Detect the intention of a sentence typed in natural language

Vizir allows you to use machine learning algorithms capable of detecting the main intention of a sentence typed in natural language. It is the most advanced technology on the market in terms of natural language understanding.

Intent detection

The best chatbots use machine learning tools to detect the intentions of sentences typed in natural language.

First: Vizir works by default with these NLU (Natural Language Understanding) algorithms. If you work with us you will find that we strongly recommend using this technology, often in addition to Elastic Search algorithms.

Define the main topics

If they are machine learning algorithms, be careful not to extrapolate: chatbots can only understand what you teach them (like little children).

So the first step is to define on a piece of paper (or rather an Excel spreadsheet) the main topics that the chatbot should understand. You can look at our Excel templates for that!

Then we’ll help you classify it correctly under different intentions.

Train intent detection in our no-code platform

Once this preparatory job is done, you must then proceed to the parameterization in Vizir.

Everything is done very simply from our no-code interface.

You start by entering your intentions as below.

Add training sentences

Once the intentions are added, your algorithms need data to learn what the given intention corresponds to.

You will need to add 5 to 10 practice sentences per entity to get started.

You can enter them manually or copy and pastethem massively from an Excel file.

How to set up an NLU

Discover our 13 articles to help you set up an NLU.

Train your NLU with user requests

Once your chatbot is launched, Vizir displays on a visual interface all the requests typed by your users to the chatbot.

You will be able to analyze the understanding of the algorithm and correct it for the sentences that were misunderstood.

This is what we call training a chatbot.