RealTime Search API Early Access
This reference guide provides an early access preview of the upcoming release of the RealTime search APIs. All content in this topic is subject to change.
Contact your Tamr account manager for access information.
Tamr RealTime’s search API enables you to quickly search for a matching entity before you create a new record in an operational system.
Tamr RealTime adds an additional datastore to your Tamr Cloud tenant: the RealTime datastore. This datastore acts as a system of reference; it is the authoritative source for your mastered entity data, including the current and historical Tamr IDs and source record IDs associated with each entity.
In a standard Tamr Cloud use case, you add source datasets to Tamr Cloud, and then add these sources to specific data products for deduplication, cleaning, enrichment, and consolidation. The resulting mastered entities and clustered source records are stored in Tamr Cloud’s working datastore. When you review, curate, and export data product data in Tamr Cloud, you are working with datasets stored in the working datastore.
With Tamr RealTime, once you are satisfied with the data in your working datastore, you can run a job to batch publish the mastered entity data to the RealTime datastore. This data is immediately available for search.
Tamr RealTime’s search capability enables you to programmatically search mastered entities stored in the RealTime datastore, by:
- Retrieving a record by a current or past Tamr ID, Tamr source record ID, or other persistent identifier. You can search for previous, out-of-date, and deleted IDs.
- Searching for a record by exact attribute values (case-sensitive).
- Searching for a record with similar attribute values. Records are retrieved using probabilistic matching, ranked by relevance.
The following diagram illustrates how the data in the RealTime datastore is updated through a data product job, and how data is retrieved from the RealTime datastore via the Tamr RealTime search API endpoints.
Tamr RealTime search endpoints include:
- Records: Retrieve a record by a specified ID, or search for a record by exact attribute values, specified as key/value pairs.
- Search: Search for records with similar attribute values, specified as key/value pairs.