Intent prediction for dialogue generation

1. A system for facilitating chat dialogue, the system comprising:

one or more processors programmed with computer program instructions that, when executed, cause operations comprising:

predicting, based on one or more actions of a user on a website, that the user will initiate a chat session with a chat application available via the website after the one or more actions of the user;

in response to the prediction of the initiation of the chat session, and before the initiation of the chat session, obtaining, via a neural network, predicted intents of the user based on activity information associated with the user, the predicted intents comprising a first set of predicted intents;

before the initiation of the chat session with the user, pre-generating questions related to the first set of predicted intents for the chat session; and

causing the questions to be presented on a user interface,

wherein causing the questions to be presented on the user interface comprises:

in response to a first confidence score associated with a first intent of the predicted intents and a second confidence score associated with a second intent of the predicted intents both satisfying a first confidence score threshold, selecting, for presentation on the user interface, first and second questions that respectively correspond to the first and second intents; and

in response to (i) the first confidence score associated with the first intent also satisfying a second confidence score threshold greater than the first confidence score threshold and (ii) the second confidence score not satisfying the second confidence score threshold, causing the first question to be presented to the user before the initiation of the chat session and not presenting the second question to the user until after the initiation of the chat session.

2. The system of claim 1 , wherein causing the questions to be presented on the user interface comprises, in response to a user selection of the first question presented to the user, causing the chat session to be initiated and the questions to be presented on the user interface.

3. The system of claim 1 , wherein causing the first question to be presented to the user before the initiation of the chat session comprises, in response to (i) the first confidence score associated with the first intent also satisfying the second confidence score threshold and (ii) the second confidence score not satisfying the second confidence score threshold, causing the first question to be presented on a sub-window of the user interface before the initiation of the chat session and not presenting the second question to the user in the sub-window.

4. A method comprising:

one or more processors programmed with computer program instructions that, when executed, cause operations comprising:

predicting, based on one or more actions of a user, that the user will initiate a dialogue session after the one or more actions of the user;

before the initiation of the dialogue session, obtaining, via a prediction model, predicted intents of the user based on (i) the prediction of the initiation of the dialogue session and (ii) activity information associated with the user, the predicted intents comprising a first set of predicted intents;

before the initiation of the dialogue session with the user, generating dialogue items related to the first set of predicted intents for the dialogue session; and

causing the dialogue items to be presented on a user interface,

wherein causing the dialogue items to be presented on the user interface comprises:

determining that a first confidence score associated with a first intent of the predicted intents satisfies a first confidence score threshold;

selecting, based the determination that the first confidence score associated with the first intent satisfies the first confidence score threshold, a first dialogue item corresponding to the first intent for presentation on the user interface;

determining that the first confidence score associated with the first intent also satisfies a second confidence score threshold greater the first confidence score threshold; and

causing, based the determination that the first confidence score associated with the first intent also satisfies the second confidence score threshold, the first dialogue item to be presented to the user before the initiation of the dialogue session.

5. The method of claim 4 , further comprising:

determining a user selection of the first dialogue item presented to the user,

wherein causing the dialogue items to be presented on the user interface comprises, based on the user selection of the first dialogue item, causing the dialogue session to be initiated and the dialogue items to be presented on the user interface.

6. The method of claim 4 , wherein causing the first dialogue item to be presented to the user before the initiation of the dialogue session comprises causing, based on the first confidence score associated with the first intent also satisfying the second confidence score threshold, the first dialogue item to be presented on a sub-window of the user interface before the initiation of the dialogue session.

7. The method of claim 4 , wherein obtaining the predicted intents comprises, in response to the prediction of the initiation of the dialogue session, and before the initiation of the dialogue session, providing the activity information associated with the user as input to the prediction model to obtain the predicted intents of the user.

8. The method of claim 4 , further comprising:

obtaining, via the user interface, a user selection of a given dialogue item of the dialogue items, the given dialogue item matching a given intent of the first set of predicted intents;

determining a feedback score associated with the given intent based on a probability category of the given intent such that (i) the feedback score is a first feedback score in response to the given intent being in a first probability category and (ii) the feedback score is a second feedback score in response to the given intent being in the second probability category; and

using, based on the user selection of the given dialogue item, the given intent and the feedback score to update one or more configurations of the prediction model.

9. The method of claim 4 , wherein the activity information associated with the user comprises page view information related to recent page views of the user.

10. The method of claim 4 , wherein the activity information associated with the user comprises service interaction information related to recent interactions of the user with one or more services or transaction information related to recent transactions of the user.

11. One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors, cause operations comprising:

one or more processors programmed with computer program instructions that, when executed, cause operations comprising:

predicting, based on one or more actions of a user, that the user will initiate a dialogue session after the one or more actions of the user;

before the initiation of the dialogue session, obtaining, via a prediction model, predicted intents of the user based on (i) the prediction of the initiation of the dialogue session and (ii) activity information associated with the user, the predicted intents comprising a first set of predicted intents;

before the initiation of the dialogue session with the user, generating dialogue items related to the first set of predicted intents;

upon the initiation of the dialogue session with the user, causing the dialogue items to be presented on a user interface;

obtaining, via the user interface, a user selection of a given dialogue item of the dialogue items, the given dialogue item matching a given intent of the first set of predicted intents;

determining a feedback score associated with the given intent based on a probability category of the given intent such that (i) the feedback score is a first feedback score in response to the given intent being in a first probability category and (ii) the feedback score is a second feedback score in response to the given intent being in the second probability category; and

using, based on the user selection of the given dialogue item, the given intent and the feedback score to update one or more configurations of the prediction model.

12. The media of claim 11 , the operations further comprising:

determining that a first confidence score associated with a first intent of the predicted intents satisfies a first confidence score threshold;

selecting, based the determination that the first confidence score associated with the first intent satisfies the first confidence score threshold, a first dialogue item corresponding to the first intent for presentation on the user interface;

determining that the first confidence score associated with the first intent also satisfies a second confidence score threshold greater the first confidence score threshold; and

causing, based the determination that the first confidence score associated with the first intent also satisfies the second confidence score threshold, the first dialogue item to be presented to the user before the initiation of the dialogue session.

13. The media of claim 12 , the operations further comprising:

determining a user selection of the first dialogue item presented to the user,

wherein causing the dialogue items to be presented on the user interface comprises, based on the user selection of the first dialogue item, causing the dialogue session to be initiated and the dialogue items to be presented on the user interface.

14. The media of claim 12 , wherein causing the first dialogue item to be presented to the user before the initiation of the dialogue session comprises causing, based on the first confidence score associated with the first intent also satisfying the second confidence score threshold, the first dialogue item to be presented on a sub-window of the user interface before the initiation of the dialogue session.

15. The media of claim 11 , wherein obtaining the predicted intents comprises, in response to the prediction of the initiation of the dialogue session, and before the initiation of the dialogue session, providing the activity information associated with the user as input to the prediction model to obtain the predicted intents of the user.

16. The media of claim 11 , wherein the activity information associated with the user comprises page view information related to recent page views of the user.

17. The media of claim 11 , wherein the prediction model comprises a neural network.