Iterative intent building utilizing dynamic scheduling of batch utterance expansion methods

A system for expansion of intent classification data utilizing batch utterance scheduling, comprising:

• a processor executing instructions stored in a memory device;
• wherein the processor:
• receives a set of unlabeled examples for intent processing by an intent builder iteratively defining an intent;
• executes machine learning logic to separate the set of examples into a first subset processed according to a first model and a second subset processed according to a second model;
• incorporates the first subset processed according to the first model into the intent builder during a building iteration and scheduling a first batch processing of the second subset processed according to the second model based on a scheduling criteria, wherein decisions taken by the intent builder on examples from the first subset are used as input during the first batch processing of the second subset;
• responsive to determining the scheduling criteria is satisfied, initiates the first batch processing of the second subset;
• and
• executing the machine learning logic to, responsive to detecting a completion of the first batch processing, utilize results from the completion to influence additional examples retrieved from the first subset and the second subset during a subsequent building iteration by the intent builder, wherein decisions taken by the intent builder on examples from the second subset, subsequent to the first batch processing, are used as input for retrieving the additional examples from the first subset.

A computer program product for expansion of intent classification data utilizing batch utterance scheduling, the computer program product comprising a non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising:

• an executable portion that receives a set of unlabeled examples for intent processing by an intent builder iteratively defining an intent;
• an executable portion that executes machine learning logic to separate the set of examples into a first subset processed according to a first model and a second subset processed according to a second model;
• an executable portion that incorporates the first subset processed according to the first model into the intent builder during a building iteration and scheduling a first batch processing of the second subset processed according to the second model based on a scheduling criteria, wherein decisions taken by the intent builder on examples from the first subset are used as input during the first batch processing of the second subset;
• an executable portion that, responsive to determining the scheduling criteria is satisfied, initiates the first batch processing of the second subset;
• and
• an executable portion that executes the machine learning logic to, responsive to detecting a completion of the first batch processing, utilize results from the completion to influence additional examples retrieved from the first subset and the second subset during a subsequent building iteration by the intent builder, wherein decisions taken by the intent builder on examples from the second subset, subsequent to the first batch processing, are used as input for retrieving the additional examples from the first subset.