08-pre-trained-entities.md

How to add Pre-trained Entities

This tutorial shows how to add Pre-Trained Entities to your Conversation Learner Model.

Video

Requirements

This tutorial requires that the general tutorial Bot is running

npm run tutorial-general

Details

Pre-Trained Entities recognize common types of Entities, such as numbers, dates, monetary amounts, and others. They work "out-of-the-box," do not require any training and their behavior cannot be changed unlike custom entities. By default, Pre-Trained Entities are multi-valued, accumulating every identified instance of the Entity.

Steps

Start on the home page in the Web UI.

Create the Model

1. Select New Model.
2. Enter PretrainedEntities for Name.
3. Select Create.

Entity Creation

1. Select Entities in the left panel, then New Entity.
2. Select Pre-Trained/datetimeV2 for Entity Type.
3. Check Multi-valued to enable the Entity accumulate one or more values. Note, Pre-Trained Entities cannot be negatable.
4. Select Create.

1. Select Actions in the left panel, then New Action.
2. Enter The date is $builtin-datetimev2 for Bot's Response....
3. Select Create.

Create the Second Action

1. Select Actions in the left panel, then New Action.
2. Enter What's the date? for Bot's Response.... Pre-Trained entities cannot be Required Entities as they are recognized by default for all utterances.
3. Enter builtin-datetimev2 for Disqualifying Entities.
4. Select Create.

Train the Model

1. Select Train Dialogs in the left panel, then New Train Dialog.
2. Enter hello for the user's utterance in the left chat panel.
3. Select Score Actions.
4. Select What's the date? from the Actions list
5. Enter today for the user's utterance in the left chat panel.
   • The today utterance is automatically recognized by pre-trained models in LUIS.
   • Hovering over values of Pre-Trained Entities shows additional data provided by LUIS.

Next steps

[!div class="nextstepaction"] Entity Resolvers