NLU Custom components possibilities

[musabukhari]

Hi all. Is it possible to create custom components that fit into the NLU pipeline that can perform tasks such as text summarization, machine translation, numerical analysis etc?

I’m more familiar with the spaCy world where you can customize your pipeline for a specific task. Wonder if thats the same case with RASA NLU. Basically, can RASA NLU be utilized for wider NLP tasks? All I’ve seen are examples of the sentiment analyzer, and i’m looking for other capabilities as well.

Thanks in advance

[tyd (Ty Dunn)]

@musabukhari Yes, it is possible to create custom components to do whatever you want. In addition to the sentiment analysis example, here are some more examples:

• How to make a Custom Printer Component in Rasa NLU
• rasa-nlu-examples
• Using Custom spaCy Components in Rasa