Method and apparatus for detection and analysis of first contact resolution failures

• US 10,242,330 B2
• Filed: 11/06/2012
• Issued: 03/26/2019
• Est. Priority Date: 11/06/2012
• Status: Active Grant

• Current Assignee: Nice Ltd
• Original Assignee: Nice Systems Limited (Nice Ltd)
• Inventors: Daya, Ezra, Hurvitz, Eyal
• Primary Examiner(s): Padot, Timothy
• Assistant Examiner(s): Garcia-Guerra, Darlene
• Application Number:
  US13/669,467
• Publication Number:
  US 20140129299A1
• Time in Patent Office: 2,331 Days
• Field of Search: 705 711, 705 712, 705 732, 705 741
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 10/0633 Workflow analysis
• CPC Class Codes: G06Q 10/0639 Performance analysis of employees; Performance analysis of enterprise or organisation operations
• CPC Class Codes: G06Q 10/06393 Score-carding, benchmarking or key performance indicator [KPI] analysis
• CPC Class Codes: G06Q 10/06395 Quality analysis or management

Abstract

The subject matter discloses a method for detection and analysis of first contact resolution failures comprising: obtaining a first plurality of digital representations, each digital representation of the first plurality is associated with a contact center interaction; determining a pair of digital representations that represent a pair of contact center interactions determined from the first plurality of digital representations according to metadata associated with the pair of contact center interactions; determining topics of each interaction of the determined pair of contact center interactions represented by the pair of digital representations; classifying the pair of contact center interactions as first contact resolution failure or first contact resolution interactions pair.

Claims

1. A method for detecting first contact resolution failures, comprising:

using a processor, converting audio interactions into textual transcripts and textual interactions or receiving textual transcripts from an e-mail server, chat server or social media server,

storing in a training corpus the converted or received textual transcripts as pairs of interaction digital representations, each pair is tagged as First Contact Resolution Failure (FCRF) or First Contact Resolution (FCR) interactions pair;

generating and storing in a storage device a topic model including a plurality of weight scores associated with corresponding key phrases, by detecting topics and corresponding sets of phrases in the interaction digital representations;

generating and storing in the storage device a First Contact Resolution Failures (FCRF) model by calculating based on the stored topic model a topic similarity score for each of the stored tagged pairs of interactions, and by executing a training module with the topic similarity score and parameters of each of the stored tagged pairs of interactions as input;

obtaining by a capturing apparatus a first plurality of digital representations, each digital representation of the first plurality is associated with a contact center interaction;

from the first plurality of digital representations, determining that a first and a second of the plurality of digital representations is a pair of digital representations that represent a pair of contact center interactions according to predefined rules applied on metadata associated with the pair of the contact center interactions, and associating the first and second interactions as candidates for contact resolution classification;

applying the stored FCRF model for classification of the pair of contact center interactions candidates;

calculating an FCRF classification score representing the probability that a customer issue was not resolved within the first interaction of the pair of interactions, and handling of the same customer issue continued in the second interaction of the pair of interactions;

comparing the FCRF classification score to a predefined threshold, thereby classifying the pair of contact center interactions candidates as a first contact resolution failure (FCRF) interaction pair if the FCRF classification score is higher than the predefined threshold or classifying the pair of contact center interactions candidates as a first contact resolution (FCR) interactions pair if the FCRF classification score is lower than the predefined threshold; and

generating a report including a list of pairs of interactions that were classified as FCRF pairs of interactions;

wherein the classification of the determined pair of contact center interactions comprises;

determining based on the stored topic model a topic vector for each digital representation of the determined pair of digital representations; and

determining a topic similarity score between the topic vectors associated with the interactions of the determined pair of digital representations.

2. The method according to claim 1, wherein each digital representation of the first plurality represents an audio interaction or a textual interaction.

3. The method according to claim 1, wherein classifying the determined pair of contact center interactions comprises:

obtaining external data associated with each digital representation of the determined pair of digital representations.

4. The method according to claim 1, further comprises:

grouping classified pairs of interactions of contact center interactions to sequences of interactions; and

determining an FCR score for each interaction sequence.

5. The method according to claim 4, further comprises:

storing the FCR scores that are associated with the sequences of interactions.

6. The method according to claim 1, further comprises:

storing the classification of the pair of contact center interactions.

7. The method according to claim 1, further comprises:

enabling a search of the classified pairs of interactions based on the FCRF classification score.

