System, method, and computer program for predicting at least one reason for a current call received from a customer

Patent metadata

Field	Value
Publication number	US-9307085-B1
Application number	US-201414192807-A
Country	US
Kind code	B1
Filing date	Feb 27, 2014
Priority date	Mar 18, 2013
Publication date	Apr 5, 2016
Grant date	Apr 5, 2016

Abstract

A system, method, and computer program product are provided for predicting at least one reason for a current call received from a customer. In operation, information associated with at least one transaction corresponding to at least one customer is identified. Additionally, relevant data associated with the at least one customer is determined, based on the at least one transaction. Further, at least one reason for a current customer call received from the at least one customer is predicted, based on the relevant data associated with the at least one customer.

First claim

What is claimed is:

1. A computer program product embodied on a non-transitory computer readable medium, comprising computer code for:
storing, by a system, information in a hardware database of the system that is associated with a plurality of transactions conducted by the customer, the information including order information, usage information, bill information, payment information, and historical call data detailing historical calls made by the customer, including for each call:
a reason for the customer's call,
issues addressed during the call,
a resolution of the call,
a duration of the call, and
a location of the customer during the call;

partitioning the stored information into a first subset and a second subset,
identifying a plurality of training algorithms,
for each of the plurality of training algorithms:
training a predictive model utilizing the training algorithm and the first subset of the stored information,
testing the trained predictive model by applying the trained predictive model to the second subset of the stored information,
evaluating a first plurality of prediction errors resulting from the application of the trained predictive model to the second subset of the stored information,
training the predictive model utilizing the training algorithm and the second subset of the stored information,
testing the trained predictive model by applying the trained predictive model to the first subset of the stored information,
evaluating a second plurality of prediction errors resulting from the application of the trained predictive model to the first subset of the stored information, and
combining the first and second plurality of prediction errors to create a combined number of prediction errors associated with the training algorithm;

selecting one of the plurality of training algorithms having a smallest combined number of associated prediction errors;
training by the system the predictive model using the selected training algorithm and the stored information; and
predicting by the system at least one reason for a current customer call received from the at least one customer, utilizing the predictive model.

2. The computer program product of claim 1 , further comprising computer code for receiving the current customer call from the at least one customer at a customer call center.

3. The computer program product of claim 1 , further comprising computer code for generating explanatory text including information about the at least one transaction that is relevant to the predicted at least one reason for the current customer call from the at least one customer.

4. The computer program product of claim 1 , further comprising computer code for determining an appropriate action for a call agent to take, based on the determined relevant data and the predicted at least one reason for the current customer call.

5. The computer program product of claim 1 , wherein the computer program product is operable such that the predictive model includes a probabilistic model capable of being utilized to predict the at least one reason for the current customer call received from the at least one customer.

6. The computer program product of claim 1 , wherein the computer program product is operable such that the predictive model includes a predictive model based on a Bayesian Belief Network.

7. The computer program product of claim 1 , wherein the computer program product is operable such that predicting the at least one reason for the current customer call received from the at least one customer includes predicting a most likely reason for the current customer call.

8. The computer program product of claim 1 , wherein the computer program product is operable such that predicting the at least one reason for the current customer call received from the at least one customer includes predicting a plurality of predicted reasons for the current customer call.

9. The computer program product of claim 1 , further comprising computer code for storing information associated with the current customer call placed by the at least one customer.

10. The computer program product of claim 1 , further comprising computer code for displaying the predicted at least one reason for the current customer call received from the at least one customer to a call agent utilizing a user interface.

11. The computer program product of claim 1 , further comprising computer code for displaying a plurality of predicted reasons for the current customer call received from the at least one customer to a call agent utilizing a user interface such that the call agent is able to proactively address any of the plurality of predicted reasons for the current customer call.

12. The computer program product of claim 1 , wherein the plurality of training algorithms include at least one expectation-maximization algorithm.

13. A method, comprising:
storing, by a system, information in a hardware database of the system that is associated with a plurality of transactions conducted by the customer, the information including order information, usage information, bill information, payment information, and historical call data detailing historical calls made by the customer, including for each call:
a reason for the customer's call,
issues addressed during the call,
a resolution of the call,
a duration of the call, and
a location of the customer during the call;

partitioning the stored information into a first subset and a second subset,
identifying a plurality of training algorithms,
for each of the plurality of training algorithms:
training a predictive model utilizing the training algorithm and the first subset of the stored information,
testing the trained predictive model by applying the trained predictive model to the second subset of the stored information,
evaluating a first plurality of prediction errors resulting from the application of the trained predictive model to the second subset of the stored information,
training the predictive model utilizing the training algorithm and the second subset of the stored information,
testing the trained predictive model by applying the trained predictive model to the first subset of the stored information,
evaluating a second plurality of prediction errors resulting from the application of the trained predictive model to the first subset of the stored information, and
combining the first and second plurality of prediction errors to create a combined number of prediction errors associated with the training algorithm;

selecting one of the plurality of training algorithms having a smallest combined number of associated prediction errors;
training by the system the predictive model using the selected training algorithm and the stored information; and
predicting by the system at least one reason for a current customer call received from the at least one customer, utilizing a processor of the system and the predictive model.

14. A system, comprising:
a memory system; and
one or more processing cores coupled to the memory system and that are each configured for:
storing, by a system, information in a hardware database of the system that is associated with a plurality of transactions conducted by the customer, the information including order information, usage information, bill information, payment information, and historical call data detailing historical calls made by the customer, including for each call:
a reason for the customer's call,
issues addressed during the call,
a resolution of the call,
a duration of the call, and
a location of the customer during the call;

partitioning the store

Assignees

• Amdocs Software Systems Ltd

Inventors

• Hanson Craig Dean
• Verego Greg
• O'Hara Joseph Patrick

Classifications

• H04M3/5235 Primary

  Dependent on call type or called number [DNIS] · CPC title

• H04M3/51

  Centralised call answering arrangements requiring operator intervention {, e.g. call or contact centers for telemarketing} · CPC title

• H04M3/5183 Primary

  Call or contact centers with computer-telephony arrangements · CPC title