Systems and methods for optimization of interactive voice recognition systems

1. A method for optimization of interactive voice recognition (IVR) system processes, the method comprising:

receiving, from the IVR system, an IVR transaction log;

receiving a plurality of IVR menus;

extracting a plurality of IVR journeys from the IVR transaction log that correspond to the plurality of IVR menus;

transforming each of the plurality of IVR journeys and corresponding prompts into a vector based on a corresponding model that is specific to the particular IVR menu for the particular IVR journey to create a plurality of IVR journey vectors;

clustering the IVR journey vectors by:

for each menu in each cluster:

a) determining a mean number of occurrences, μ, of a particular menu in the plurality of IVR journeys,

b) determining a standard deviation, σ, of mean number of occurrences of a particular menu, μ, from the plurality of IVR journeys,

c) determining a mean number of occurrences, X, of a particular menu in the clustered IVR journey vectors,

d) labeling the particular cluster with three Z-score values, the three Z-score values are selected from all Z-score values for the IVR journeys with menus having the of a particular cluster of the menus IVR journeys within the cluster having the highest Z score; and

determining one or more desired optimization parameters based on the clustered IVR journey vectors.

2. The method of claim 1 wherein the IVR transaction log is received from the IVR system.

3. The method of claim 1 wherein the plurality of IVR menus are menus of interest.

4. The method of claim 1 wherein each corresponding model is based on a deep multimodal sequence auto-encoder (DMSA) that was trained with previous IVR transaction logs from the IVR system.

5. The method of claim 1 wherein the clustering is K-Means clustering.

6. The method of claim 5 wherein an optimal number of clusters is a value of K that maximizes a Silhouette metric.

7. The method of claim 1 wherein the one or more optimization parameters includes a containment rate.

8. One or more non-transitory computer-readable storage media comprising instructions that are executable to cause one or more processors to:

receive an IVR transaction log;

receive a plurality of IVR menus;

extract a plurality of IVR journeys from the IVR transaction log that correspond to the plurality of IVR menus;

transform each of the plurality of IVR journeys and corresponding prompts into a vector based on a corresponding model that is specific to the particular IVR menu for the particular IVR journey to create a plurality of IVR journey vectors;

cluster the IVR journey vectors by;

for each menu in each cluster:

a) determine a mean number of occurrences, μ, of a particular menu in the plurality of IVR journeys,

b) determine a standard deviation, σ, of mean number of occurrences of a particular menu, μ, from the plurality of IVR journeys,

c) determine a mean number of occurrences, X, of a particular menu in the clustered IVR journey vectors,

d) label the particular cluster with three Z-score values, the three Z-score values are selected from all Z-score values for the IVR journeys with menus having the of a particular cluster of the menus IVR journeys within the cluster having the highest Z score; and

determine one or more desired optimization parameters based on the clustered IVR journey vectors.

9. The one or more non-transitory computer-readable storage media of claim 8 where the instructions when executed further cause one or more processors to receive the IVR transaction log from the IVR system.

10. The one or more non-transitory computer-readable storage media of claim 8 wherein the plurality of IVR menus are menus of interest.

11. The one or more non-transitory computer-readable storage media of claim 8 wherein each corresponding model is based on a deep multimodal sequence auto-encoder (DMSA) that was trained with previous IVR transaction logs from the IVR system.

12. The one or more non-transitory computer-readable storage media of claim 8 wherein the clustering is K-Means clustering.

13. The one or more non-transitory computer-readable storage media of claim 8 wherein an optimal number of clusters is a value of K that maximizes a Silhouette metric.

14. The one or more non-transitory computer-readable storage media of claim 8 wherein the one or more optimization parameters includes a containment rate.