-
Notifications
You must be signed in to change notification settings - Fork 3
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add abstract class for dialogue policy #238
base: main
Are you sure you want to change the base?
Conversation
Coverage Report •
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM.
The dialogue policy generates the next dialogue act of an agent based on the | ||
current dialogue state (last user's utterance and dialogue history). It defines | ||
the flow of the conversation, i.e., what steps an agent must take at every | ||
stage. The annotations of the dialogue act represent what the agent must elicit, | ||
recommend, or inform. The output of the dialogue policy is converted to a | ||
natural language response by the natural language generator. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@NoB0 Is this policy meant only for the conversational agent or could it also be used for a user simulator? In the latter case, this text needs to be updated.
For example, in the context of recommendation system, the dialogue act with | ||
intent ELICIT is generated if the agent does not store any user preferences. | ||
For the user intent REVEAL, the dialogue policy triggers the generation of an | ||
item recommendation (dialogue act with intent RECOMMEND). |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is not needed.
This method is most likely used in receive_utterance() in the agent to | ||
generate the agent's response. Dialogue act prediction is composed of | ||
two steps: intent prediction and annotation prediction. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is not needed (we don't need to make assumptions about how the agent is implemented).
Fixes #156