Configure Smart Ticket

User Role: Administrator

Smart Ticket provides the following two out-of-box Smart Ticket (auto-classification) configurations:

  • Standard category field
  • Service category field

These out-of-box configurations are best practices based on the out-of-box data. You can use or modify these configurations, or you can add new configurations that best reflect your business needs.

Add a new Smart Ticket task

In the out-of-box system, two Smart Ticket configuration tasks ("Standard category field" and "Service category field") are used for Smart Ticket by default. You can choose to use these out-of-box Smart Ticket tasks, or you can create new Smart Ticket tasks.

To add a new Smart Ticket task, follow these steps:

  1. From the System Navigator, click System Administration > Ongoing Maintenance > Smart Analytics > Smart Ticket.

  2. Select Blank from the drop-down list, and then click Add.

    Note You can also select one of the out-of-box templates ("Category" or "Affected Service") from the drop-down list, and then click Add to create a new Smart Ticket task based on the template.

  3. Type the task name for the new Smart Ticket configuration.
  4. Go to the Configurations tab.
  5. Select a module for auto-classification. For example, Interaction.
  6. In the Training Sample Query field, define a query to refine the sample data. The default value is category~="service catalog", which means the data that is not in the "service catalog" category can be selected as the training samples.

  7. Select the target fields to be automatically filled by SM Smart Analytics. You can select up to three levels. For example, Category, Subcategory, and Area.

  8. Select the source fields that the auto-classification is based on. For example, title and description.

  9. In the Training Optimization tab, modify the settings for training optimization.

    Note We recommend that you keep the default settings. For more information on improving accuracy for Smart Ticket, see Improving accuracy for Smart Ticket.

    Setting Description
    Training Samples Per Category

    The maximum records to be used as the training samples for each category.

    Default: 200

    Test Data Coverage

    The percentage of records out of the total source data that are used to test the trained system.

    Default: 5

    Source Data Coverage

    The percentage of records out of the total source data that are used to train the system.

    Default: 90

    Training Method
    • Choose "use best terms" for a faster training process if you have huge data volume.
    • Choose "use training documents" for a higher accuracy with a slower training process.

    Default: use training documents

    Adjust Term Weight From Test Result

    Select this option to automatically adjust the term weight for some terms in some categories based on testing result.

    Default: Disabled

    Remove Low Weight Document

    After the training is finished by using the "training documents" method, check the weight of every training document, and then remove the low-weight training documents from the training sample pool.

    Default: Disabled

    Weight Threshold

    The threshold to remove the low weight training documents, after finish training by using the "training documents" training method.

    Min Number of Training Samples

    The minimum number of the training documents in a category. Use this parameter to ensure that a certain number of training samples will not be removed when the system removes the low weight training documents.

  10. Click Add. The new auto-classification task is now added to the Current Configuration List.
  11. Modify the Smart Ticket form (idol.quick.new.interaction) to use the new Smart Ticket task that you just created.

    Note This step is only required for operators to create Smart Interaction from index.do. The new auto-classification task will take effect directly on the user requests from SRC and ESS after a training is performed.

Perform training and testing

To perform a training for a Smart Ticket task, follow these steps:

  1. From the System Navigator, click System Administration > Ongoing Maintenance > Smart Analytics > Smart Ticket.

  2. Click the task name of a Smart Ticket configuration. The Smart Ticket Task screen appears.
  3. Click the Training button to start training this auto-classification.

    Tip You can click Refresh Status to view the latest training status.

  4. When the training is done, click Testing. When the testing is finished, you can view an estimated result of the accuracy for this auto-classification in the Testing Result field.

Note Smart Analytics does not support duplicated category names that are different only in text cases (For example, two categories named as "AVAILABILITY" and "availability"). If such categories exist during the training, the system logs an error and aborts the training process.

Tip The quality of the sample data is critical to the accuracy of the auto-classification. To refine your sample data, you can define a query in the Training Sample Query field under the Configurations tab. For more best practices to improve accuracy, see Improving accuracy for Smart Ticket.

Tip If you disable or enable the Multi-Company mode for Service Manager, you need to delete the existing Smart Ticket configuration tasks and re-create them before you perform training.

Apply a rule-based training

You can append the rule-based analysis on top of the meaning based analysis. The typical scenario is that if one particular record has the same relevancy within several categories, you can append a rule to one specific category to improve the categorization accuracy.

"Rule Field Name" is where you can specify the field based on which you define the rule.

