Creating training data sets for BMC Helix Business Workflows

To utilize BMC Helix Cognitive Automation capabilities to predict case templates and categories for cases, as an administrator, you train the cognitive service to understand your data and use it in the application.

Important

To leverage the cognitive capabilities, you must have a license to use BMC Helix Cognitive Automation.

As an administrator, you create a data set to train the cognitive service from sample data or from your application data. The following table describes the types of training data sets that you can create:

Training data set typeData set contents
CSV Data Set

Uses data stored in a CSV file.

Note: A CSV training data set file must have at least 5 rows, and it must not contain more than 25,000 intent examples associated with 2,000 intents

BMC Helix Innovation Suite Data Set

Uses your application record definitions and record fields.

For applications that do not require continuous cognitive service training, train the cognitive service by using data from CSV files.

For applications that are used in a business environment that changes often, you need to continuously train the cognitive service. For such applications, you provide the cognitive service training data from your applications. This approach helps the cognitive service to get the updated data and provide suggestions according to the business changes.

To create a training data set

  1. As an administrator, log in to BMC Helix Innovation Studio and navigate to the Administration tab.
  2. Select Case Management > Cognitive Training Data Set.
  3. On the Auto-classification Training and Evaluation tab, in the Data Sets section, click New, and perform one of the following steps:
    • To create a data set that uses a CSV file, select CSV Data Set, and in the New CSV Data Set pane, complete the fields as described in the following table:

      FieldDescription
      Data Set NameEnter a name for your training data set.
      DescriptionEnter a description that provides details about your training data set.
      TrainingTypeThe IBM Watson Conversation for natural language classification option is populated automatically and you cannot change this option.
      CSV FileClick Choose File and upload your training data set.
      LocaleSelect the locale in which you want to display your training data set.
      Training DataSelect the percentage of the CSV data that you want to use as training data. 
      Test DataThe percentage of CSV data that you want to use as test data is automatically calculated according to the Training Data percentage. 
    • To create a data set that uses the application data, select Innovation Suite Data Set, and in the New Training Data Set pane, complete the fields as described in the following table:

      FieldDescriptionValues to add for template predictionValues to add for category prediction
      Data Set NameEnter a name for your training data set.Not ApplicableNot Applicable
      DescriptionEnter a description that provides details about your training data set.Not ApplicableNot Applicable
      TrainingTypeThe IBM Watson Conversation for natural language classification option is populated automatically. You cannot change this option.Not ApplicableNot Applicable
      Record Definition NameSelect a record to provide data to the cognitive service.CaseCase
      FilterSelect the Company ID for which you want to create the data sets. Enables the cognitive service to create data sets specific to a company. For example, Calbro Services.

      To map the correct trained data set with the correct company, you can use the following naming convention while creating your data set:

      • Category datasetName for companyName
        For example, Category Data Set 1 for Calbro Services.

      • Template datasetName for companyName
        For example, Template Data Set 1 for Calbro Services.

      <Company ID><Company ID>
      Text FieldsClick Add/Remove Text Fields and select the field from the list of available fields. The text fields enable the cognitive service to use the data from the selected field to predict the case templates.SummarySummary
      Category FieldsClick Add/Remove Category Fields and select the field from the list of available fields. The category fields enable the cognitive service to use the data from the Summary field, and to return the Template IDs for the predicted case templates and Category IDs for predicted categories.

      Notes:

      • For predicting templates, if you use a field other than Template Name in the Category Fields, the cognitive service fails to predict the case templates.
      • For predicting categories, if you use a field other than Category Tier 1Category Tier 2, and Category Tier 3 in the Category Fields, the cognitive service fails to predict the case templates. Ensure that you add the fields in the correct order, such as, Category Tier 1Category Tier 2, and Category Tier 3.
      Template Name

      Category Tier 1

      Category Tier 2

      Category Tier 3

      Training DataSelect the percentage of the CSV data that you want to use as training data. Not ApplicableNot Applicable
      Test DataThe percentage of CSV data that you want to use as test data is automatically calculated according to the Training Data percentage. Not ApplicableNot Applicable
  4. Click Save.
    The new training data set is displayed in the Training Data Sets section. You can delete the training data set or create a copy of the existing training data set.

After you create the training data set, you can train and test the cognitive service. Learn more about it in  Training and testing the cognitive service for a custom application Open link .

Where to go from here

Enabling template and category prediction


Was this page helpful? Yes No Submitting... Thank you

Comments