METHOD AND APPARATUS FOR FACILITATING CUSTOMER INTENT PREDICTION

• US 20170116177A1
• Filed: 10/26/2016
• Published: 04/27/2017
• Est. Priority Date: 10/26/2015
• Status: Active Grant

• Current Assignee: [24]7.ai, Inc.
• Original Assignee: 24/7 Customer Incorporated
• Inventors: WALIA, Anmol
• Granted Patent: US 10,579,834 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3335 Syntactic pre-processing, e.g. stopword elimination, stemming
• CPC Class Codes: G06F 40/129 Handling non-Latin characters, e.g. kana-to-kanji conversion
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/242 Dictionaries
• CPC Class Codes: G06F 40/295 Named entity recognition
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• 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 3/527 Centralised call answering arrangements not requiring operator intervention
• CPC Class Codes: H04M 7/0045 where the text-based messaging service is an instant messaging service

Abstract

A computer-implemented method and an apparatus facilitate customer intent prediction. The method includes receiving natural language communication provided by a customer on at least one enterprise related interaction channel. Textual data corresponding to the natural language communication is generated by converting one or more non-textual portions in the natural language communication to a text form. One or more processing operations are performed on the textual data to generate normalized text. The normalized text is configured to facilitate interpretation of the natural language communication. At least one intention of the customer is predicted, at least in part, based on the normalized text and a reply is provisioned to the customer based on the predicted intention. The reply is provisioned to the customer on the at least one enterprise related interaction channel in response to the natural language communication.

Claims

1. A computer-implemented method, comprising:

receiving, by a processor, natural language communication provided by a customer on at least one enterprise related interaction channel;

when the natural language communication comprises one or more non-textual portions, converting, by the processor, the one or more non-textual portions to a text form to generate textual data corresponding to the natural language communication, the processor using textual content associated with the natural language communication to configure the textual data when the natural language communication does not comprise non-textual portions;

performing, by the processor, at least one processing operation on the textual data to generate normalized text corresponding to the natural language communication, the processor configuring the normalized text to facilitate interpretation of the natural language communication provided by the customer;

predicting, by the processor, at least one intention of the customer, at least in part, based on the normalized text corresponding to the natural language communication; and

causing, by the processor, a provisioning of a reply to the customer based on the at least one intention, the reply provisioned to the customer on the at least one enterprise related interaction channel in response to the natural language communication.

2. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by;

determining if the textual data comprises one or more regularly used expressions;

when the textual data comprises one or more regularly used expressions, determining if each regularly used expression from among the one or more regularly used expressions is expressed in a respective predetermined format; and

replacing a current format of at least one regularly used expression with the respective predetermined format if the at least one regularly used expression is not expressed in the respective predetermined format, wherein replacing the current format with the respective predetermined format facilitates, at east in part, generation of the normalized text.

3. The method of claim 2, wherein a regularly used expression from among the one or more regularly used expressions comprises any of a date expression, a time expression, a currency amount expression, an email expression, a phone number expression, and an account number expression.

4. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by removing non-English characters from the textual data, to facilitate, at least in part, generation of the normalized text.

5. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by substituting symbols with equivalent word representations to facilitate, at least in part, generation of the normalized text.

6. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by;

determining if the textual data comprises at least one word corresponding to any of a name of an individual, a relation of an individual, a profession of an individual, a gender of an individual, or a location of an individual; and

substituting the at least one word in the textual data with a respective word class to facilitate, at least in part, generation of the normalized text.

7. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by;

determining if one or more words from among a plurality of words configuring the textual data are extensions of word stems; and

replacing the one or more words with corresponding word stems to facilitate, at least in part, generation of the normalized text.

8. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by removing one or more stop-words present in the textual data to facilitate, at least in part, generation of the normalized text.

9. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by replacing at least one of abbreviations and acronyms in the textual data with corresponding full-word representations to facilitate, at least in part, generation of the normalized text.

10. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by replacing two or more consecutive white spaces, tab delimiters, and newline characters in the textual data with single spaces to facilitate, at least in part, generation of the normalized text.

