3 steps to convert chatbot training data between different NLP Providers

QBox is a great tool for assessing the quality of your chatbot’s training data, but did you know it can also be used to convert training data from one NLP provider’s format to another?

This easy conversion means you can start with a JSON file containing your intents and entities in the IBM Watson format, and seconds later have a ZIP file containing that same information that can be imported into DialogFlow.

QBox currently supports IBM Watson, Microsoft LUIS, QnA-Maker, DialogFlow, Wit, RASA and Amazon LEX and you can do this for free.

Here’s how to convert your training data

Step 1: Export your training data from your existing provider.

You should end up with a JSON or ZIP file, depending on the provider. If you are unsure, we’ve put together instructions on how to do this on our Get Started page.

Step 2: Configure and run a test on QBox

From within QBox, create a new job, and upload the training data file you’ve just exported. You can choose any names you prefer for the job name and project name fields.

Scroll down to the “Providers to run job against” section and select the providers you want to convert to. The native provider for the file you uploaded will also be selected by default, but you can uncheck this to make the conversion process run faster.

Click Create job and wait a few minutes for the test to run.

Step 3: Download the converted file

Once the test has finished, you’ll see it listed with any other tests in the project. If you selected multiple NLP providers to run the test against in the earlier step, then you’ll see an intermediate screen listing the results for each provider, select the one you want to review first.

Once on the test results page, select the download icon at the top of the screen. You’ll see an option to download your converted training data. This file can now be imported into your other NLP provider.

Hope you enjoy this easy guide on how to move from one provider to another in 2 minutes.