Schema Based DPO Pipeline

To utilize the Schema-Based DPO Pipeline, navigate to the DPO Pipeline and select Schema Based. It currently supports CSV files as reference.

Once you choose the pipeline, select "Add Preset" to start.

Fill in the preset name and prepare your CSV file (along with a header and sample values).

After uploading the CSV, you'll see a list of columns extracted from the file. You can then assign various tasks such as:

1. Filter Display Name: Change the display name of the filter if necessary.

2. Column Mapping: Determines if a CSV column will serve as a filter column, a field of interest (that will be searched), both a filter and a field of interest, or none.

3. Data Types

   1. String: Represents text data.

   2. Number: Represents numerical data, including integers and floats.

   3. Date: Represents date values.

   4. Time: Represents time values.

   5. Boolean: Represents true or false values.

4. Sub type: Determine the subtype to check whether the value is single-select or multi-select.

Once you are confident with the configuration, select the save button in the upper right to proceed with knowledge base creation.

Refer here for a guide on creating a knowledge base.

Refer here for a guide on using chat filter