What is an Intelligence?

Mallu Updated by Mallu

Overview

A Dataset or Intelligence is the main part of your bot. That's where all the training, test and translation phrases are. It's also where you'll find the general settings.

On the Weni homepage, when you click on Artificial Intelligence the platform will present all the intelligences linked to your project.

By clicking on the "Public Intelligences" card you will find the list of all intelligences created so far by other users. For each of them, we can see the name, author, languages and categories.

Click an intelligence to enter the Dataset main page.

Researching an Intelligence

You can search for a specific dataset by typing the name in the search bar at the top of the homepage.

Categories

A dataset category refers to what context area is the training related to. In the example above, we searched for the name "COVID-19" that are related to Health. Use categories for other users to easily identify the purpose of your intelligence.

How do Languages work at Weni?

One of Weni's core features is its ability to have multiple languages for the same intelligence!

This allows the user to perform the same intelligence training for several different languages, so the bot can be used in several countries within the same context without changing their assertiveness in predictions, because it has the same base of training phrases!

At Weni, you can translate your dataset into over 150 languages in a very easy way!

Quick Test

When training on Weni you may find the need to test your dataset in an easy way. To help you with that, BotHub has a quick test functionality inside the intelligences.

To access it, enter the repository and click in the tab "Quick Test" on the right of the screen.

Then, type your input phrase, and the prediction will happen based on the dataset trained in this repository.

The quick test brings us some important information about the intelligence, which are:

  • the intent, entities, and confidence for the input;
  • a debug option that shows us the weight of each word on the phrase in the prediction;
  • the raw data for the prediction (JSON format);

How did we do?

Repository - Overview

Intents and Entities

Contact