System and method for discovering and exploring concepts and root causes of events

• US 10,061,822 B2
• Filed: 07/26/2013
• Issued: 08/28/2018
• Est. Priority Date: 07/26/2013
• Status: Active Grant

• Current Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Original Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Inventors: Lev-Tov, Amir, Faizakof, Avraham, Ollinger, David, Konig, Yochai
• Primary Examiner(s): Truong, Dennis
• Application Number:
  US13/952,470
• Publication Number:
  US 20150032746A1
• Time in Patent Office: 1,859 Days
• Field of Search: 707999006, 707999003, 707778, 707737, 704 9, 37926506, 706 12, 706 47, 715254
• CPC Class Codes: G06F 16/243 Natural language query formulation
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence

Abstract

A method for determining a cause of events detected in a plurality of interactions includes: identifying, on a processor, a plurality of elements in the interactions; detecting, on the processor, a plurality of sequences of elements in the interactions; mining, on the processor, the plurality of sequences for generating a set of supported patterns; computing, on the processor, association rules from the set of supported patterns; and returning the computed association rules.

Claims

1. A method for determining a cause of events detected in a plurality of interactions between customers and agents of a contact center, the method comprising:

identifying, on a processor, a plurality of elements in the interactions, each element of the elements corresponding to a cluster of semantically related sentences;

detecting, on the processor, a plurality of sequences of elements in the interactions;

mining, on the processor, the plurality of sequences for generating a set of supported patterns, the mining comprising, for each sequence of the plurality of sequences;

computing a first metric for the sequence in accordance with the number of times the sequence appears in the interactions;

comparing the computed first metric with a first threshold; and

adding the sequence to the set of supported patterns in response to the computed first metric being greater than the first threshold;

condensing, on the processor, the set of supported patterns by removing sequences having elements repeated within the sequence to generate a set of condensed supported patterns, each of the condensed supported patterns corresponding to one or more supported patterns of the set of supported patterns with one or more elements removed, wherein the first metric of each of the condensed supported patterns is greater than or equal to the first metrics of each of the corresponding one or more supported patterns;

automatically analyzing, on the processor, the set of condensed supported patterns to generate association rules; and

returning the computed association rules for identifying the cause of an event of the events detected in the plurality of interactions.

2. The method of claim 1, wherein the elements comprise defined topics and identified concepts.

3. The method of claim 2, wherein the identified concepts are derived by:

filtering, on the processor, the interactions based on intervals;

creating, on the processor, a plurality of sentences from the filtered interactions;

computing, on the processor, a saliency of each the sentences;

pruning away, on the processor, sentences with low saliency for generating a set of informative sentences;

clustering, on the processor, the sentences of the set of informative sentences for generating a plurality of sentence clusters, each of the clusters corresponding to a concept of the concepts;

computing, on the processor, a saliency of each of the clusters; and

naming, on the processor, each of the clusters.

4. The method of claim 1, wherein the detecting the sequences of elements comprises:

sorting the identified elements in each interaction by timestamp within the interaction for generating the plurality of sequences; and

condensing each of the sequences for removing repeated consecutive elements.

5. The method of claim 1, wherein the computing the association rules from the set of condensed supported patterns comprises:

computing an association rule for each sequence pattern in the condensed supported patterns, the computing the association rule comprising, in response to the sequence pattern comprising more than one element;

dividing the sequence pattern into a first portion and a second portion, the second portion comprising the last element in the sequence pattern;

generating a proposed association rule, the proposed association rule being a logical implication from the first portion to the second portion;

computing a confidence of the proposed association rule;

in response to the computed confidence being greater than a threshold confidence level, storing the proposed association rule and moving the last element in the first portion to the second portion and iteratively repeating generating the proposed association rule and computing the confidence; and

in response to the computed confidence being less than a threshold confidence level, ending analysis of the sequence pattern and returning a previously stored proposed association rule as the association rule or returning no association rule in response to no such proposed association rules being stored.

6. The method of claim 5, further comprising computing a lift and a saliency of the association rule.

7. The method of claim 5, further comprising computing a lift and a saliency of the proposed association rule.

8. A method for determining a root cause of an event detected in a plurality of interactions, the method comprising:

receiving, on a processor, a query for the root cause of the event;

searching, on the processor, a plurality of association rules, each of the association rules comprising one or more first portion elements and one or more second portion elements, each of the first portion elements and each of the second portion elements corresponding to a cluster of semantically related sentences and each of the first portion elements and each of the second portion elements appearing once in the corresponding association rule, each of the association rules being a logical implication from the first portion to the second portion, each of the association rules being automatically computed, by the processor, from sequences of events automatically detected, by the processor, in a plurality of interactions between customers and agents of a contact center, the first portion and the second portion being among the detected events, each of the sequences of events used to generate the association rules being mined from the interactions and condensed to remove sequences having elements repeated in the sequence, the sequences having a corresponding first metric exceeding a first threshold value, the first metric being computed in accordance with the number of times the corresponding sequence of the sequences appears in the interactions; and

returning, from the processor, one or more association rules matching the query for identifying the root cause of the event, wherein the second portion elements of each of the matching association rules comprise the event, the one or more association rules being sorted in accordance with at least one of a confidence value, a lift value, or a saliency value.

