Add a dataset

Add the data from a table to a data catalog as a dataset through generative AI by using the Now Assist Data Kit application. Adding a dataset is required to create and publish a data collection.

시작하기 전에

Role required: sn_data_kit.admin

프로시저

1. Navigate to All > Now Assist Data Kit > Home.
2. Navigate to Discover datasets and select Get started.
3. On the Datasets tab, select New.
4. Select where to curate data from.
   • I'll import data from Instance table
   • I'll import data from my computer
5. On the Choose data form, select the table and columns.
   If a column does not appear as a possible selection, then the field is not a supported data type. For example, Watch List fields are the glide_list datatype, which is not supported, so Watch List is not a selectable field.
6. 옵션: Select Add column via scripting if you have columns, such as work notes or comments, that aren't stored in the table.
   This is an example script for adding worknotes.

   (function generate(current) {
   //Current is the Gliderecord object of current record.
   // //Sample script for adding worknotes
   // var gr = new GlideRecord("sys_journal_field");
   // gr.addQuery('name', current.getTableName());
   // gr.addQuery('element_id', current.getUniqueValue());
   // gr.query();
   // var records = [];
   // while (gr.next()) {
   //     var obj = {
   //         "value": gr.getValue("value"),
   //         "type": gr.getValue("element"),
   //         "created": gr.getValue("sys_created_on"),
   //         "created_by": gr.getValue("sys_created_by")
   //     };
   //     records.push(obj);
   // }
   // return records;
   })(current);

7. Select Edit filter condition.
8. Review the records and select Continue.
9. On the form, fill in the fields.

   표 1. Add dataset info form

   Field	Description
   Dataset name	Name of the dataset.
   Dataset description	Description of the dataset.

10. Add tags to identify the dataset.
11. Navigate to the Data governance section and select each check box.
    

    • I'm assuring to use data responsibly for AI Evaluation

    • Scan for personally identifiable or information sensitive data before creating datasets. You can turn this off if you prefer.

      주:

      If you opt in, your data is scanned for sensitive data like names or email addresses using vault service. After the scan, records will be highlighted and give you an option to anonymize them. You can also choose to scan the dataset after it is generated.
12. Select Generate dataset.
    Fields of any data type are stored as strings when converted to a data asset.
    The dataset is added to the data assets.

다음에 수행할 작업

After your dataset is added to the data catalog, you can choose to create a smaller dataset by creating a derived dataset or adding a ground truth to your existing data set. For more information, see Create a derived dataset or Add a ground truth to each dataset record.