Review the following best practices and solutions to problems that you might experience when working with the natural language processing (NLP) capability of Pega Platform to analyze incoming messages in your conversational channels, such as email or chat.
- General troubleshooting tips for NLP
When you encounter a problem while working with the natural language processing (NLP) features of your application, take these steps to identify and understand the underlying cause.
- Topics or entities are not detected
You might notice that your application does not detect topics or entities correctly in incoming emails or chat messages. Common causes of issues with topic or entity detection include misconfiguration or insufficient model training. Learn about the solutions to these problems.
- Email parser does not detect the body, signature, or disclaimer
If the email parser does not properly recognize the body, signature, or disclaimer in emails, it might indicate that the underlying pxEmailParser model is not sufficiently trained with training data from your domain.
- Language is not detected
If you notice that the system does not properly detect the language of incoming emails or chat messages, use the following tips to resolve the problem.
- Running out of memory when using NLP
If you have high memory consumption or exceed memory usage when using text analytics in Pega Platform, use the following best practices to improve your system performance.
- Training data for NLP models is missing
If you notice that training data for models used in natural language processing (NLP) is not available, inadequate configuration of the analytics repository might be the cause.
Previous topic Values of the pxOutcomeTime property are not unique in the Oracle database Next topic General troubleshooting tips for NLP