System and method for semantically exploring concepts

• US 10,446,135 B2
• Filed: 12/11/2017
• Issued: 10/15/2019
• Est. Priority Date: 07/09/2014
• Status: Active Grant

• Current Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Original Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Inventors: Faizakof, Avraham, Lev, Yoni, Lev-Tov, Amir, Konig, Yochai
• Primary Examiner(s): Guerra-Erazo, Edgar X
• Application Number:
  US15/837,980
• Publication Number:
  US 20180102126A1
• Time in Patent Office: 673 Days
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

A method for detecting and categorizing topics in a plurality of interactions includes: extracting, by a processor, a plurality of fragments from the plurality of interactions; filtering, by the processor, the plurality of fragments to generate a filtered plurality of fragments; clustering, by the processor, the filtered fragments into a plurality of base clusters; and clustering, by the processor, the plurality of base clusters into a plurality of hyper clusters.

Claims

1. An analytics system configured to automatically detect and categorize topics in a plurality of interactions between customers and agents of a contact center during one or more time periods, the interactions comprising a plurality of phrases, the analytics system comprising:

a processor; and

a memory, wherein the memory has instructions stored thereon that, when executed by the processor, cause the processor to;

extract a plurality of fragments of the phrases from the interactions in accordance with one or more extraction rules, the extraction rules being automatically generated based on sequences of parts of speech found in a set of key fragments of the fragments by;

tagging words or phrases of the key fragments with corresponding parts-of-speech tags to generate sequences of parts-of-speech tags;

identifying one or more of sequences of parts-of-speech tags appearing at least a threshold number of times among the key fragments; and

outputting the identified one or more of sequences of parts-of-speech tags as the one or more extraction rules;

filter the fragments to generate a filtered plurality of fragments by removing fragments based on at least one of;

a frequency with which the removed fragments appear among the fragments;

a saliency of the removed fragments; and

a stop list;

cluster the filtered plurality of fragments into a plurality of first clusters, each of the first clusters comprising a plurality of semantically similar fragments corresponding to a detected topic of the topics in the interactions;

cluster the first clusters into a plurality of second clusters, each of the second clusters corresponding to a categorization of the topics based on semantic similarity; and

output a hierarchy of concepts in accordance with the filtered plurality of fragments clustered into the first clusters of topics and the second clusters of categorizations of topics.

2. The analytics system of claim 1, wherein the instructions that cause the processor to extract the fragments from the interactions comprise instructions that, when executed by the processor, cause the processor to:

receive text corresponding to the interactions;

tag portions of the text based on parts of speech; and

extract fragments from the text in accordance with the one or more extraction rules.

3. The analytics system of claim 2, wherein the interactions comprise speech between the customers and the agents of the contact center, and

wherein the text corresponding to the interactions comprises an output of an automatic speech recognition engine, the output being generated by processing speech of at least one of the customers and speech of at least one of the agents from at least one of the interactions through the automatic speech recognition engine.

4. The analytics system of claim 1, wherein the memory further has stored thereon instructions that, when executed by the processor, cause the processor to label a first cluster of the first clusters by:

extracting a plurality of noun phrases from the first cluster;

computing a distribution of probabilities of stems of the noun phrases; and

identifying a label noun phrase of the noun phrases, the label noun phrase having a highest probability based on the distribution of probabilities of stems of the noun phrases.

5. The analytics system of claim 1, wherein the instructions that cause the processor to cluster the first clusters into the second clusters comprise instructions that, when executed by the processor, cause the processor to:

compute a plurality of semantic distances between pairs of the first clusters; and

cluster the first clusters into the second clusters in accordance with the semantic distances.

6. The analytics system of claim 5, wherein the instructions that cause the processor to compute the semantic distances between the pairs of the first clusters comprise instructions to compute a semantic distance of the semantic distances based on semantic similarities between the pairs of the first clusters and co-occurrence of fragments in the pairs of the first clusters.

7. The analytics system of claim 1, wherein the memory further has stored thereon instructions that, when executed by the processor, cause the processor to output the hierarchy of concepts by:

generating a visualization of the hierarchy of concepts; and

outputting the visualization of the hierarchy of concepts.

15. The system of claim 1, wherein the memory further stores instructions that, when executed by the processor, cause the processor to further automatically generate the one or more extraction rules by:

validating the one or more extraction rules in accordance with a precision and a recall of each extraction rule against the key fragments.

8. A method for performing analytics to automatically detect and categorize topics in a plurality of interactions between customers and agents of a contact center during one or more time periods, the interactions comprising a plurality of phrases, the method comprising:

extracting, by a processor of an analytics system, a plurality of fragments of the phrases from the interactions in accordance with one or more extraction rules, the extraction rules being automatically generated based on sequences of parts of speech found in a set of key fragments of the fragments by;

tagging words or phrases of the key fragments with corresponding parts-of-speech tags to generate sequences of parts-of-speech tags;

identifying one or more of sequences of parts-of-speech tags appearing at least a threshold number of times among the key fragments; and

outputting the identified one or more of sequences of parts-of-speech tags as the one or more extraction rules;

filtering, by the processor, the fragments to generate a filtered plurality of fragments by removing fragments based on at least one of;

a frequency with which the removed fragments appear among the fragments;

a saliency of the removed fragments; and

a stop list;

clustering, by the processor, the filtered plurality of fragments into a plurality of first clusters, each of the first clusters comprising a plurality of semantically similar fragments corresponding to a detected topic of the topics in the interactions;

clustering, by the processor, the first clusters into a plurality of second clusters, each of the second clusters corresponding to a categorization of the topics based on semantic similarity; and

outputting, by the processor, a hierarchy of concepts in accordance with the filtered plurality of fragments clustered into the first clusters of topics and the second clusters of categorizations of topics.

9. The method of claim 8, further comprising:

receiving text corresponding to the interactions;

tagging portions of the text based on parts of speech; and

extracting fragments from the text in accordance with the one or more extraction rules.

10. The method of claim 9, wherein the interactions comprise speech between the customers and the agents of the contact center, and

wherein the text corresponding to the interactions comprises an output of an automatic speech recognition engine, the output being generated by processing speech of at least one of the customers and speech of at least one of the agents from at least one of the interactions through the automatic speech recognition engine.

11. The method of claim 8, further comprising:

extracting a plurality of noun phrases from the first clusters;

computing a distribution of probabilities of stems of the noun phrases; and

identifying a label noun phrase of the noun phrases, the label noun phrase having a highest probability based on the distribution of probabilities of stems of the noun phrases.

12. The method of claim 8, further comprising:

computing a plurality of semantic distances between pairs of the first clusters; and

clustering the first clusters into the second clusters in accordance with the semantic distances.

13. The method of claim 12, wherein the computing a plurality of semantic distances between pairs of the first clusters comprises computing a semantic distance of the semantic distances based on semantic similarities between the pairs of the first clusters and co-occurrence of fragments in the pairs of the first clusters.

14. The method of claim 8, wherein the outputting the hierarchy of concepts comprises generating and outputting a visualization of the hierarchy of concepts.

16. The method of claim 8, further comprising validating the one or more extraction rules in accordance with a precision and a recall of each rule against the key fragments.