METHOD AND APPARATUS FOR FACILITATING VOICE USER INTERFACE DESIGN

• US 20150012829A1
• Filed: 06/30/2014
• Published: 01/08/2015
• Est. Priority Date: 07/02/2013
• Status: Active Grant

• Current Assignee: [24]7.ai, Inc.
• Original Assignee: 24/7 Customer Incorporated
• Inventors: SRIVASTAVA, Vaibhav, BROWN, Kathy L.
• Granted Patent: US 9,733,894 B2
• Time in Patent Office: Days
• US Class Current: 715/728
• CPC Class Codes: G06F 3/167 Audio in a user interface, e.g. using voice commands for navigating, audio feedback
• CPC Class Codes: G06F 40/20 Natural language analysis semantic analysis of natural language G06F40/30
• CPC Class Codes: G06F 8/38 for implementing user interfaces
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning

Abstract

A computer implemented method and an apparatus for facilitating voice user interface (VUI) design are provided. The method comprises identifying a plurality of user intentions from user interaction data. The method further comprises associating each user intention with at least one feature from among a plurality of features. One or more features from among the plurality of features are extracted from natural language utterances associated with the user interaction data. Further, the method comprises computing a plurality of distance metrics corresponding to pairs of user intentions from among the plurality of user intentions. A distance metric is computed for each pair of user intentions from among the pairs of user intentions. Furthermore, the method comprises generating a plurality of clusters based on the plurality of distance metrics. Each cluster comprises a set of user intentions. The method further comprises provisioning a VUI design recommendation based on the plurality of clusters.

Claims

1. A computer-implemented method, comprising:

identifying, by an apparatus, a plurality of user intentions from user interaction data;

associating, by the apparatus, each user intention from among the plurality of user intentions with at least one feature from among a plurality of features, wherein one or more features from among the plurality of features are extracted from natural language utterances associated with the user interaction data;

computing, by the apparatus, a plurality of distance metrics corresponding to pairs of user intentions from among the plurality of user intentions, wherein a distance metric is computed for each pair of user intentions from among the pairs of user intentions and, wherein the distance metric is indicative of a measure of similarity between user intentions in a corresponding pair of user intentions;

generating, by the apparatus, a plurality of clusters based on the plurality of distance metrics, wherein each cluster from among the plurality of clusters comprises a set of user intentions from among the plurality of user intentions; and

provisioning, by the apparatus, a voice user interface (VUI) design recommendation based on the plurality of clusters.

2. The method of claim 1, wherein the user interaction data comprises interaction data generated during interactions between a plurality of users and customer support representatives.

3. The method of claim 1, wherein a feature from among the one or more features extracted from the natural language utterances corresponds to one of a word, a bigram, a trigram, a n-gram and a part of speech (POS) tag.

4. The method of claim 3, wherein a feature from among the plurality of features corresponds to a feature-based rule.

5. The method of claim 1, further comprising:

constructing, by the apparatus, a category-feature matrix comprising a plurality of categories and the plurality of features, wherein each category from among the plurality of categories corresponds to a user intention from among the plurality of user intentions, and wherein the category-feature matrix is configured to facilitate a visualization of an association of the at least one feature from among the plurality of features with each category.

6. The method of claim 1, wherein the measure of similarity corresponds to a correlation measure between the at least one feature associated with each user intention in the corresponding pair of user intentions.

7. The method of claim 1, wherein the plurality of distance metrics is computed using at least one algorithm from among a cosine similarity based algorithm, a Pearson correlation based algorithm, a Spearman rank correlation based algorithm, a Kendall'"'"'s tau based algorithm, a city block distance based algorithm, a Euclidian distance based algorithm and a custom algorithm derived from business logic and designer intuition.

8. The method of claim 7, wherein two or more distance metrics are computed for each pair of user intentions using two or more algorithms from among the at least one algorithm and, wherein a weighted average of the two or more distance metrics is performed to compute the distance metric for each pair of user intentions.

9. The method of claim 1, wherein the plurality of clusters is generated using at least one clustering algorithm from among a K-means algorithm, a self-organizing map based algorithm, a self-organizing feature map based algorithm, a density-based spatial clustering algorithm and an optics clustering based algorithm.

10. The method of claim 1, wherein a value of a distance metric corresponding to each pair of user intentions in the set of user intentions is less than a first pre-defined threshold value.

11. The method of claim 10, further comprising:

identifying, by the apparatus, at least one user intention in the set of user intentions with associated values of distance metrics between the at least one user intention and remaining user intentions in the set of user intentions to be less than a second pre-defined threshold value, the second pre-defined threshold value being less than the first pre-defined threshold value; and

splitting the at least one user intention, by the apparatus, to generate two or more new user intentions.

12. The method of claim 1, further comprising:

facilitating addition of new user intentions to the set of user intentions till a pre-determined stopping condition is met.

13. The method of claim 1, further comprising:

arranging, by the apparatus, user intentions corresponding to one or more clusters from among the plurality of clusters in a hierarchical structure to configure a plurality of levels of user intentions for the one or more clusters.

14. The method of claim 13, wherein provisioning the VUI design recommendation comprises generating a visual representation of the plurality of clusters, wherein the one or more clusters from among the plurality of clusters are associated with a plurality of levels of user intentions.