"Apply Rule" lists all the categories, where you can choose the target category and set the value for the rule you want to append.

For example, suppose there are two affected services, "printer_San Diego" and "printer_Shanghai". You can define the rule field as "Primary Contact Location City". Then, set value "San Diego" to the "printer_San Diego", and set value "Shanghai" to the "printer_Shanghai". With this rule, if the contact person for the new coming record is from the San Diego office, the record will be automatically filled with "printer_San Diego" as the affected service.

To apply a rule-based training for an auto-classification, follow these steps:

  1. From the System Navigator, click System Administration > Ongoing Maintenance > Smart Analytics > Smart Ticket.

  2. Click a task name of a Smart Ticket configuration. The Smart Ticket Task screen appears.
  3. Go to the Rule Base tab.

  4. In the Rule Field Name field, specify the field name based on which you define the rule.
  5. Click Apply Rule, and then click Search. A list containing all the categories appears, where you can choose the target category and set the value for the rule that you want to apply.
  6. Click a category.
  7. In the Rule Field Value field, set the value for the rule that you want to apply.
  8. Click the Apply Rule button.

Perform tuning in the Smart Ticket definition

Another way to improve the accuracy of Smart Ticket is to perform tuning continually for the Smart Ticket definition.

To perform tuning in the Smart Ticket definition, follow these steps:

  1. Service Desk agents select tuning candidates during their daily work:

    1. In an interaction record, update the fields suggested by Smart Ticket if the suggested values are incorrect, such as category or affected service.
    2. After the interaction is closed, from the interaction record, click More > Add to Tuning Records to add this record as a tuning candidate for Smart Ticket.

      Note The Add to Tuning Records option is only available when an interaction is in the "Closed" status.

  2. A system administrator tunes Smart Ticket after a period of time to increase the accuracy:

    1. From the System Navigator, click System Administration > Ongoing Maintenance > Smart Analytics > Smart Ticket.
    2. Click a task name of a Smart Ticket configuration. The Smart Ticket Task screen appears.
    3. Go to the Tuning tab.
    4. Click Manage Tuning Records to open Tuning Records where you can find all the tuning candidates.
    5. Delete the meaningless or inappropriate records. The rest of records will be used in tuning Smart Ticket.
    6. Click the Tuning button to start the tuning process.

Configure Smart Ticket for multi-company

SM Smart Analytics supports multi-tenancy. When multi-company mode is enabled in Service Manager, you can configure specific Smart Ticket task to apply to multiple companies when applicable. The Smart Ticket configuration takes effect on these companies individually by segregating their data in Smart Analytics database.

To specify the companies in a Smart Ticket configuration, follow these steps:

  1. From the System Navigator, click System Administration > Ongoing Maintenance > Smart Analytics > Smart Ticket.

  2. Click a task name of a Smart Ticket configuration. The Smart Ticket Task screen appears.
  3. Click the Multiple Company tab, and then do one of the following:

    • Click Add Company to add companies to this configuration.

      Note A training is needed if you add a new company.

    • Click Remove Company to remove companies from this configuration.

Tip If you are unable to see the Multiple Company tab, see the related topic in Troubleshooting: Smart Analytics setup.

Note Mandanten in Smart Ticket supports only multiple company.

Configure Smart Ticket for OCR

  1. From the System Navigator, click System Administration > Ongoing Maintenance > Smart Analytics > Smart Ticket.

  2. From the toolbar, click More or the More Actions button, and then select Tailoring. The Smart Ticket Settings page opens.
  3. Update the following fields as needed.

    Field Name Description
    Use FFT (Fast Fourier Transform) to remove the grid noise during OCR

    The default value is true.

    When selected, the system uses Fast Fourier Transform (FFT) to remove the grid noise during OCR. However, using FFT consumes around 400 MB memory for each image and makes the OCR process slower. You can disable this feature if you want to speed up the system performance.

    Maximum image size allowed to perform OCR without resizing (pixel):

    The default value is 2000 pixels. It is recommended that you keep the default value unchanged.

    Specifies the maximum pixels of an image that is allowed to perform OCR without resizing. Images that larger than the specified value will be resized before Smart Analytics performs OCR.

    The images size after resizing (pixel):

    The default value is 700 pixels.

    Specifies the number of pixels to which an image is re-sized when the image size exceeds the maximum image size allowed to perform OCR without resizing.

    To ensure best OCR quality, the recommended value for this field is from 600 to 1200.