Skip to main content

         This documentation site is for previous versions. Visit our new documentation site for current releases.      

This content has been archived and is no longer being updated.

Links may not function; however, this content may be relevant to outdated versions of the product.

Defining the training set and training the text extraction model

Updated on March 11, 2021

In the Sample construction step of the text extraction model creation wizard, select the data to use to train the model and the data to use to test the model's accuracy. In the Model creation step, build the model.

During the training process of a text extraction model, the Conditional Random Fields (CRF) algorithm is applied on the training data and the model learns to predict labels. The data that you designate for testing is not used to train the model. Instead, Pega Platform uses this data to compare whether the labels that you defined (for example, Person, Location, and so on) match the labels that the model predicted.
  1. If you want to keep the split between the training and testing data as defined in the file that you uploaded, in the Construct training and test sets using field, select User-defined sampling based on "Type" column.
  2. If you want to ignore the split that is defined in the file and customize that split according to your business needs, perform the following actions:
    1. Select Uniform sampling.
    2. In the Training set field, specify the percentage of records that is randomly assigned to the training sample.
  3. Click Next.
  4. In the Model creation step, make sure that the Conditional Random Fields check box is selected.
  5. Click Next.
    Result: The model training and testing process starts.

Have a question? Get answers now.

Visit the Support Center to ask questions, engage in discussions, share ideas, and help others.

Did you find this content helpful?

Want to help us improve this content?

We'd prefer it if you saw us at our best. is not optimized for Internet Explorer. For the optimal experience, please use:

Close Deprecation Notice
Contact us