Back to results
US 11,900,327 B2Granted
Evaluation adjustment factoring for bias
AssigneeCapital One Services, LLC
InventorsLin Ni Lisa Cheng, Vyjayanthi Vadrevu, Xiaoguang Zhu
Life of this patent
Estimating the term…
Abstract
In some implementations, a device may obtain interaction data for an evaluator in a communication session with an individual based on interactions of the evaluator with the individual during the communication session, wherein the evaluator is to provide the evaluation of the individual. The device may obtain evaluation information identifying the evaluation of the individual by the evaluator. The device may determine a degree of bias of the evaluator toward the individual during the communication session based on at least the interaction data. The device may adjust the evaluation of the individual based on the degree of bias of the evaluator.
Claims20
A system for adjusting an evaluation to factor for bias, the system comprising:
- one or more memories;
- and
- one or more processors, communicatively coupled to the one or more memories, configured to:
- train, using historical data, at least one of a first machine learning model or a second machine learning model associated with bias identification;
- monitor a communication session between an individual and an evaluator that is to provide the evaluation of the individual;
- generate interaction data for the evaluator based on interactions of the evaluator with the individual during the communication session,
- wherein the one or more processors, when generating the interaction data, are configured to input at least one of video data, image data, or audio data, obtained during the communication session, into the first machine learning model to cause the first machine learning model to output the interaction data, and
- wherein the interaction data identifies data associated with at least one of:
- a facial expression of the evaluator,
- a gesture of the evaluator,
- a speech cadence of the evaluator, or
- wording used by the evaluator;
- obtain evaluation information identifying the evaluation of the individual by the evaluator;
- determine indications of bias of the evaluator toward the individual during the communication session in connection with a plurality of categories associated with the individual,
- wherein the one or more processors, when determining the indications of bias of the evaluator, are configured to input the interaction data into the second machine learning model to cause the second machine learning model to output the indications of bias;
- and
- adjust the evaluation of the individual in connection with a plurality of metrics, wherein the evaluation is adjusted by a plurality of different amounts based on the indications of bias.
14 more elements
A method of adjusting an evaluation to factor for bias, comprising:
- obtaining, by a device, interaction data for an evaluator in a communication session with an individual based on interactions of the evaluator with the individual during the communication session,
- wherein the evaluator is to provide the evaluation of the individual, and
- wherein the interaction data identifies data associated with at least one of:
- a facial expression of the evaluator,
- a gesture of the evaluator,
- a speech cadence of the evaluator, or
- wording used by the evaluator;
- obtaining, by the device, evaluation information identifying the evaluation of the individual by the evaluator;
- training, by the device and by using historical data, a machine learning model associated with bias identification;
- determining, by the device and by using the machine learning model, indications of bias of the evaluator toward the individual during the communication session in connection with a plurality of categories associated with the individual and based on input of the interaction data into the machine learning model;
- and
- adjusting, by the device, the evaluation of the individual in connection with a plurality of metrics, wherein the evaluation is adjusted by a plurality of different amounts based on the indications of bias.
9 more elements
A non-transitory computer-readable medium storing a set of instructions for adjusting an evaluation to factor for bias, the set of instructions comprising:
- one or more instructions that, when executed by one or more processors of a device, cause the device to:
- train at least one of a first machine learning model or a second machine learning model associated with bias identification;
- monitor a telecommunication session between an individual and an evaluator that is to provide the evaluation of the individual;
- generate, based on inputting at least one of video data, image data, or audio data, obtained during the telecommunication session, into the first machine learning model, interaction data for the evaluator based on interactions of the evaluator with the individual during the telecommunication session,
- wherein the interaction data identifies data associated with at least one of:
- a facial expression of the evaluator,
- a gesture of the evaluator,
- a speech cadence of the evaluator, or
- wording used by the evaluator;
- obtain evaluation information identifying the evaluation of the individual by the evaluator;
- determine, based on inputting the interaction data and the evaluation information into the second machine learning model, a respective degree of bias of the evaluator toward the individual during the telecommunication session in connection with one or more categories associated with the individual;
- and
- adjust the evaluation of the individual proportionate to the respective degree of bias of the evaluator in connection with a particular metric, wherein the evaluation is adjusted by a particular amount based on the respective degree of bias.
10 more elements
Dependencies are inferred from the claim text.