Intents are collections of identified user intents (e.g. what does the user want).
The user input is scored against all example sentences and Cognigy.AI will identify the intent with the highest score as the intent for this input if it is above the configured thresholds.
The training menu is available under the "NLU" tab in the flow editor. Any flow that requires training will display a red point next to its name in the list of flows and on the NLU tab in the flow editor.
Flows that are only processed as an attached flow do not need to be individually trained (the exclamation mark can be ignored) as the training of the attaching flow will pull in the intents of the attached flows and the training will be done there. Flows that are used separately with intent mapping need to be trained.
Cognigy NLU requires pre-training of the intent mapping model.
To train your model go to the Intents tab in your Flow Editor and click the blue TRAIN INTENTS button.
A new Task will begin and the spinning task icon indicates training is in progress. While new models are trained the latest version of your flow will fall back to the most recently trained model.
Successful training is indicated with a green success notification. It is possible that training may fail, this will be indicated by a red error notification. Expand the error message to learn more, for example, you may have to add more example sentence training data to complete training successfully.
Any time adjustments are made to the training data, the NLU model must be retrained to incorporate the changes. The following actions also trigger the need to train the intents (the red exclamation point next to the Train Intents icon will appear):
- creating, editing and deleting an intent
- attaching a Lexicon
- editing a Lexicon
- attaching and detaching a Flow
This is where creators are given the power to add unique levels of customized understanding to virtual agents which enable them to react intelligently to any message received by an end user.
Each Intent can have any number of trained example sentences. These sentences should be a broad range of possible user inputs that are expected to activate the Intent. Building a comprehensive example sentence list is a critical step in the process of building a smart AI service.
Rather than writing similar sentences for different entity names i.e.
I want to order Pizza, I want to order Pasta...build and attach a Lexicon to the flow that can be Annotated into your sentence.
A minimum of 5 example sentences are required for each Machine Learning Intent to ensure the NLU model can be sufficiently trained. If the minimum of 5 example sentences is not reached, a yellow warning icon will be displayed to prompt the user for further sentences. Any attempt at training the intents when less than 5 example sentences have been added will result in an error message being displayed.
Intent example sentences should be as distinct and unique as possible. If there is too much similarity and overlap with other intents, then the intent mapping will become less predictable and, if configured, intent confirmation sentences will be triggered more frequently.
Evaluation of Example Sentences
Find out more about how Cognigy.AI evaluates example sentences on the Intent Analyzer Page.
When writing example sentences, it is possible to add System Slots and attached Lexicons to the sentence structure by using the Annotations feature.
If you have configured Lexicons with similar items (e.g. pizza = FOOD and cake = FOOD), you don't need to enter similar sentences for both. Build a general sentence that will be configured with Annotations to recognize your custom Lexicon keyphrases e.g.
I want to order some FOOD.
Simply highlight the word that should be mapped as a slot or lexicon and click the "+" at the right end of the text field to open the Annotations menu. The annotations menu provides options to select the type of slot that should be mapped to the highlighted word.
Example Sentence Annotation
Find out more about recognizing Slots and Lexicons in example sentences on the Annotations Page.
Intents can fall into one of three thresholds:
The Intent has a score higher than the confirmation threshold and is considered "confirmed"
The Intent has a score higher than the reconfirmation threshold, but lower than the confirmation threshold and must be reconfirmed by the system
The Intent has a score lower than the reconfirmation threshold and isn't considered a valid result for this input
You can change these thresholds in Agent Settings.
In case you haven't defined a confirmation sentence for your Machine Learning intent the reconfirmation threshold is used to confirm an intent.
So the reconfirmation threshold is used in two different ways depending on the existence of a confirmation sentence:
- With confirmation sentence: the reconfirmation threshold triggers the reconfirmation sentence
- Without confirmation sentence: the reconfirmation threshold confirms the input
If an Intent has been marked as Reconfirmation Needed and it is the highest found Intent, Cognigy AI will ask the user the question set in the
Confirmation Sentence property for the Intent. If the user confirms the question with a positive answer, Cognigy AI will remember the answer for this user and not ask the user again.
How Reconfirmation Works
- User: "I want to order a snack"
- Cognigy AI scores the Intent
- Cognigy AI: "Do you mean you want to order a pizza?"
- User: "Yes"
- Cognigy AI executes the sentence as if the user had said "I want to order pizza", plus remembers the confirmation for the future
If a certain number of users confirm the sentence, the sentence will be added to the list of example sentences trained to this Intent.
Updated about a year ago
|Example Sentence Annotation|