15. The method of claim 14, wherein the VUI design recommendation is configured to support refinement in structure arrangement based on at least one of business logic and designer intuition.

16. The method of claim 14, wherein the VUI design recommendation is configured to facilitate an unambiguous design of a VUI for an interactive voice response (IVR) system.

17. The method of claim 16, further comprising:

provisioning, by the apparatus, appropriate dialog with prompts corresponding to the VUI design recommendation for facilitating interaction with a plurality of users.

18. The method of claim 17, further comprising:

receiving, by the apparatus, a request for interaction by a user from among the plurality of users over a communication network;

querying the user, by the apparatus, for determining a corresponding user intention; and

providing, by the apparatus, an optimized interaction journey experience based on the VUI design recommendation and the determined user intention.

19. An apparatus comprising:

at least one processor; and

a memory having stored therein machine executable instructions, that when executed by the at least one processor, cause the apparatus to;

identify a plurality of user intentions from user interaction data;

associate each user intention from among the plurality of user intentions with at least one feature from among a plurality of features, wherein one or more features from among the plurality of features are extracted from natural language utterances associated with the user interaction data;

compute a plurality of distance metrics corresponding to pairs of user intentions from among the plurality of user intentions, wherein a distance metric is computed for each pair of user intentions from among the pairs of user intentions and, wherein the distance metric is indicative of a measure of similarity between user intentions in a corresponding pair of user intentions;

generate a plurality of clusters based on the plurality of distance metrics, wherein each cluster from among the plurality of clusters comprises a set of user intentions from among the plurality of user intentions; and

provision a VUI design recommendation based on the plurality of clusters.

20. The apparatus of claim 19, wherein the user interaction data comprises interaction data generated during interaction between a plurality of users and customer support representatives.

21. The apparatus of claim 19, wherein a feature from among the one or more features extracted from the natural language utterances corresponds to one of a word, a bigram, a trigram, a n-gram and a POS tag.

22. The apparatus of claim 21, wherein a feature from among the plurality of features corresponds to a feature-based rule.

23. The apparatus of claim 19, wherein the apparatus is further caused to:

construct a category-feature matrix comprising a plurality of categories and the plurality of features, wherein each category from among the plurality of categories corresponds to a user intention from among the plurality of user intentions, and wherein the category-feature matrix is configured to facilitate a visualization of an association of the at least one feature from among the plurality of features with each category.

24. The apparatus of claim 19, wherein the measure of similarity corresponds to a correlation measure between the at least one feature associated with each user intention in the corresponding pair of user intentions.

25. The apparatus of claim 19, wherein the plurality of distance metrics is computed using at least one algorithm from among a cosine similarity based algorithm, a Pearson correlation based algorithm, a Spearman rank correlation based algorithm, a Kendall'"'"'s tau based algorithm, a city block distance based algorithm, a Euclidian distance based algorithm and a custom algorithm derived from business logic and designer intuition.

26. The apparatus of claim 25, wherein two or more distance metrics are computed for each pair of user intentions using two or more algorithms from among the at least one algorithm and, wherein a weighted average of the two or more distance metrics is performed to compute the distance metric for each pair of user intentions.

27. The apparatus of claim 19, wherein the plurality of clusters is generated using at least one clustering algorithm from among K-means algorithm, a self-organizing map based algorithm, a self-organizing feature map based algorithm, a density-based spatial clustering algorithm and an optics clustering based algorithm.

28. The apparatus of claim 19, wherein a value of a distance metric corresponding to each pair of user intentions in the set of user intentions is less than a first pre-defined threshold value.

29. The apparatus of claim 28, wherein the apparatus is further caused to:

identify at least one user intention in the set of user intentions with associated values of distance metrics between the at least one user intention and remaining user intentions in the set of user intentions to be less than a second pre-defined threshold value, the second pre-defined threshold value being less than the first pre-defined threshold value; and

split the at least one user intention to generate two or more new user intentions.

30. The apparatus of claim 29, wherein the apparatus is further caused to:

facilitate addition of new user intentions to the set of user intentions till a pre-determined stopping condition is met.

31. The apparatus of claim 19, wherein the apparatus is further caused to:

arrange user intentions corresponding to one or more clusters from among the plurality of clusters in a hierarchical structure to configure a plurality of levels of user intentions for the one or more clusters.

32. The apparatus of claim 31, wherein provisioning the VUI design recommendation comprises generating a visual representation of the plurality of clusters, wherein the one or more clusters from among the plurality of clusters are associated with a plurality of levels of user intentions.

33. The apparatus of claim 32, wherein the VUI design recommendation is configured to support refinement in structure arrangement based on at least one of business logic and designer intuition.

34. The apparatus of claim 32, wherein the VUI design recommendation is configured to facilitate an unambiguous design of a VUI for an IVR system.

35. The apparatus of claim 34, wherein the apparatus is further caused to:

provision appropriate dialog with prompts corresponding to the VUI design recommendation for facilitating interaction with a plurality of users.

36. The apparatus of claim 35, wherein the apparatus is further caused to:

receive a request for interaction by a user from among the plurality of users over a communication network;

query the user for determining a corresponding user intention; and

provide an optimized interaction journey experience based on the VUI design recommendation and the determined user intention.