Process for automated deployment of natural language

• US 8,027,457 B1
• Filed: 12/01/2006
• Issued: 09/27/2011
• Est. Priority Date: 12/01/2005
• Status: Active Grant

• Current Assignee: Concentrix CVG Customer Management Delaware LLC (Synnex Corporation)
• Original Assignee: CMG Utah, Inc.
• Inventors: Coy, Cordell, Combs, Joseph
• Primary Examiner(s): Matar; Ahmad
• Assistant Examiner(s): Pope; Kharye
• Application Number:
  US11/565,702
• Time in Patent Office: 1,761 Days
• Field of Search: 370/401, 370/352, 704/244, 704246-257, 705/7, 379/265.09, 379/265, 379/266
• US Class Current: 379/265.09
• CPC Class Codes: H04M 2203/355 Interactive dialogue design tools, features or methods
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

There is disclosed a system and method for monitoring and updating an interactive voice response system through the use of a human agent who reviews the responses provided by a computer system and makes recommendations which are then used as data for updating the interactive voice response system. There is further disclosed a user interface for allowing the human agent to review the responses provided by the computer system, and to provide alternate responses which could be implemented by the computer system.

Claims

1. A method comprising the steps of:

a) receiving a statement during an interaction with an automated response system;

b) sending the statement to an agent for response;

c) receiving;

i) a first response to the statement, wherein the first response to the statement is determined by the agent based on the statement; and

ii) a second response to the statement, wherein the second response to the statement is determined automatically based on the statement;

d) implementing the first response to the statement

e) storing data comprising the first response and the second response;

f) terminating the interaction with the automated response system; and

g) subsequent to termination of the interaction, determining whether to identify the stored data based at least in part on the second response, wherein the second response was not implemented in response to the statement.

2. The method of claim 1 wherein the first response is determined by a live agent;

wherein the statement is received from a caller; and

wherein the second response is determined automatically;

wherein the second response is not implemented.

3. The method of claim 1 wherein storing data comprises storing data in a database comprising a plurality of records having a first field corresponding to the first response and a second field corresponding to the second response and wherein selectively identifying comprises querying the database.

4. The method of claim 3 wherein the querying step comprises selecting records in which the first field is different from the second field.

5. The method of claim 1 wherein selectively identifying takes place automatically.

6. The method of claim 1 wherein selectively identifying the stored data comprises altering the stored data.

7. The method of claim 6 wherein altering the stored data comprises moving the stored data.

8. The method of claim 6 wherein altering the stored data comprises augmenting the stored data.

9. A method comprising the steps of:

a) receiving a statement from a caller during an interaction between the caller and an automated response system;

b) involving an agent in the interaction, wherein involving the agent comprises transparently conferencing the agent into the interaction such that the caller is not aware of the agent'"'"'s involvement in the interaction;

c) simultaneously providing the statement to the agent and to an automated speech recognizer during the interaction between the caller and the automated response system;

d) receiving;

1) a first interpretation of the statement from the agent;

2) a second interpretation of the statement from the automated speech recognizer;

e) implementing a response to the statement, wherein the response to the statement is determined using either the first interpretation of the statement or the second interpretation of the statement;

f) storing data comprising the first interpretation of the statement and the second interpretation of the statement;

g) terminating the interaction between the caller and the automated response system; and

h) subsequent to termination of the interaction between the caller and the automated response system, determining whether to identify the stored data based on information comprising an identifying interpretation of the statement, wherein the identifying interpretation is either the first interpretation of the statement or the second interpretation of the statement, and wherein the identifying interpretation was not used in determining the response to the statement.

10. The method of claim 9 wherein involving the agent comprises muting an audio input.

11. The method of claim 9 wherein involving the agent comprises selecting the agent from an agent list.

12. The method of claim 9 wherein:

a) the agent is a live agent; and

b) in the event the first interpretation of the statement and the second interpretation of the statement are not the same, the response is determined using the second interpretation of the statement.

13. The method of claim 9 wherein:

a) determining whether to identify the stored data comprises determining whether the first interpretation of the statement is different from the second interpretation of the statement;

b) the stored data is identified if the first interpretation of the statement is different from the second interpretation of the statement; and

c) if the stored data is identified, it is used to update the automated speech recognizer.

14. A system comprising:

a) an agent terminal;

b) an automated interaction system;

c) a reusable intermediary system;

d) a computerized input interpreter; and

e) a computer readable medium having stored thereon computer executable instructions operable to;

configure the reusable intermediary system to allow an agent to log into the reusable intermediary system using the agent terminal;

configure the automated interaction system such that, in response to an external input, the automated interaction system sends a request to the reusable intermediary system for allocation of a resource for processing the external input;

configure the reusable intermediary system such that, in response to the request from the automated interaction system, the reusable intermediary system integrates the agent into processing the external input;

configure the automated interaction system to send a response to the external input both to the source of the external input and, through the reusable intermediary system, to the agent terminal;

configure the automated interaction system to send a second external input having the same source as the first external input to the computerized input interpreter, and to the agent terminal via the reusable intermediary system;

configure the computerized input interpreter to send, to the automated interaction system, a first output generated in response to the second external input;

configure the automated interaction system to send the first output to the agent terminal via the reusable intermediary system;

configure the reusable intermediary system and the agent terminal such that the agent terminal allows the agent to, using the agent terminal, determine a second output based on the second external input wherein the agent terminal and reusable intermediary system are configured such that the agent is allowed, using the agent terminal, to determine the second output before the first output is sent to the agent terminal, and such that the second output is sent from the agent terminal to the automated interaction system via the reusable intermediary system;

configure the automated interaction system to send a second response to the source of the external input based on the first output and the second output; and

if the first output and the second output are different, update the computerized input interpreter based at least in part on the second output.

15. A method of operating a system comprising an IVR, an agent terminal, a reusable Wizard of Oz system, and an automatic speech recognizer comprising the steps of:

a) allowing an agent to log into the reusable Wizard of Oz system using the agent terminal;

b) receiving a call from a caller at the IVR;

c) in response to a request from the IVR, bridging the agent to the call;

d) from the IVR, playing a prompt for the caller;

e) sending a description of the prompt from the IVR to the agent via the reusable

Wizard of Oz system and the agent terminal;

f) receiving, at the IVR, a statement from the caller;

g) from the IVR, sending the statement to the automatic speech recognizer;

h) from the automatic speech recognizer, sending an interpretation of the statement to the IVR;

i) from the IVR, sending the interpretation of the statement to the agent via the reusable Wizard of Oz system and the agent terminal;

j) allowing the agent to determine an agent interpretation of the statement via the agent terminal;

k) sending the agent interpretation of the statement from the agent terminal to the IVR via the reusable Wizard of Oz system;

l) based on the interpretation of the statement, determining a response to the statement from the caller;

m) via the IVR, responding to the statement from the caller using the determined response;

n) via the reusable Wizard of Oz system, automatically saving the interpretation of the statement and the agent interpretation of the statement;

o) if the interpretation of the statement and the agent interpretation of the statement are different, updating the automatic speech recognizer with a model based at least in part on the agent interpretation of the statement.