About the Talend Trust Score™ with Snowflake

1. When you add a dataset from a Snowflake connection, a copy of the DQ Java libraries and semantic dictionary is sent to Snowflake to perform the validity checks.
   Verify that the pushdown parameter is in the JDBC URL of the Snowflake connection. For example:

   jdbc:snowflake://account.snowflakecomputing.com/?db=MY_DB&schema=PUBLIC&warehouse=MY_WAREHOUSE_WH&runProfile=sql 

   For more information, see Adding the pushdown parameter to a Snowflake connection.
   Information noteTip: You can use the crawler to retrieve multiple tables and views. Using the Talend APIs, you can also automate the quality compute of . For more information, see Scheduling a crawler run.
2. The DQ Java libraries are defined as Java UDFs.

   If you apply data quality rules to a dataset from Talend Cloud Data Inventory, they are also natively computed in Snowflake using the UDFs.

3. To calculate the Talend Trust Score™ in Snowflake, the following steps occur:

   1. The semantic discovery defines the nature and the format of the data. The semantic type of each column of the dataset is checked by analyzing a sample of up to 10,000 rows. By default, the sample contains the first rows, called the Head sample. The rows can also be chosen randomly, called the Random sample.

   2. The data validity and completeness: the records are checked against the semantic types to determine whether the fields are valid or invalid. If the fields do not match a semantic type, they are checked against the native types.

      Using the JDBC URL from Talend Cloud Data Inventory, the validity and completeness are computed from the entire table in Snowflake.

      From the dataset overview in Talend Cloud Data Inventory, you can preview a sample of the Snowflake table and retrieve valid and invalid records. This sample contains up to 10,000 records.

      The data quality bars in the column and dataset headers represent the quality of the entire table.

   3. Talend Trust Score™: the validity and completeness are calculated for the entire dataset in Snowflake. You can find the history in Talend Cloud Data Inventory, as well as the Talend Trust Score™ evolution from the dataset overview.

      You can also retrieve the Talend Trust Score™ from the dataset list and the data console.

4. The sample is sent to Talend Cloud Data Inventory and the Talend Trust Score™ of the entire dataset is calculated as follows:
   • The validity and completeness are natively computed in Snowflake on the entire table. The data quality rules compliance is also performed on the entire table.
   • The popularity, discoverability, and usage are calculated in Talend Cloud Data Inventory. For more information on each axis, see Checking the Talend Trust Score™.

You now have the Talend Trust Score™ with five axes for your dataset.