As you have now identified some issues in your NLU performance it is time to augment the training data to improve the NLU performance.
This step only applies if you used the Test Case Wizard to download the full training data from your NLU engine in the previous step of this tutorial. Otherwise you will have to apply the changes to your training data in your NLU engine itself - here are links to the documentation of some of the supported providers:
- [Microsoft LUIS](https://docs.microsoft.com/en-us/azure/cognitive-services/luis/luis-how-to-add-intents#working-with-an-individual-utterance)
- [IBM Watson](https://console.bluemix.net/docs/services/assistant/intents.html#editing-intents)
- [Google Dialogflow](https://dialogflow.com/docs/intents/training-phrases)
- [Wit](https://wit.ai/docs/quickstart).ai
- [Amazon Lex](https://docs.aws.amazon.com/lex/latest/dg/ex-utterances.html)
- [RASA](https://rasa.com/docs/nlu/master/dataformat/)
- [QnA Maker](https://docs.microsoft.com/en-us/azure/cognitive-services/qnamaker/how-to/edit-knowledge-base#add-a-qna-pair)
With the help of Botium Coach you now have decided what actions to take
-
in general, there are three possible actions to augment your training data:
-
Add additional user examples for specific intents
-
Remove user examples from intents
-
Move user examples from one intent to the other
You can use the Botium Box Test Case Designer to perform these actions.
-
Use Paraphraser to Generate New User Examples
Botium Box includes a paraphraser to quickly generate new user examples based on given ones. After adding a handful of user examples to the utterance list, click on the Paraphrase it! button to get a couple of suggestions for additional user examples and select the ones you want to use.
Comments
0 comments
Please sign in to leave a comment.