Train and use the similarity solution definition for issue assignment prediction

Train and use the machine learning solution by activating the Governance, Risk, and Compliance Predictive Intelligence plugin. The solution enables the system to automatically suggest correct assignees to route issues. These suggestions reduce the time spent on deciding who should own the issue.

Before you begin

Make sure that the following plugins are activated.

• The Policy and Compliance Management plugin
• The Risk Management plugin
• GRC Profiles plugin must be activated
• The Predictive Intelligence plugin. For more information, see Install Predictive Intelligence.

Modify the following GRC properties with the sn_grc.admin role.

• Issue assignee suggestion based on property (Similarity Analysis).
• Machine Learning Solution for prediction of issue assignee property (ml_x_snc_sn_grc_pred_intel_global_similarity_solution_definition_for_assigned_to_for_issue).

Role required: ml_admin

Procedure

1. Navigate to Predictive Intelligence > Similarity > Solution Definitions.
2. Select the Similarity Solution Definition for Assigned To for Issue solution.

   Table 1. Similarity Definition form

   Field	Value
   Label	Unique name for the similarity solution.
   Name	This field is automatically set to a system-assigned, read-only name based on your value in the Label field.
   Word Corpus	Existing word corpus that is relevant to your solution. For this use case, select the Word Corpus for Issue Assignment word corpus. For information on limits to the number of records per table for Word Corpus use, see Predictive Intelligence properties. Note: Starting from the Washington DC release, a word corpus isn't needed for similarity solutions. For more information, see Create a word corpus.
   Table	Table that contains the records that you want to train against. Based on the table value, the number of records matching your filter conditions is displayed as a link. Select this link to view the list of records. For this use case, the field is automatically set to the Issue [sn_grc_issue] table. Do not modify this value for this solution definition.
   Fields	Columns in your Table that are likely to help in identifying the appropriate assignee. The information in these fields is used as training data. In this use case, the Short description and the Description fields are selected by default. These fields contain the text of the issue records whose owner you're trying to identify. Note: You can modify the fields selected here if there are other important fields on the issue record in your database, such that these fields can help in finding out similar issues for predicting the issue owners. The fields you select should not have empty values.
   Test Table	Table that contains the record that you want to predict for. For this use case, the field is automatically set to the Issue [sn_grc_issue] table. Note: The number of records which the Similarity window can retrieve is limited to 10. This field must not be modified for this solution definition.
   Test Fields	Fields which are used as input during prediction. In this use case, select Short description and Description. Note: You can modify the fields selected here if there are other non-empty important fields on the issue record in your database, such that these fields can help in finding out similar issues for predicting the issue owners.
   Filter	Filter that applies conditions to the records that you're using as an information source for training. For example in this use case, you can set an Assigned to [is] [not empty] condition because historical Assigned to values are needed to provide suggestions for future issues.
   Processing Language	Dominant language of the dataset that you are training on the solution definition. If the dataset language is English, choose English. By default, English processing is applied to all datasets. For example, if you select Italian, the system processes the data in both Italian and English. Note: The term processing indicates some of the language-specific steps that are used as part of training a solution. These steps include tokenizing words, removing stop words, and stemming.
   Stopwords	List of stopwords. When you select your processing language, the system automatically adds a Stopwords list for that language. For example, if your processing language is Italian, the Default Italian Stopwords option appears. The Default English Stopwords list is also included. You can add your own custom list of stopwords. For more information, see Create a custom stopwords list.
   Training Frequency	Select a retraining frequency. The available options range from Run once up to Every 180 days.
   Update Frequency	Frequency of how often you want to refresh the data that you use to retrieve your similarity results. For example, new issues typically occur frequently throughout the day. If you have issue records that are open, you may want to select an update frequency of Every 15 minutes. This frequency can increase the likelihood that newly opened records are included in the refresh.

3. Click the appropriate button for the solution definition.

   Option	Description
   Save	Save your solution definition record so that you can return to it later.
   Update and Retrain	Modify your solution definition and train it.