Building machine learning entity extraction models
Use Pega Platform machine learning capabilities to create entity extraction models to recognize named entities.
- Define an entity extraction model in which to accommodate the entities trained as a result of machine learning. For more information, see Creating entity extraction models.
- Ensure that the system locale language settings are set to UTF-8.
- Specify a repository for text analytics models. For more information, see Specifying a database for Prediction Studio records.
By using models that are based on the Conditional Random Fields (CRF) algorithm, you can extract information from unstructured data and label it as belonging to a particular group. For example, if the document that you want to analyze mentions Galaxy S8, the entity extraction model classifies that as Phone.
- Preparing data for entity extraction
In the Source selection step of the entity extraction model creation wizard, select the extraction type and provide the data for training and testing of your entity extraction model.
- Defining the training set and training the entity extraction model
In the Sample construction step of the entity 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.
- Accessing entity extraction model evaluation reports
After you build the model, you can evaluate it by using various accuracy measures, such as F-score, precision, recall, and so on. You can view the model evaluation report in the application or you can download that report to your directory. You can also view the test results for each record.
- Saving the entity extraction model
After the model has been created, you can export the binary file that contains the model to your directory and store it for future use. You can also create a specialized rule that contains the model. That rule can be used in text analyzers in Pega Platform.
- Best practices for creating entity extraction models
Use extraction analysis to detect and classify named entities into predefined categories, for example, names of people, locations, organizations, and so on.