or reject suggestions (done by you).
6. Repeat if necessary (done by you).

Steps 1 and 5 involve human feedback and input.
Steps 2-4 are where Tamr learns and generates suggested mappings.

You can use the sliding bar to choose how similar attributes for Tamr suggestions should be. You can also add new datasets to the project and repeat this process by running steps 2-4 again, and accepting or rejecting Tamr suggestions.