System and method for analysis and adjustment of speech-enabled systems

1. A method for analyzing and adjusting the performance of a speech-enabled application, the speech-enabled application receiving user utterances and associating each user utterance with an action-object based on one or more salient terms in the user utterance that are associated with the action-object, comprising:

selecting a plurality of user utterances that were previously received by the speech-enabled application;

associating one of a plurality of action-objects with each of the selected user utterances;

for each action-object, determining a first percentage of the utterances associated with the action-object that include at least one of the salient action terms associated with the action-object and a second percentage of the utterances associated with the action-object that include at least one of the salient object terms associated with the action-object; and

if the first percentage does not exceed a selected threshold, adjusting the one or more salient action terms associated with the action-object; and

if the second percentage does not exceed a selected threshold, adjusting the one or more salient object terms associated with the action-object.

2. The method of claim 1 , wherein selecting a plurality of user utterances that were previously received by the speech-enabled application comprises selecting all user utterances that were received by the speech-enabled application during a selected period of time.

3. The method of claim 1 , wherein selecting a plurality of user utterances that were previously received by the speech-enabled application comprises selecting the last X user utterances received by the speech-enabled application, wherein X comprises an integer.

4. The method of claim 1 , further comprising transcribing the selected user utterances to allow for text searching for salient terms.

5. The method of claim 1 , wherein associating one of a plurality of action-objects with each selected user utterance comprises manually associating an action-object with each user utterance.

6. The method of claim 1 , wherein associating one of a plurality of action-objects with each selected user utterance comprises determining the action-object that was ultimately associated with each user utterance when the user utterance was received by the speech-enabled application.

7. The method of claim 1 , wherein associating one of a plurality of action-objects with each selected user utterance comprises:

generating a table including the selected user utterances and the action-object associated with each user utterance; and

sorting the table by action-object.

8. The method of claim 1 , wherein adjusting the one or more salient action terms associated with the action-object comprises:

determining what words most frequently occur in the user utterances associated with the action-object;

selecting a new salient action term to associate with the action-object based on the determination of the most frequently occurring words.

9. The method of claim 8 , further comprising disassociating one or more salient action terms previously associated with the action-object based on the determination of the most frequently occurring words.

10. The method of claim 1 , wherein the speech-enabled application comprises a call center.

11. Software for analyzing and adjusting the performance of a speech-enabled application, the speech-enabled application receiving user utterances and associating each user utterance with an action-object based on one or more salient terms in the user utterance that are associated with the action-object, the software embodied in computer-readable media and when executed operable to:

receive a selection of a plurality of user utterances that were previously received by the speech-enabled application;

associate one of a plurality of action-objects with each of the selected user utterances;

for each action-object, determine a first percentage of the utterances associated with the action-object that include at least one of the salient action terms associated with the action-object and a second percentage of the utterances associated with the action-object that include at least one of the salient object terms associated with the action-object;

if the first percentage does not exceed a selected threshold, adjust one or more salient action terms associated with the action-object; and

if the second percentage does not exceed a selected threshold, adjust one or more salient object terms associated with the action-object.

12. The software of claim 11 , wherein receiving a selection of a plurality of user utterances that were previously received by the speech-enabled application comprises receiving a selection of all user utterances that were received by the speech-enabled application during a selected period of time.

13. The software of claim 11 , wherein receiving a selection of user utterances that were previously received by the speech-enabled application comprises receiving a selection of the last X user utterances received by the speech-enabled application, wherein X comprises an integer.

14. The software of claim 11 , further operable to transcribe the selected user utterances to allow for text searching for salient terms.

15. The software of claim 11 , wherein associating one of a plurality of action-objects with each selected user utterance comprises receiving a manual association of an action-object with each user utterance.

16. The software of claim 11 , wherein associating one of a plurality of action-objects with each selected user utterance comprises determining the action-object that was ultimately associated with each user utterance when the user utterance was received by the speech-enabled application.

17. The software of claim 11 , wherein associating one of a plurality of action-objects with each selected user utterance comprises:

generating a table including the selected user utterances and the action-object associated with each user utterance; and

sorting the table by action-object.

18. The software of claim 11 , further operable to:

determine what words most frequently occur in the user utterances associated with the action-object; and

suggest one or more new salient action or object terms to associate with the action-object based on the determination of the most frequently occurring words.

19. The software of claim 18 , further operable to suggest one or more salient action or object terms previously associated with the action-object to disassociate with the action-object based on the determination of the most frequently occurring words.

20. The software of claim 11 , wherein adjusting one or more salient action or object terms associated with the action-object comprises receiving instructions from a user of the software and adjusting one or more salient action or object terms associated with the action-object based on the instructions.

21. The software of claim 11 , wherein the speech-enabled application comprises a call center.

22. A method for analyzing and adjusting the performance of a speech-enabled call center, the speech-enabled call center receiving user utterances and associating each user utterance with an action-object based on one or more salient action terms and one or more salient object terms in the user utterance that are associated with the action-object, comprising:

selecting a plurality of user utterances that were previously received by the speech-enabled call center;

transcribing the selected user utterances;

associating one of a plurality of action-objects with each of the selected user utterances;

for each action-object, determining a first percentage of the utterances associated with the action-object that include at least one of the salient action terms associated with the action-object and a second percentage of the utterances associated with the action-object that include at least one of the salient object terms associated with the action-object;

if the first percentage does not exceed a selected threshold, adjusting the one or more salient action terms associated with the action-object; and

if the second percentage does not exceed a selected threshold, adjusting the one or more salient object terms associated with the action-object.

23. A method for analyzing and adjusting the performance of a speech-enabled application, the speech-enabled application receiving user utterances and associating each user utterance with an action-object based on one or more salient terms in the user utterance that are associated with the action-object, comprising:

selecting a plurality of user utterances that were previously received by the speech-enabled application;

associating one of a plurality of action-objects with each of the selected user utterances;

for each action-object, determining the percentage of the utterances associated with the action-object that include at least one of the salient terms associated with the action-object; and

if the percentage does not exceed a selected threshold, adjusting the one or more salient terms associated with the action-object;

wherein the threshold is selected to maximize both the percentage of utterances associated with an action-object that include at least one of the salient terms associated with the action-object and the accuracy with which the speech-enabled application associates the user utterances with the action-object.

24. Software for analyzing and adjusting the performance of a speech-enabled application, the speech-enabled application receiving user utterances and associating each user utterance with an action-object based on one or more salient terms in the user utterance that are associated with the action-object, the software embodied in computer-readable media and when executed operable to:

receive a selection of a plurality of user utterances that were previously received by the speech-enabled application;

associate one of a plurality of action-objects with each of the selected user utterances;

for each action-object, determine the percentage of the utterances associated with the action-object that include at least one of the salient terms associated with the action-object; and

if the percentage does not exceed a selected threshold, adjust one or more salient terms associated with the action-object;

wherein the threshold is selected to maximize both the percentage of utterances associated with an action-object that include at least one of the salient terms associated with the action-object and the accuracy with which the speech-enabled application associates the user utterances with the action-object.

25. The method of claim 1 , wherein adjusting the one or more salient object terms associated with the action-object comprises:

determining what words most frequently occur in the user utterances associated with the action-object;

selecting a new salient object term to associate with the action-object based on the determination of the most frequently occurring words.

26. The method of claim 9 , further comprising disassociating one or more salient object terms previously associated with the action-object based on the determination of the most frequently occurring words.