Periodic refresh of chatbot from virtual assistant timeline analytics of user conversations
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12210862-B2
Publication Date
2025-01-28
Expiration Date
Abstract
Aspects of the present disclosure relate generally to chatbot performance and, more particularly, to periodically refreshing chatbots from timeline analytics of user online conversations to improve performance. In embodiments, a method includes: receiving, by a computing device, a plurality of conversation transcripts generated from a plurality of versions of a chatbot; determining, by the computing device, a plurality of changes of a plurality of attributes of intents between the plurality of the versions of the chatbot; identifying, by the computing device, at least one intent to update from the plurality of changes of the plurality of attributes of the intents to improve performance of the chatbot; and generating, by the computing device, another version of the chatbot that includes the at least one intent updated from the plurality of changes of the plurality of attributes of the intents to improve the performance of the chatbot.
Core Innovation
A method receives, by a computing device, from a memory, a plurality of conversation transcripts generated from a plurality of versions of a chatbot. Each of the plurality of versions of the chatbot is trained using first training data comprising a plurality of attributes of intents, where each version is based on one or more variations in the plurality of attributes of intents. Second training data is received for at least one intent from a plurality of changes of the plurality of attributes of intents.
The computing device executes a training procedure to update a natural language classification model of the chatbot using the second training data. In response to executing the training procedure, the computing device generates another version of the chatbot having the updated natural language classification model that includes the at least one intent updated from the plurality of changes of the plurality of attributes of intents to improve performance of the chatbot. The another version of the chatbot is stored in the memory and deployed from the memory.
In related embodiments, the plurality of conversation transcripts are associated with a plurality of different time periods by a chatbot, where each different time period is associated with versions of the chatbot based on variations in the plurality of attributes of intents. Candidate intents are selected for update based on change in intent attributes, including intents with a greatest increase in change and intents with a greatest decrease in change, and the updated intents are included in a generated version of the chatbot for improved performance.
Claims Coverage
The document includes three independent claims. Across these claims, the coverage centers on receiving conversation transcripts from multiple chatbot versions or time periods, deriving second training data for selected intents based on changes in intent attributes, retraining a natural language classification model, generating an updated chatbot version including the updated intents, and deploying the updated version to improve performance.
Update a natural language classification model using second training data derived from intent attribute changes
receiving, by the computing device, second training data for at least one intent from a plurality of changes of the plurality of attributes of intents; executing, by the computing device, a training procedure to update a natural language classification model of the chatbot using the second training data
Generate and deploy another version of the chatbot including updated intent(s) to improve performance
generating, by the computing device, in response to executing the training procedure, another version of the chatbot having the updated natural language classification model that includes the at least one intent updated from the plurality of changes of the plurality of attributes of intents to improve performance of the chatbot; storing, by the computing device, the another version of the chatbot in the memory; and deploying, by the computing device from the memory, the another version of the chatbot
Training-and-deploy on a non-transitory computer-readable medium using intents with greatest increase and greatest decrease in change
receive, from the memory, second training data for at least one first intent from a plurality of intents with a greatest increase in change from a plurality of changes of the plurality of attributes of intents associated with the plurality of conversation transcripts from the plurality of different time periods by the chatbot and at least one second intent from the plurality of intents with a greatest decrease in change from the plurality of changes of the plurality of attributes of intents associated with the plurality of conversation transcripts from the plurality of different time periods by the chatbot; generate, by executing a training procedure that updates a natural language classification model of the chatbot using the second training data, a version of the chatbot that includes the at least one first intent and the at least one second intent updated from the plurality of intents with the greatest increase in change and the greatest decrease in change, respectively, to improve performance of the chatbot; store the version of the chatbot in the memory; and deploy, from the memory, the version of the chatbot
System for receiving transcripts from versions or time periods and deploying an updated chatbot version
receive, from a memory, a plurality of conversation transcripts generated from a plurality of different versions of a chatbot, each of the plurality of different versions of the chatbot is trained using first training data comprising a plurality of attributes of intents, wherein each of the plurality of different versions of the chatbot is based on one or more variations in the plurality of attributes of intents, wherein a plurality of changes of the plurality of attributes of intents are associated with the plurality of conversation transcripts from the plurality of different versions of the chatbot; receive, from the memory, second training data for at least one intent from the plurality of changes of the plurality of attributes of intents; execute a training procedure to update a natural language classification model of the chatbot using the second training data; generate, in response to executing the training procedure, another version of the chatbot having the updated natural language classification model that includes the at least one intent updated from the plurality of changes of the plurality of attributes of intents; store the another version of the chatbot in the memory; and deploy, from the memory, the another version of the chatbot
Overall, the independent claims cover updating a chatbot by retraining a natural language classification model using second training data tied to changes in intent attributes derived from conversation transcripts associated with multiple chatbot versions or time periods, then generating and deploying an updated chatbot version that includes selected updated intents to improve performance.
Stated Advantages
Improve performance of the chatbot.
Documented Applications
Periodically refreshing a chatbot using timeline analytics of user conversation transcripts across different chatbot releases or time periods, selecting intents for update based on change in intent attributes, and deploying a refreshed chatbot version.
Interested in licensing this patent?