Creation, training, and analysis

Before using the commands from the examples presented in this section, it is necessary to set the environment variable with the NLU service URL provided by CPQD. In this case, you can simply do something like this:

export NLU_HOST=https://address:port

In the provided commands, the access authorization field is not included, but it must be sent in all commands.

• Sending Access Authorization

To use an NLU agent, we must first create the agent.

• Create an Agent

After creation, it’s already possible to use the sentiment analysis. However, to be able to use the remaining functionalities, it’s necessary to configure and train the agent.

• Configure the agent

• Train the Agent

Training may take a few seconds or several minutes, depending on the size of the agent.

After being trained, the agent can be used for sentence analysis.

If the agent’s training encounters an error, the previous model is retained. Therefore, it’s possible to use sentence analysis while the agent is being fixed. For more information about model states during training, please refer to Model status.

Summary of the steps:

• Access Authorization
• Create an Agent
• Sentiment Analysis
• Configure agent with JSON File
• Training the agent
• Conversation Analysis