Capital One Services, LLC

Kunlaya Soiaporn

Rui Zhang

Victor Alvarez Miranda

Erik Thomas Mueller

Scott Karp

Arturo Hernandez Zeledon

Vinay Igure

Pamela Katali

Kwan-Yuet Ho

Tanushree Luke

Methods and systems are described for generating dynamic conversational responses using two-tier machine learning models. The dynamic conversational responses may be generated in real time and reflect the likely goals and/or intents of a user. The two-tier machine learning model may include a first tier that determines an intent cluster based on a feature input, and a second tier that determines a specific intent from the cluster.

Systems and methods for a two-tier machine learning model for generating conversational responses

In certain embodiments, intent prediction and dialogue generation may be facilitated. In some embodiments, a chat initiation request may be obtained from a user. The latest activity information associated with the user may be provided to a prediction model to obtain a first set of predicted intents of the user. For each intent of the first set of predicted intents, a candidate question may be selected from a question set based on the candidate question matching the intent. In some embodiments, the candidate questions may be simultaneously presented on the chat interface.

Intent prediction for dialogue generation

In certain embodiments, pre-chat intent prediction and dialogue generation may be facilitated. In some embodiments, a chat initiation request may be obtained from a user. The latest activity information associated with the user may be provided to a prediction model to obtain a first set of predicted intents of the user. For each intent of the first set of predicted intents, a candidate question may be selected from a question set based on the candidate question matching the intent. Within ten seconds of the chat initiation request, the candidate questions may be simultaneously presented on the chat interface. Based on negative feedback obtained via the chat interface with respect to the candidate questions, additional candidate questions matching the second set of predicted intents may be presented on the chat interface. In some embodiments, the negative feedback may be provided to the prediction model to update configurations of the prediction model.

Pre-chat intent prediction for dialogue generation

In certain embodiments, pre-chat intent prediction and dialogue generation may be facilitated. In some embodiments, a chat initiation request may be obtained from a user. The latest activity information associated with the user may be provided to a prediction model to obtain predicted intents of the user. For each intent of the current intents, a candidate question may be selected from a question set based on the candidate question matching the intent. Within ten seconds of the chat initiation request, the candidate questions may be simultaneously presented on the chat interface. A user selection of a first question of the candidate questions may be obtained via the chat interface responsive to the presentation of the candidate questions. Based on the user selection and the first question matching a first intent of the predicted intents, the first intent may be provided as reference feedback for the prediction model.

Growing community of inventors

McLean, VA, United States of America

Kunlaya Soiaporn