Configure Match and Merge > Training adaptive AI match models > Training the Adaptive AI match model
  

Training the Adaptive AI match model

Before you can use an Adaptive AI match model for match predictions, you must train and publish the Adaptive AI match model. Before you start training the Adaptive AI match model, you can review and add the match fields that you want to use in the training process. When you start the training process, the training data set is processed to generate record pairs. To train the Adaptive AI match model, label these record pairs that are presented to you in batches.
The training metrics, such as accuracy, precision, and recall, are calculated after a batch of record pairs is labeled. When you reach the desired level of accuracy, precision, and recall, complete the training process.
Note: The training process uses all the data that is available in the MDM SaaS data store.
To train an Adaptive AI match model, you must label the record pairs from the training data set. You can label the record pairs in multiple batches, up to a maximum of 10 batches. Each batch might contain record pairs in multiple sections that are grouped based on exact match fields. For example, you might find a section for record pairs that match exactly based on the Address field and another section for record pairs that match exactly based on the Country field.
Record pairs without exact matches are also grouped in a separate section. You must label 30 record pairs with conclusive labels, Match and Not a Match, within each batch. You can complete labeling a batch with 30 conclusive pairs or continue labeling up to 90 pairs to improve match accuracy.

Label record pairs for training an Adaptive AI match model

Label record pairs in bulk within a section or individually after assessing them. After labeling the required number of record pairs in all batches, you can complete the training process.
    1Open a business entity, and click the Match tab.
    The Match Models page appears.
    2Click the match model that was identified for training the Adaptive AI match model, and click the Adaptive AI Match Model tab.
    The Adaptive AI Match Model page appears.
    3In the Match Fields section, review the match fields.
    The fields that the Directed AI match rules and candidate selection criteria use appear as match fields. If you don't want to use these fields, you can remove them from the training.
    4To remove any of the match fields, perform the following tasks:
    1. aIn the Match Fields section, select the field that you want to remove.
    2. bClick Actions > Remove.
      3. The selected field is removed from the training.
    5To add additional match fields, click Manage Fields.
    6In the Manage Fields dialog box, select the match fields that you want to add, and click OK.
    The selected fields are added for training the model.
    7After you review and add the match fields, click Start Training.
    The training process starts preparing the data for training.
    8After the training data preparation completes, click Start Labeling.
    The Label Record Pairs in Batch page appears. The record pairs are grouped into sections based on exact match fields that paired them.
    9To label record pairs, perform the following tasks:
    1. aExpand a section containing the record pairs you want to label.
      2. Each section contains up to 90 record pairs.
      The following image shows a section that lists record pairs matched exactly by the Address Line 1 field:
      The image shows the record pairs that match exactly based on the PostalAddress.addressLine1 field. Users can label each pair as Match, Not Sure, and Not a Match. After labeling the record pairs, users can save and close the page or finish labeling the batch.
    2. bSelect a record pair and label it with one of the following labels:
      3. Label
      Description
      Not a Match
      A conclusive label. Indicates that the records in the pair are different.
      Not Sure
      An inconclusive label. Indicates that you are uncertain about your decision.
      Match
      A conclusive label. Indicates that the records in the pair are the same.
      Note: Optionally, if you are confident about the record pairs, you can select multiple record pairs within a section and label them in bulk.
    3. cTo label record pairs in bulk, select the record pairs that you want to label, and click Actions.
      4. The following image shows multiple record pairs selected within a section for bulk labeling:
      The Labeling - Batch 1 page displays multiple record pairs that are selected and set for bulk labeling using the labels, Match, Not Sure, and Not a Match.
    4. dSelect a label.
      5. The selected record pairs are labeled based on the selected value.
    5. eTo reset the label assigned to a record pair, select the record pair, and click Actions > Reset.
      6. The label assigned to the selected record pair is reset.
    6. fTo view the details of a record pair, select the record pair and click Actions > View Details. The image shows a record pair selected on the Adaptive AI model training page to view its details.
      7. The Details of Record Pair dialog box appears.The image shows the Details for Record Pair #1 dialog box, which displays the batch completion status and the labels Match, Not Sure, and Not a Match. It also shows the match fields and the records in the pair.
      Optionally, you can continue labeling the record pair without navigating to the Label Record Pairs in Batch page.
    7. gTo label the record pair in the Details of Record Pair dialog box, select a label.
    8. hTo label the next record pair, click Next.
    9. iAfter you complete labeling all record pairs in a section, expand another section and continue labeling until you have selected conclusive labels for 30 record pairs in the batch.
      10. After you select conclusive labels for 30 record pairs, the labeling for the batch is complete. However, you can label up to 90 record pairs to improve the match accuracy. To review the labeled record pairs, click Previous and Next.
    10To complete labeling the batch, click Finish Batch.
    The accuracy of the Adaptive AI match model is calculated.
    11To label the next batch of record pairs, click Next Batch, and label record pairs by repeating step 9.
    12To pause and resume labeling later, perform the following steps:
    1. aTo pause labeling, on the Label Record Pairs in Batch page, click Save and Close.
    2. bTo resume labeling, on the Match Training page, click Continue Labeling, and label record pairs by repeating step 9.
    13After you complete labeling the required number of batches and reach an acceptable accuracy, precision, and recall, click Finish Training.
    The training process completes and saves the Adaptive AI match model as part of the match model.

Keyboard shortcuts

When you view the details of a record pair, you can use keyboard shortcuts to label it and also navigate to other pairs in a batch. You can use the keyboard shortcuts in the Details of Record Pair dialog box.
The following table lists the available keyboard shortcuts:
Action
Shortcut Key
Label a record pair as match.
M
Label a record pair as not sure.
U
Label a record pair as not a match.
N
Reset a labeled record pair.
R
Navigate to previous pair.
Left arrow
Navigate to next pair.
Right arrow