Natural Language Understanding

DRUID NLU gives bot authors the possibility to allow their chatbots use semantic-based technology to recognize user intent with a high degree of accuracy and also better handle complex human conversations.

Semantics means the meaning and interpretation of words, sentence structure and semantic connection between words. Semantic text classification algorithms assign a meaning to every word in a text, and then compute the meanings of sentences based on the meanings of those words and how they are put together. These semantic-based driven algorithms reach better classification performances enabling chatbots to understand the meaning of what the user says.

When to use NLU?

Use NLU when you have a knowledgebase bot with articles (QnAs) which have more than 100 intents.

Enable NLU on the bot

In DRUID, NLU is disabled by default. To enable NLU, on the NLU menu, click Configurations. Click the Intents tab and click on the Thresholds and parameters section header.

Add the parameter NLU.NER.Classification.ModelType and choose the desired model type for NER classification, then click the Save icon () displayed inline.

Note: The parameter NLU.NER.Classification.ModelType is available starting with DRUID version 7.18. In earlier versions, if NLU.NER.Classification.UseSemantic was set to "true", it will appear in the NLU Experimental Parameters section. You can remove it or switch to the new NLU.NER.Classification.ModelType parameter.

For DRUID versions prior to 7.18, add the parameter NLU.NER.Classification.UseSemantic, set the value to “true” and click the Save icon () displayed inline.

Click the Save button at the bottom of the page to save the settings, then click the Train button to train the bot.