Natural Language Processing (NLP) text analyzers
Incoming messages from Facebook Messenger are processed through a series of text analyzers to determine what the user wants to accomplish. These text analyzers are processed in a sequential order.
Pega Customer Service provides five text analyzers for Intelligent Virtual Assistant for Facebook Messenger. The following table describes these text analyzers and their purpose.
|Text analyzer||Purpose||Modifications recommended?|
|Pre-NLP > Pre-NLP Processing||Manages the status of an Intelligent Virtual Assistant for Facebook Messenger to understand whether the Messenger is currently in the middle of a service case or is interacting with a live agent.||No|
|In-Case Escapes > NLP||Natural Language Processing for circumstances where the user is in the middle of a service case. Prevents customer responses for case queries from being interpreted as commands in the broader NLP definition.||Yes|
|Global NLP > NLP||Natural Language Processing for when the customer is not in the middle of a service case. All queries/responses and case commands should be provided for this NLP taxonomy.||Yes|
|Post NLP > Post-NLP processing||Manages commands that have special functions after the message has been received, such as the queuing of commands or escalation to chat.||Yes|
|Simple > Simple||A case-sensitive text analyzer that interprets the exact text entered in the message and compares it to the literal names of commands.||No|
Pre-Natural Language Processing (NLP) and post-NLP processing allows the system to handle different scenarios, such as deciding whether the user has escalated to an agent or determining if a user is currently in the process of executing a case flow. NLP analyzers are linked to the Pega Natural Language Processing engine and require configuration to map the language that you anticipate the customer will be using to the commands that you want to execute.
The Simple text analyzer interprets the exact words that are entered and checks if there is an exact match for the requests. This is useful for commands that you want to take in, such as those that are preceded with the word _cmd. You do not need to modify any of the text analyzers except for updating the taxonomies that are related to the NLP analyzers.
For NLP configuration, you can configure either a rule based model or free text model. You can use the sample rule based model as a starting point and modify it per your business needs. For more information on configuring a NLP taxonomy, see Requirements and best practices for creating a taxonomy for rule-based classification analysis.