How can I get training data for my chatbot? + more FAQs

Getting Training Data for Chatbots and Assistants

Getting a good training data set is crucial for building a chatbot or assistant. We have a couple of datasets to help you get started on the Community Forum.

It’s also a very good idea to pretend to be the bot yourself.

But the most important factor is always real user feedback. At Rasa we strongly believe that real conversations are more important than hypothetical ones. The best data of all is the result of real humans interacting with your AI.

Is there a GUI for labeling data?

Open Source

  • rasa-trainer-ui is an open source tool for annotating NLU data.
  • Articulate is an open source platform for building conversational interfaces.
  • Rasa UI is a web application which uses Rasa NLU as its backend.

Commercially Supported

Rasa Platform comes with a graphical interface for creating both Rasa NLU and Rasa Core data.

Building Advanced chatbots and assistants

Where can I learn the fundamentals of chatbots?

Rasa has developed a course on building chatbots in python together with DataCamp.

What does it mean to build machine-learning based dialogue?

In our experience, rule-based dialogue systems do not scale. There has been a lot of research on using machine learning to overcome these limitations, this post goes into a lot more depth.

What are some examples of great bots built with Rasa?

There are hundreds of great bots out there built with Rasa, built by everyone from startups, to NGOs, to the Fortune 500. We are working on a directory, so email hi@rasa.com if you’d like to be included!

  • Meekan is a popular slack bot for scheduling meetings.
  • Tia is a women’s health company with a chatbot as part of the core experience.
  • Helvetia is an insurance company that lets you report bike thefts via a Facebook Messenger chatbot.

I need help developing a bot

Rasa works with a number of partners who can help you with your project. Reach out hi@rasa.com and let us know what you’re looking for, and we’ll recommend a partner.

How many languages does Rasa support?

It depends on your exact setup. Rasa NLU has a number of different pipelines. The tensorflow_embedding pipeline supports any language that you can tokenize, whereas the spacy based pipelines work with any existing spaCy language model. Rasa Core is independent of which language you use.

Can I create custom models for NLU and Core?

Yes! One of the big advantages of using open source code is that you can customise everything. There is no one-size-fits-all solution to machine learning problems. You can build custom Rasa Core models by creating your own Policy, and for Rasa NLU by creating Custom Components .

Deploying Chatbots and Voice apps in Enterprises

Is there a way to get enterprise-grade support?

Yes. Rasa Platform comes with enterprise-grade support and a choice of SLAs to suit your requirements.

How can I connect Rasa to my backend enterprise systems?

The best way to do this is with Custom Actions.

How do I deploy Rasa in production?

We recommend docker for production deployments. For the Rasa Stack we provide pre-built docker images on docker hub. `Rasa Platform <http://legacy-docs.rasa.com/docs/platform/`_ is also containerized, and runs either on premise or on your private cloud. It ships with production-ready container orchestration.

How can non-engineers contribute to bot development?

Rasa Platform is our enterprise product.

How can I add a Rasa Bot to my website?

There are multiple open source projects for adding a Rasa-compatible chat widget to a website.

How can I create a custom integration for Rasa?

It’s straightforward to implement a custom channel (e.g. for company-specific chat software). Check out the docs here.

I’ve already built a bot with the Rasa Stack. Can I easily upgrade to Rasa Platform?

Yes. Rasa Platform is installed alongside Rasa Core and NLU. Rasa Platform can work with your already existing Rasa NLU and Rasa Core servers.

About Rasa the Company

How does Rasa make money if everything is open source?

Rasa Platform is a paid product which we offer in addition to the open source Rasa Stack.

How do I get access to your roadmap?

As part of our customer success program, companies have access to our roadmap and we work closely with them to prioritize upcoming features and get early feedback.

How can I get in touch?

If you have a question about how to use Rasa, the Rasa Community is the best place to help. For bug reports and feature requests, please go to GitHub. For everything else please email hi@rasa.com