As a curator, you can manage schema mapping, mastering, golden records, and categorization projects. You provide the initial data expertise to validate and train Tamr models for these projects.
In the schema mapping workflow, you design the unified schema by mapping the attributes, or columns, in input datasets to unified attributes, which trains the Tamr model for recommending subsequent mappings.
In mastering projects, you are responsible for:
- Defining the blocking model to optimize the number of pairwise comparisons Tamr makes to identify matching and non-matching records.
- Labelling an initial set of record pairs as match or no-match to train the Tamr model.
- Assigning reviewers to evaluate Tamr’s suggestions for matching and non-matching record pairs. You then verify reviewers’ responses, and apply the feedback to train the Tamr model further.
- Reviewing Tamr’s suggestions for records that belong to the same cluster.
In golden records projects, you are responsible for configuring the rules that consolidate a cluster of records from a mastering project into a single golden record with data values that best represent the single entity represented by that cluster.
In a categorization project, you are responsible for:
- Uploading and, if necessary, editing the taxonomy used to categorize records.
- Categorizing a subset of records to train the Tamr model.
- Assigning reviewers to categorize records, verifying reviewers’ responses, and applying the feedback to train the Tamr model.
You can also define transformations in schema mapping, mastering, and categorization projects to modify the unified dataset. For example, you can concatenate first and last name fields, remove whitespace, or apply consistent date formatting.
Updated 25 days ago