11. The method of claim 1, further comprising:

performing, by the processor, a processing operation from among the at least one processing operation by;

performing a spelling check of words configuring the textual data; and

when one or more words with incorrect spellings are identified in the textual data during the spelling check, correcting the incorrect spellings of one or more words in the textual data to facilitate, at least in part, generation of the normalized text.

12. The method of claim 11, further comprising:

performing the spelling check by;

generating at least one list of words based on predefined criteria;

configuring a set of parameters for performing the spelling check;

for each word in the textual data, determining if the word exists in a dictionary;

when the word exists in the dictionary, checking if the word is present in the at least one list of words;

classifying the word if the word is present in the at least one list of words;

when the word does not exist in the dictionary, identifying one or more suggestions for the word;

scoring each suggestion from among t e one or more suggestions based on an order of relevance; and

correcting the incorrect spellings based on scores associated with the one or more suggestions.

13. The method of claim 12, further comprising:

generating an n-gram model;

using said n-gram model to determine a context of the word in the textual data; and

scoring each suggestion from among the one or more suggestions based on the context of the word.

14. The method of claim 13, further comprising:

scoring the each suggestion with at least one of an error model and a statistical language model.

15. The method of claim 1, further comprising:

determining, by the processor, one or more recommendations for providing a personalized treatment to the customer based on the predicted at least one intention; and

provisioning the reply to the customer based on the one or more recommendations.

16. The method of claim 1, further comprising:

receiving, by the processor, multi-modal data corresponding to customer interaction on the at least one enterprise related interaction channel, wherein the multi-modal data comprises the natural language communication,

17. 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;

receive natural language communication provided by a customer on at least one enterprise related interaction channel;

when the natural language communication comprises one or more non-textual portions, convert the one or more non-textual portions to a text form to generate textual data corresponding to the natural language communication, wherein textual content associated with the natural language communication is used to configure the textual data when the natural language communication does not comprise non-textual portions;

perform at least one processing operation on the textual data to generate normalized text corresponding to the natural language communication, the normalized text facilitating interpretation of the natural language communication provided by the customer;

predict at least one intention of the customer, at least in part, based on the normalized text corresponding to the natural language communication; and

cause a provisioning of a reply to the customer based on the at least one intention, the reply provisioned to the customer on the at least one enterprise related interaction channel in response to the natural language communication.

18. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to;

determine if the textual data comprises one or more regularly used expressions;

when the textual data comprises one or more regularly used expressions, determine if each regularly used expression from among the one or more regularly used expressions is expressed in a respective predetermined format; and

replace a current format of at least one regularly used expression with the respective predetermined format if the at least one regularly used expression is not expressed in the respective predetermined format to facilitate, at least in part, generation of the normalized text.

19. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to remove non-English characters from the textual data to facilitate, at least in part, generation of the normalized text.

20. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to substitute symbols with equivalent word representations to facilitate, at least in part, generation of the normalized text.

21. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to;

determine if the textual data comprises at least one word corresponding to any of a name of an individual, a relation of an individual, a profession of an individual, a gender of an individual, or a location of an individual; and

substitute the at least one word in the textual data with a respective word class to facilitate, at least in part, generation of the normalized text.

22. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to;

determine if one or more words from among a plurality of words configuring the textual data are extensions of word stems; and

replace the one or more words with corresponding word stems to facilitate, at least in part, generation of the normalized text.

23. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to remove one or more stop-words present in the textual data to facilitate, at least in part, generation of the normalized text.

24. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to replace at least one of abbreviations and acronyms in the textual data with corresponding full-word representations to facilitate, at least in part, generation of the normalized text.

25. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation, the apparatus is caused to replace two or more consecutive white spaces, tab delimiters, and newline characters in the textual data with single spaces to facilitate, at least in part, generation of the normalized text.

26. The apparatus of claim 17, further comprising:

for performing a processing operation from among the at least one processing operation_;

the apparatus is caused to;

perform a spelling check of words configuring the textual data; and

when one or more words with incorrect spellings are identified in the textual data during the spelling check, correct the incorrect spellings of one or more words in the textual data to facilitate, at least in part, generation of the normalized text.