8. The method according to claim 1, further comprises enabling a search of the sequences of interactions based on sequences of interactions FCR scores.

9. The method according to claim 1, further comprises:

generating a report based on pairs of interactions and their associated FCRF classification scores.

10. The method according to claim 1, further comprises:

generating a report based on sequences of interactions and FCR scores of said sequences of interactions.

11. The method according to claim 1, further comprises obtaining a second plurality of digital representations each digital representation is associated with a contact center interaction;

determining a topic model for the second plurality of digital representations;

determining a pair of digital representations that represent a pair of contact center interactions determined from the second plurality of digital representations according to metadata that is associated with the interactions;

tagging each pair of digital representations;

determining topic vectors for each digital representation of the pairs of digital representations;

determining topic similarity scores between each two topic vectors associated with the pairs of digital representations;

determining mutual topic vectors for each digital representation of the pairs of digital representations; and

generating an FCRF model based on the topic similarity scores and/or the mutual topic vectors and/or obtained external data that is associated with the interactions.

12. The method according to claim 11, wherein tagging of each pair of digital representations is based on pattern-based information extraction.

13. The method according to claim 1, wherein determining a topic similarity score is determined according to Kullback-Leibler divergence function.

19. The method according to claim 1, further wherein determining a topic similarity score is determined according to the formula:

20. The method according to claim 1, further comprising determining a mutual topic vector according to the formula:

21. The method according to claim 1, further comprising generating a classification vector according to the formula CV(n):

{[TS(n), MV(n), ED(n)]},

wherein;

n represents the n-th pair of interactions,

TS(n) represents the topic similarity of the n-th pair,

MV(n) represents the mutual vector of the n-th pair, and

D(n) represents external data obtained for the n-th pair.

14. A system for detecting first contact resolution failures, comprising:

a storage device configured to store a training corpus of pairs of interaction digital representations,

at least one hardware processor configured to execute code, the code comprising instructions for;

converting audio interactions into textual transcripts and textual interactions or receiving textual transcripts from an e-mail server, chat server or social media server;

storing in the training corpus the converted or received textual transcripts as pairs of interaction digital representations, wherein each pair is tagged as First Contact Resolution Failure (FCRF) or First Contact Resolution (FCR) interactions pair;

generating and storing in the storage device a topic model including a plurality of weight scores associated with corresponding key phrases, by detecting topics and corresponding sets of phrases in the interaction digital representations;

generating and storing in the storage device a First Contact Resolution Failures (FCRF) model by calculating based on the stored topic model a topic similarity score for each of the stored tagged pairs of interactions, and by executing a training module with the topic similarity score and parameters of each of the stored tagged pairs of interactions as input;

obtaining by a capturing apparatus a first plurality of digital representations, each digital representation of the first plurality is associated with a contact center interaction;

from the first plurality of digital representations, determining that a first and a second of the plurality of digital representations is a pair of digital representations that represent a pair of contact center interactions according to predefined rules applied on metadata associated with the pair of the contact center interactions, and associating the first and second interactions as candidates for contact resolution classification; and

applying the stored FCRF model for classification of the pair of contact center interactions candidates;

calculating an FCRF classification score representing the probability that a customer issue was not resolved within the first interaction of the pair of interactions, and handling of the same customer issue continued in the second interaction of the pair of interactions;

comparing the FCRF classification score to a predefined threshold, thereby classifying the pair of contact center interactions candidates as a first contact resolution failure (FCRF) interaction pair if the FCRF classification score is higher than the predefined threshold or classifying the pair of contact center interactions candidates as a first contact resolution (FCR) interactions pair if the FCRF classification score is lower than the predefined threshold; and

generating a report including a list of pairs of interactions that were classified as FCRF pairs of interactions;

wherein the classification of the determined pair of contact center interactions comprises;

determining based on the stored topic model a topic vector for each digital representation of the determined pair of digital representations; and

determining a topic similarity score between the topic vectors associated with the interactions of the determined pair of digital representations.

15. The system according to claim 14, further comprising an additional analysis component configured to perform additional filtering of the contact center interactions.

16. The system according to claim 14, further comprising an FCRF search module configured to search for pairs of interactions classified as FCRF pairs of interactions.

17. The system according to claim 14, further comprising an FCRF search module configured to search for pairs of interactions classified as FCR pairs of interactions.

18. The system according to claim 14, further comprising an FCRF report component configured to generate a report comprising a list of pair of interactions classified as FCRF pairs of interactions.