Open source contribution

Hi team Earlier I was using Dialogflow by google. In Dialogflow there are some impressive functionality like

  1. automatically recognizing entities like country, time, city. ,
  2. no need to write stories, just one to one mapping but have a context.
  3. intents like small talk are already integrated.

out of these, I’ve implemented some in rasa. So I’m thinking of contributing. Please let me know how can I do this. This is my first-time contribution to any open source. And anyone already working on the similar things some resources will be helpful. Also, any documentation for the code will be helpful.

@Emma @amn41 @Juste @akelad, please assist

1 Like

Hey @capgos17

Firstly, it’s awesome to hear that you want to contribute to Rasa Open Source!

To get started with code contributions, it would be best to check out this guide: How to make a code contribution.

As mentioned in the guide, it’s important to first consider if you want to contribute to an existing issue, or work on a potentially new feature. Both of these options are covered in the guide, however, our contributor dashboard on Github is also a key resource to help you check current open issues we’d like support with.

We also have a contributor program, designed to recognise efforts and achievements within the community. Since you plan on starting your contributor journey soon, I definately encourage you to take a look at those perks too. :slight_smile:

Hope that helps, let me know if you have further questions.

Hey @capgos17 before contributing these things it would be good if you create issues for your proposed features. E.g. 1 + 2 are already doable with Rasa Open Source - or do you have something different in mind?

Hey, @akelad thanks for that.

Can I maintain the context without the story? That’s what I want to achieve. And I know entity extraction is present in rasa but for the same, we need to put them in nlu data. For ex: if I have to extract the name of a country “India” I’ve to put the same in nlu data. I just want to add these common entities to the default rasa init

Yes you can maintain context without the story, that’s what the MappingPolicy is for.

As for common entities like countries, you can use pretrained entity extractors for that. Like SpacyEntityExtractor for example. rasa init is just designed to get you up and running quickly as of now. We do however also have a training data repo where we’d be happy to accept any contributions around training data: GitHub - RasaHQ/NLU-training-data: Crowd sourced training data for Rasa NLU models