27. The apparatus of claim 17, further comprising:

for performing the spelling check, the apparatus is caused to;

facilitate generation of at least one list of words based on predefined criteria;

facilitate configuration of a set of parameters for performing the spelling check;

for each word in the textual data, determine if the word exists in a dictionary;

when the word exists in the dictionary, check if the word is present in the at least one list of words and facilitating classification of the word if the word is present in the at least one list of words; and

when the word does not exist in the dictionary, identify one or more suggestions for the word, score each suggestion from among the one or more suggestions based on an order of relevance, and perform the correction of the incorrect spellings based on scores associated with the one or more suggestions.

28. The apparatus of claim 27, wherein the apparatus is further caused to:

generate an n-gram model for determining a context of the word in the textual data; and

score each suggestion from among the one or more suggestions based on the context of the word.

29. The apparatus of claim 17, wherein the apparatus is further caused to:

determine one or more recommendations for providing a personalized treatment to the customer based on the predicted at least one intention, and

provision the reply to the customer based on the one or more recommendations.

30. An apparatus, comprising:

at least one communication interface configured to receive natural language communication provided by a customer on at least one enterprise related interaction channel;

a textual data generator configured to convert one or more non-textual portions to a text form to generate textual data corresponding to the natural language communication if the natural language communication comprises one or more non-textual portions, wherein textual content associated with the natural language communication configures the textual data if the natural language communication does not comprise non-textual portions;

a normalization module configured to perform at least one processing operation on the textual data to generate normalized text corresponding to the natural language communication, the normalized text configured to facilitate interpretation of the natural language communication provided by the customer; and

a prediction module configured to predict at least one intention of the customer, at least in part, based on the normalized text corresponding to the natural language communication,

wherein the at least one communication interface is caused to provision a reply to the customer based on the at least one intention, the reply provisioned to the customer on the at least one enterprise related interaction channel in response to the natural language communication.

31. The apparatus of claim 30, wherein the normalization module further comprises:

a regular expression module configured to replace a current format of at least one regularly used expression in the textual data with a respective predetermined format if the at least one regularly used expression is not expressed in the respective predetermined format;

a character removal module configured to remove non-English characters from the textual data;

a symbol substitution module configured to substitute symbols in the textual data with equivalent word representations;

a word class substitution module configured to substitute at least one word in the textual data with a respective word class;

a stemming module configured to replace one or more words in the textual data with corresponding word stems;

a stop-word removal module configured to remove stop-words in the textual data;

a short form replacement module configured to replace at least one of abbreviations and acronyms in the textual data with corresponding full-word representations;

a white space removal module configured to replace two or more consecutive white spaces, tab delimiters and newline characters in the textual data with single spaces; and

a spell checker module configured to perform a spelling check of words configuring the textual data and correcting incorrect spellings of one or more words in the textual data if the one or more words with incorrect spellings are identified in the textual data.

32. The apparatus of claim 31, wherein for performing the spelling check, the spell checker module is caused to:

facilitate generation of at least one list of words based on predefined criteria;

facilitate configuration of a set of parameters for performing the spelling check;

for each word in the textual data determine if the word exists in a dictionary;

if the word exists in the dictionary, Check if the word is present in the at least one list of words and facilitating classification of the word if the word is present in the at least one list of words; and

if the word does not exist in the dictionary, identify one or more suggestions for the word, wherein each suggestion from among the one or more suggestions is scored based on an order of relevance, and wherein the correction of the incorrect spellings is performed based on scores associated with the one or more suggestions.

33. The apparatus of claim 32, wherein the spell checker module is further caused to:

generate an n-gram model for determining a context of the word in the textual data; and

score each suggestion from among the one or more suggestions based on the context of the word.

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

determine one or more recommendations for providing a personalized treatment to the customer based on the predicted at least one intention, wherein the reply is provisioned to the customer based on the one or more recommendations.