Dialogue flow optimization and personalization

• US 10,455,088 B2
• Filed: 10/21/2015
• Issued: 10/22/2019
• Est. Priority Date: 10/21/2015
• Status: Active Grant

• Current Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Original Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Inventors: Tapuhi, Tamir, Konig, Yochai, Lev-Tov, Amir, Faizakof, Avraham, Lev, Yoni
• Primary Examiner(s): Bezuayehu, Solomon G
• Application Number:
  US14/919,675
• Publication Number:
  US 20170118336A1
• Time in Patent Office: 1,462 Days
• Field of Search: 379 8801- 8819
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: H04M 3/4938 comprising a voice browser which renders and interprets, e.g. VoiceXML
• CPC Class Codes: H04M 3/5166 in combination with interactive voice response systems or voice portals, e.g. as front-ends
• CPC Class Codes: H04M 7/0012 Details of application programming interfaces [API] for telephone networks; Arrangements which combine a telephonic communication equipment and a computer, i.e. computer telephony integration [CPI] arrangements

Abstract

A method for generating a dialog tree for an automated self-help system of a contact center from a plurality of recorded interactions between customers and agents of the contact center includes: computing, by a processor, a plurality of feature vectors, each feature vector corresponding to one of the recorded interactions; computing, by the processor, similarities between pairs of the feature vectors; grouping, by the processor, similar feature vectors based on the computed similarities into groups of interactions; rating, by the processor, feature vectors within each group of interactions based on one or more criteria, wherein the criteria include at least one of interaction time, success rate, and customer satisfaction; and outputting, by the processor, a dialog tree in accordance with the rated feature vectors for configuring the automated self-help system.

Claims

1. A method for generating a dialogue tree for an automated self-help system of a contact center, the method comprising:

recognizing, by a processor, speech in a plurality of recorded interactions between customers and agents of the contact center to generate recognized text, the recognized text comprising a plurality of phrases, the phrases being classified into a plurality of clusters, each cluster corresponding to an identifiable concept within the recorded interactions comprising a plurality of semantically similar phrases corresponding to the identifiable concept;

computing, by the processor, a plurality of feature vectors, each feature vector corresponding to one of the plurality of recorded interactions, each feature vector identifying one or more clusters corresponding to phrases found in the recognized text of the corresponding one of the recorded interactions;

computing, by the processor, similarities between pairs of the feature vectors based on a similarity threshold;

grouping, by the processor, similar feature vectors based on the computed similarities into groups of interactions;

rating, by the processor, feature vectors within each group of interactions based on one or more criteria, wherein the criteria include at least one of interaction time, success rate, and customer satisfaction; and

outputting, by the processor, a dialogue tree in accordance with the rated feature vectors for configuring the automated self-help system.

2. The method of claim 1, wherein the computing the similarity between a first feature vector and a second feature vector of the pair of feature vectors based on a similarity threshold comprises:

summing the number of clusters appearing in both the first feature vector and the second feature vector to initialize a similarity metric;

adjusting the similarity metric based on clusters appearing in only one of the pair of feature vectors; and

normalizing the similarity metric.

3. The method of claim 1, further comprising computing the success rate by:

identifying, by the processor, clusters corresponding to phrases indicating successful resolution of an issue associated with the interaction; and

computing, by the processor, the success rate based on the number of the recorded interactions that include clusters indicating successful resolution of the issue.

4. The method of claim 1, further comprising computing the customer satisfaction rate by:

associating each interaction with ratings from a survey; and

computing the customer satisfaction based on the ratings.

5. The method of claim 1, further comprising identifying a plurality of customer features associated with the customer associated with each of the recorded interactions.

6. The method of claim 5, further comprising:

partitioning the plurality of recorded interactions based on a plurality of groups of personalization features;

computing a dialogue tree for each of the groups of personalization features to compute a plurality of dialogue trees; and

outputting the plurality of dialogue trees for configuring the automated self-help system.

7. The method of claim 5, wherein the dialogue tree comprises a plurality of agent nodes arranged in a plurality of agent layers and a plurality of customer nodes arranged in a plurality of customer layers,

wherein at least one of the customer nodes is coupled to a plurality of successive agent nodes, and

wherein each of a plurality of edges between the at least one of the customer nodes and the successive agent nodes includes a condition corresponding to a group of personalization features.

8. The method of claim 1, further comprising outputting a plurality of suggestions for a successive node of a tree based on an input prior node and the rated feature vectors.

9. A system comprising

a processor;

and

a memory, wherein the memory stores instructions that, when executed by the processor, cause the processor to;

recognize speech in a plurality of recorded interactions between customers and agents of a contact center to generate recognized text, the recognized text comprising a plurality of phrases, the phrases being classified into a plurality of clusters, each cluster corresponding to an identifiable concept within the recorded interactions comprising a plurality of semantically similar phrases corresponding to the identifiable concept;

compute a plurality of feature vectors, each feature vector corresponding to one of the plurality of recorded interactions, each feature vector identifying one or more clusters corresponding to phrases found in the recognized text of the corresponding one of the recorded interactions;

compute similarities between pairs of the feature vectors based on a similarity threshold;

group similar feature vectors based on the computed similarities into groups of interactions;

rate feature vectors within each group of interactions based on one or more criteria, wherein the criteria include at least one of interaction time, success rate, and customer satisfaction; and

output a dialogue tree in accordance with the rated feature vectors for configuring an automated self-help system.

10. The system of claim 9, wherein the instructions that cause the processor to compute the similarity between a first feature vector and a second feature vector of the pair of feature vectors based on a similarity threshold include instructions that, when executed by the processor, cause the processor to:

sum the number of clusters appearing in both the first feature vector and the second feature vector to initialize a similarity metric;

adjust the similarity metric based on clusters appearing in only one of the pair of feature vectors; and

normalize the similarity metric.

11. The system of claim 9, wherein the instructions that cause the processor to compute the success rate include instructions that, when executed by the processor, cause the processor to:

identify clusters corresponding to phrases indicating successful resolution of an issue associated with the interaction; and

compute the success rate based on the number of the recorded interactions that include clusters indicating successful resolution of the issue.

12. The system of claim 9, wherein the instructions that cause the processor to compute the customer satisfaction rate include instructions that, when executed by the processor, cause the processor to:

associate each interaction with ratings from a survey; and

compute the customer satisfaction based on the ratings.

13. The system of claim 9, wherein the memory further stores instructions that, when executed by the processor, cause the processor to identify a plurality of customer features associated with the customer associated with each of the recorded interactions.

14. The system of claim 13, wherein the memory further stores instructions that, when executed by the processor, cause the processor to:

partition the plurality of recorded interactions based on a plurality of groups of personalization features;

compute a dialogue tree for each of the groups of personalization features to compute a plurality of dialogue trees; and

output the plurality of dialogue trees for configuring the automated self-help system.

15. The system of claim 13, wherein the dialogue tree comprises a plurality of agent nodes arranged in a plurality of agent layers and a plurality of customer nodes arranged in a plurality of customer layers,

wherein at least one of the customer nodes is coupled to a plurality of successive agent nodes, and

wherein each of a plurality of edges between the at least one of the customer nodes and the successive agent nodes includes a condition corresponding to a group of personalization features.

16. The system of claim 9, wherein the memory further stores instructions that, when executed by the processor, cause the processor to output a plurality of suggestions for a successive node of a tree based on an input prior node and the rated feature vectors.