9. The method of claim 8, wherein each of the association rules has a corresponding confidence value,

wherein the query comprises a confidence threshold, and

wherein the returning the one or more association rules matching the query further comprises returning association rules matching the query that have confidence values exceeding the confidence threshold.

10. The method of claim 8, wherein each of the association rules has a corresponding lift value and a corresponding saliency value,

wherein the query comprises at least one of a lift threshold and a saliency threshold, and

wherein the returning the one or more association rules matching the query further comprises returning association rules matching the query that have lift values or saliency values exceeding the lift threshold or the saliency threshold.

11. A computer system comprising:

a processor; and

a memory, wherein the memory has stored thereon instructions that, when executed by the processor, causes the processor to compute association rules between events detected in a plurality of interactions between customers and agents of a contact center by;

identifying a plurality of elements in the interactions, each element of the elements corresponding to a cluster of semantically related sentences;

detecting a plurality of sequences of elements in the interactions;

mining the plurality of sequences for generating a set of supported patterns, the mining comprising, for each sequence of the plurality of sequences;

computing a first metric for the sequence in accordance with the number of times the sequence appears in the interactions;

comparing the computed first metric with a first threshold; and

adding the sequence to the set of supported patterns in response to the computed first metric being greater than the first threshold;

condensing the set of supported patterns by removing sequences having elements repeated within the sequence to generate a set of condensed supported patterns, each of the condensed supported patterns corresponding to one or more supported patterns of the set of supported patterns with one or more elements removed, wherein the first metric of each of the condensed supported patterns is greater than or equal to the first metrics of each of the corresponding one or more supported patterns;

automatically analyzing the set of condensed supported patterns to generate association rules; and

returning the computed association rules for identifying a cause of an event of the events detected in the plurality of interactions.

12. The computer system of claim 11, wherein the elements comprise defined topics and identified concepts.

13. The computer system of claim 12, wherein the identified concepts are derived by:

filtering, on the processor, the interactions based on intervals;

creating, on the processor, a plurality of sentences from the filtered interactions;

computing, on the processor, a saliency of each the sentences;

pruning away, on the processor, sentences with low saliency for generating a set of informative sentences;

clustering, on the processor, the sentences of the set of informative sentences for generating a plurality of sentence clusters, each of the clusters corresponding to a concept of the concepts;

computing, on the processor, a saliency of each of the clusters; and

naming, on the processor, each of the clusters.

14. The computer system of claim 11, wherein the processor is further configured to compute association rules between events detected in a plurality of interactions by:

sorting the identified elements in each interaction by timestamp within the interaction for generating the plurality of sequences; and

condensing each of the sequences for removing repeated consecutive elements.

15. The computer system of claim 11, wherein the computing the association rules from the set of condensed supported patterns comprises:

computing an association rule for each sequence pattern in the condensed supported patterns, the computing the association rule comprising, in response to the sequence pattern comprising more than one element;

dividing the sequence pattern into a first portion and a second portion, the second portion comprising the last element in the sequence pattern;

generating a proposed association rule, the proposed association rule being a logical implication from the first portion to the second portion;

computing a confidence of the proposed association rule;

in response to the computed confidence being greater than a threshold confidence level, storing the proposed association rule and moving the last element in the first portion to the second portion and iteratively repeating generating the proposed association rule and computing the confidence; and

in response to the computed confidence being less than a threshold confidence level, ending analysis of the sequence pattern and returning a previously stored proposed association rule as the association rule or returning no association rule if no such proposed association rules were stored.

16. The computer system of claim 15, wherein the computing the association rule further comprises computing a lift and a saliency of the association rule.

17. The computer system of claim 15, wherein the computing the association rule further comprises computing a lift and a saliency of the proposed association rule.

18. A computer system comprising:

a processor; and

a memory, wherein the memory has stored thereon instructions that, when executed by the processor, causes the processor to respond to a query for the root cause of an event by;

receiving the query for the root cause of the event;

searching a plurality of association rules, each of the association rules comprising one or more first portion elements and one or more second portion elements, each of the first portion elements and each of the second portion elements corresponding to a cluster of semantically related sentences and each of the first portion elements and each of the second portion elements appearing once in the corresponding association rule, each of the association rules being a logical implication from the first portion to the second portion, each of the association rules being automatically computed, by the processor, from sequences of events automatically detected, by the processor, in a plurality of interactions between customers and agents of a contact center, the first portion and the second portion being among the detected events, each of the sequences of events used to generate the association rules being mined from the interactions and condensed to remove sequences having elements repeated in the sequence, the sequences having a corresponding first metric exceeding a first threshold value, the first metric being computed in accordance with the number of times the corresponding sequence of the sequences appears in the interactions; and

returning one or more association rules matching the query for identifying the root cause of the event, wherein the second portion elements of each of the matching association rules comprise the event, the one or more association rules being sorted in accordance with at least one of a confidence value, a lift value, or a saliency value.