Create a Private Set Intersection task

更新时间:
复制 MD 格式

Private Set Intersection (PSI) finds the intersection of sample data from multiple parties. PSI uses secure computing technology to find the intersection of ID columns while the raw data remains on each party's local node. The output is stored on the node of each participant and is displayed as a virtual wide table in the multi-party secure modeling console.

Note

You must use PSI to find the intersection of ID columns, even if the data from all participants has already been sample-matched.

Prerequisites

Each participating organization has offline samples available.

Procedure

  1. Log on to the multi-party secure modeling console.

  2. In the navigation pane on the left, click Model Development > Private Set Intersection.

  3. On the Private Set Intersection page, click Private Set Intersection in the upper-right corner. Then, configure the form and the PSI task.

    Note

    If you are creating a PSI task for the first time, click Create in the center of the page to open the Create Private Set Intersection page.

    1. Configure the form information.

      Parameter

      Description

      Node

      The node where the raw data is located. The two data tables must be on different nodes.

      Name

      Select an active data table on the node.

      Association Key

      Select one or more association keys for the PSI task from the keys defined in the current data table.

    2. Configure the PSI task.

      Parameter

      Description

      Virtual wide table name

      The name of the virtual wide table that is generated after the PSI task is successfully created. The name must meet the following requirements:

      • It must start with a letter and can contain only letters, digits, and underscores (_).

      • It must be no more than 32 characters long.

      Output file path

      The physical path for the PSI output. You can enter a custom path or click System Generated to get an available path.

  4. Click OK to execute the PSI task.

    Processing the samples may take some time. A task with several hundred thousand samples typically takes a few minutes to complete.

  5. View the results.

    On the Private Set Intersection page, check the Running Status to the right of the virtual wide table. Alternatively, you can click Running Configuration to the right of the virtual wide table to view the task details on the PSI details page.

    Note
    • If the task is successful, the system automatically generates a virtual wide table. You can use this table for multi-party secure modeling.

    • If the task fails, review the error message, check the local data on the node, and verify that the data table in the offline sample is active.

What to do next

Start model training