System and method for implementing an artificially intelligent virtual assistant using machine learning

• US 10,572,801 B2
• Filed: 11/22/2017
• Issued: 02/25/2020
• Est. Priority Date: 11/22/2017
• Status: Active Grant

• Current Assignee: Clinc, Inc.
• Original Assignee: Clinc, Inc.
• Inventors: Mars, Jason, Tang, Lingjia, Laurenzano, Michael, Hauswald, Johann, Hill, Parker
• Primary Examiner(s): Zhen, Li B.
• Assistant Examiner(s): Lee, Tsu-Chang
• Application Number:
  US15/821,010
• Publication Number:
  US 20190156198A1
• Time in Patent Office: 825 Days
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 3/044 Recurrent networks, e.g. Hopfield networks
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/047 Probabilistic or stochastic networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: G06N 3/088 Non-supervised learning, e.g. competitive learning
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
• CPC Class Codes: G06N 5/025 Extracting rules from data
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03

Abstract

Systems and methods for implementing an artificially intelligent virtual assistant includes collecting a user query; using a competency classification machine learning model to generate a competency label for the user query; using a slot identification machine learning model to segment the text of the query and label each of the slots of the query; generating a slot value for each of the slots of the query; generating a handler for each of the slot values; and using the slot values to: identify an external data source relevant to the user query, fetch user data from the external data source, and apply one or more operations to the query to generate response data; and using the response data, to generate a response to the user query.

Claims

1. A machine learning-based system that implements an artificially intelligent virtual agent the system comprising:

a cache memory storing at least one historical user query posed by a given user to the machine learning-based system at a point earner in time;

a user interface that receives a real-time query from the given user;

one or more networked computing devices that implement the machine learning-based system including;

a trained machine learning model that performs;

(a) a first classification of the real-time query by the given user, the first classification includes an output of a competency classification label identifying at least one area of aptitude required for generating a response to the given user based on the real-time query;

(b) contemporaneous with the first classification, a second classification of the real-time query by the given user that is distinct from the first classification indicating whether the real-time user query comprises a follow-on query relative to the at least one historical user query of the given user that is stored in the cache memory;

wherein;

based on the second classification of the real-time query that includes an output of a follow-on query classification label by the trained machine learning model, the machine learning-based system automatically performs;

(1) a search of the cache memory, the search comprising at least data relating to the competency classification label output from the first classification and the follow-on query classification label of the second classification;

(2) an identification from within the cache memory of at least one historical query of the given user based on the search;

based on the identification of the at least one historical query of the given user, the machine learning-based system;

retrieves, from a memory device, a historical response to the at least one historical query of the given user;

the machine learning-based system further;

generates one or more new slot values based on the real-time query of the given user;

updates one or more historical slot values of the historical response to the at least one historical query of the given user with the one or more new slot values;

outputs a communication hi response to the real-time query of the given user, the communication comprising the historical response updated with the one or more new slot values.

2. The system of claim 1, wherein:

the user query is related to the cognate historical query based on a determined similarity between areas of aptitude the competency classifications labels of the real-time query and the historical query of the given user.

3. The system of claim 1, wherein:

the real-time query is related to the historical query based on the determined second classification label applied to the real-time query.

4. The system of claim 1, wherein:

the second classification label triggers a second mode of operating the machine learning-based system distinct from a first mode in which the second classification label is not produced, wherein the second mode includes an additional query handling process that involves searching for and identifying the historical query and updating the response data of the historical query with slot values derived for the real-time query.

5. The system of claim 1, wherein:

identifying the historical query includes configuring

a first subroutine that retrieves a history of prior queries of the given user; and

a second subroutine that filters the history of prior queries based on the competency classification label of the real-time query of the given user.

6. The system of claim 5, wherein:

identifying the historical query includes configuring

a third subroutine that filters the history of prior queries based on the one or more slot classification labels of the real-time query;

after filtering the history of prior queries, populating one or more prior queries candidates; and

selecting one of the one or more prior queries candidates as the historical query.

7. The system of claim 1, wherein:

the competency classification label corresponds to one competency of a plurality of areas of competencies of the machine learning-based system, and wherein a competency relates to a subject area of comprehension or aptitude of an artificially intelligent virtual assistant for which the artificially intelligent virtual assistant functions to interact with or provide the historical response having the updates of the one or more slot values to the given user based on the real-time query.

8. The system of claim 1, wherein:

slot classification labels relate a plurality of predetermined slot labels that describes segments of the text of the real-time query.

9. The system of claim 1, wherein:

the observables extraction system that uses, at least, each of the computer-executable operations to generate a plurality of handlers, wherein each of the plurality of handlers comprise one or more computer-executable subroutines for handling user data based on the program-comprehensible objects or values for each of the slots of user input data.

10. The system according to claim 1, wherein

the trained machine learning classifier is trained using samples of follow-on questions and/or follow-on commands from one or more data sources.