Interactive voice response system with a real time conversation scoring module

1. A method for interpreting sentiment of a conversation in real-time, the method comprising:

storing a plurality of sentiment signals;

receiving a first portion of the conversation, said conversation comprising a plurality of interaction pairs, each interaction pair comprising an utterance and a system response;

for each utterance included in each interaction pair, determining an utterance sentiment category, said determining comprising:

assigning a sentiment signal to each utterance, said assigned sentiment signal being selected from the plurality of sentiment signals; and

determining an utterance sentiment category for each utterance by executing a multi-class classifier operation on the assigned sentiment signal;

based on the identified utterance sentiment categories, classifying the first portion of the conversation in one of a plurality of conversation sentiment categories using a sequential module; and

outputting a conversation sentiment category that corresponds to the first portion of the conversation.

2. The method of claim 1 , wherein the first portion of the conversation is received, from a caller, at an interactive voice response system, and based on the conversation sentiment category, the caller is routed to a predetermined interactive response module.

3. The method of claim 2 , wherein the conversation sentiment category is very negative, negative, empathy needed or normal.

4. The method of claim 3 , wherein, when the conversation sentiment category is very negative, the predetermined interactive response module is a very negative response module.

5. The method of claim 3 , wherein, when the conversation sentiment category is negative, the predetermined interactive response module is a negative response module.

6. The method of claim 3 , wherein, when the conversation sentiment category is empathy needed, the predetermined interactive response module is an empathy needed response module.

7. The method of claim 3 , wherein, when the conversation sentiment category is normal, the predetermined interactive response module is a normal response module.

8. The method of claim 1 , wherein the sentiment signals comprise negation, rejection, mistake, urgent request, affirmation, empathy needed, repetition, total repetition, similar utterances, partial repetition, previous response, previous intent, previous intent status and previous input.

9. An interactive voice response system for interpreting sentiment of a conversation in real-time, the interactive voice response system comprising:

a receiver operable to receive:

a plurality of sentiment signals; and

a first portion of a conversation, said first portion of the conversation comprising a plurality of interaction pairs, each interaction pair comprising an utterance and a system response;

an utterance sentiment category determination module, said utterance sentiment category determination module comprising:

an assignment module, said assignment module operable to assign a subset of the plurality of sentiment signals to each utterance;

a multi-class classifier operation module, said multi-class classifier operation module operable to execute a multi-class classifier operation on each assigned subset of the plurality of sentiment signals; and

a sub-determination module that determines an utterance sentiment category for each utterance based on the execution of the multi-class classifier operation;

a sequential module operable to determine a conversation sentiment category based on the determined utterance sentiment categories and an order of the utterance sentiment categories; and

an output module operable to output the conversation sentiment category.

10. The interactive voice response system of claim 9 , wherein:

the receiver receives a second portion of the conversation, said second portion of the conversation comprising a subsequent utterance; and

the system further comprises a response module, and the response module determines a response to the subsequent utterance, the response being based at least in part on the conversation sentiment category.

11. The interactive voice response system of claim 9 , wherein the sequential module determines the conversation sentiment category based on weighing the positions of the determined utterance sentiment categories within the first portion of the conversation.

12. The interactive voice response system of claim 9 , further comprising a plurality of predetermined interactive response modules, wherein:

the first portion of the conversation is received from a caller;

based on the conversation sentiment category, the interactive voice response system routes the caller to a predetermined interactive response module, selected from the plurality of predetermined interactive response modules.

13. The method of claim 12 , wherein the conversation sentiment category is very negative, negative, empathy needed or normal.

14. The method of claim 13 , wherein, when the conversation sentiment category is very negative, the predetermined interactive response module is a very negative response module.

15. The method of claim 13 , wherein, when the conversation sentiment category is negative, the predetermined interactive response module is a negative response module.

16. The method of claim 13 , wherein, when the conversation sentiment category is empathy needed, the predetermined interactive response module is an empathy needed response module.

17. The method of claim 13 , wherein, when the conversation sentiment category is normal, the predetermined interactive response module is a normal response module.

18. The method of claim 9 , wherein the sentiment signals comprise negation, rejection, mistake, urgent request, affirmation, empathy needed, repetition, total repetition, similar utterances, partial repetition, previous response, previous intent, previous intent status and previous input.