How to enhance your bot’s intelligence by entity annotation

How to enhance your bot’s intelligence by entity annotation

Using entities to easily extract important information from ongoing conversations
Last Updated February 14, 2023
Table of Contents

Entity annotation teaches the bot to recognize context, named entities, and key phrases in the text. It is a vital task for Natural Language Understanding (NLU) workflow.

As discussed in the Entities overview article, entities can be named either ‘category‘ or any other name that is used in a request flow. Here, we are going to explain how to use each type of entities.

The category entity

Let’s say you’ve created the category entity and added the values ‘Smart Watches’, ‘Mobile Covers’ and ‘Headsets’ to it. Now when a user types his request, and the bot interprets the search intent, it is a good idea to narrow the categories in which the bot searched for results. This can be done through entity annotation as follows:

  1. Go to your model.
  2. Select the search intent.
  3. Then open the Annotate Entities button on the top right.
  4. On each example, click the key words you want the bot to detect, and select the category that matches them.

ℹ️ The entities’ names should be in the English language for the model to train successfully.

Request entities

The other important use of entities is automatically understanding a customer’s request and converting it to a transaction. For example, in Telecommunications industry, let’s assume you already have a service Item called ‘Transfer’ that contains some Options (questions) to ask the user when he tries to make a transfer. These questions are:

1- From which number?
2- To which number?
3- What’s the type of transfer units?
4- What’s the amount to be transferred?

Entity annotation help the bot detect all the answers at once when a user types: “I want to ransfer 50 GB from xxxxxxxxxxx to xxxxxxxxxxx” without having to ask each question again in the flow. This can be done following the same annotation steps mentioned above:

  1. Go to your model.
  2. Select the required request intent.
  3. Then open the Annotate Entities button on the top right.
  4. On each example, click the key words that represent a specific entity, and select the entitiy type and value that matches them as shown in the below figure.

ℹ️ After you’re done, you must train the AI model for the changes to take effect. Go to Artificial Intelligence > Models, then click the brain icon next to the model.

Want to learn more?
Was this article helpful?