# US 12,214,661 B2 · Question-and-answer system and method of controlling the same

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,214,661 B2
- Document ID: US12214661B2
- Status: Granted
- Kind: B2
- Application number: 17/673,287
- Filed: 2022-02-16
- Published: 2022-08-18
- Granted: 2025-02-04
- Earliest claimed: 2021-02-18
- Assignees: Hyundai Motor Company; Kia Corporation
- Applicants: Hyundai Motor Company; Kia Corporation
- Inventors: Cheoneum Park; Mirye Lee; Donghyeon Kim; Cheongjae Lee; Sung Wook Kim
- Web page: https://patentgrep.com/patent/US12214661B2

## CPC

- G06F 16/3329: Natural language query formulation
- G06N 3/04: Architecture, e.g. interconnection topology
- B60K 35/10: Input arrangements, i.e. from user to vehicle, associated with vehicle functions or specially adapted therefor
- B60K 35/28: characterised by the type of the output information, e.g. video entertainment or vehicle dynamics information; characterised by the purpose of the output information, e.g. for attracting the attention of the driver
- B60K 35/80: Arrangements for controlling instruments
- G06F 16/3344: {using natural language analysis}
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/047: Probabilistic or stochastic networks
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 5/04: Inference or reasoning models
- B60K 35/26: using acoustic output
- B60K 2360/148: Instrument input by voice
- B60K 2360/161: Explanation of functions, e.g. instructions
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- B60K: ARRANGEMENT OR MOUNTING OF PROPULSION UNITS OR OF TRANSMISSIONS IN VEHICLES; ARRANGEMENT OR MOUNTING OF PLURAL DIVERSE PRIME-MOVERS IN VEHICLES; AUXILIARY DRIVES FOR VEHICLES; INSTRUMENTATION OR DASHBOARDS FOR VEHICLES; ARRANGEMENTS IN CONNECTION WITH COOLING, AIR INTAKE, GAS EXHAUST OR FUEL SUPPLY OF PROPULSION UNITS IN VEHICLES

## Abstract

A question-and-answer system providing an appropriate answer for a vehicle-related FAQ by using deep learning, and corresponding method are provided. The question-and-answer system includes: a memory that stores a plurality of representative questions to match a plurality of answers corresponding to the plurality of representative questions, respectively; a learning module configured to output a representative question corresponding to an input sentence from among the stored plurality of representative questions; and an output module configured to search the memory or an answer that matches the output representative question and output the found answer. The learning module is configured to perform multi task learning using a plurality of extended sentences for the plurality of representative questions as input data, and using the plurality of representative questions corresponding to the plurality of extended sentences, respectively, and a plurality of categories to which the plurality of extended sentences belong, respectively, as output data.

## Claims

### Claim 1

A question-and-answer system comprising:
a memory in which a plurality of representative questions are stored to match a plurality of answers corresponding respectively to the plurality of representative questions;
a learning module configured to output a representative question corresponding to an input sentence from among the stored plurality of representative questions; and
an output module configured to search the memory for an answer that matches the output representative question and output the searched answer;
wherein the learning module is configured to perform multi task learning by using a plurality of extended sentences for the plurality of representative questions as input data, and by using the plurality of representative questions corresponding respectively to the plurality of extended sentences, and a plurality of categories to which the plurality of extended sentences belong, respectively, as output data;
wherein the learning module is further configured to:
perform the multi task learning by using the plurality of extended sentences as input data, and by using the plurality of representative questions, the plurality of categories, and a plurality of named entities included in the plurality of extended sentences, respectively, as output data;
in the multi task learning, classify a representative question corresponding to the input data from among the stored plurality of representative questions, a category to which the input data belongs from among the plurality of categories, and a named entity included in the input data from among the plurality of named entities;
calculate a loss value of the classified representative question, a loss value of the classified category, and a loss value of the classified named entity; and
adjust a weight of a deep learning model used for the multi-tasking learning based on the three calculated loss values.

### Claim 2 (depends on 1)

The question-and-answer system of claim 1, wherein the plurality of representative questions stored in the memory includes frequently asked questions (FAQs) related to a vehicle.

### Claim 3 (depends on 1)

The question-and-answer system of claim 1, wherein the plurality of named entities includes terms related to a vehicle.

### Claim 4 (depends on 1)

The question-and-answer system of claim 1, wherein the learning module includes:
an encoding module configured to encode an input sequence corresponding to the input data;
a question classifier configured to classify the representative question based on output of the encoding module;
a category classifier configured to classify the category based on the output of the encoding module; and
a named entity classifier configured to classify the named entity based on the output of the encoding module.

### Claim 5 (depends on 4)

The question-and-answer system of claim 4, wherein the encoding module includes:
a first encoding layer configured to perform global encoding on the input sequence; and
a second encoding layer configured to perform sequential encoding on output of the first encoding layer.

### Claim 6 (depends on 5)

The question-and-answer system of claim 5, wherein the learning module further includes a variational inference network that is configured to apply a variational inference on output of the encoding module,
wherein output of the variational inference network is input to the category classifier.

### Claim 7

A method of controlling a question-and-answer system, the method comprising:
storing a plurality of representative questions to match a plurality of answers corresponding respectively to the plurality of representative questions;
performing multi task learning by using a plurality of extended sentences for the plurality of representative questions as input data, and by using the plurality of representative questions corresponding respectively to the plurality of extended sentences, and a plurality of categories to which the plurality of extended sentences belong, respectively, as output data;
in response to the multi task learning being completed, determining a representative question corresponding to an input sentence uttered by a user from among the stored plurality of representative questions based on a result of the multi-task learning; and
determining an answer that matches the determined representative question from among the stored plurality of answers;
wherein performing the multi task learning includes performing the multi-task learning by using the plurality of extended sentences as input data, and by using the plurality of representative questions, the plurality of categories, and a plurality of named entities included in the plurality of extended sentences, respectively, as output data;
wherein performing the multi task learning further includes classifying a representative question corresponding to the input data from among the stored plurality of representative questions, a category to which the input data belongs from among the plurality of categories, and a named entity included in the input data from among the plurality of named entities, and
wherein performing the multi task learning additionally includes calculating a loss value of the classified representative question, a loss value of the classified category, and a loss value of the classified named entity, and adjusting a weight of a deep learning model used for the multi-tasking learning based on the three calculated loss values.

### Claim 8 (depends on 7)

The method of claim 7, wherein the stored plurality of representative questions includes frequently asked questions (FAQs) related to a vehicle.

### Claim 9 (depends on 7)

The method of claim 7, wherein the plurality of named entities includes terms related to a vehicle.

### Claim 10 (depends on 7)

The method of claim 7, wherein performing the multi task learning includes performing global encoding and sequential encoding on an input sequence corresponding to the input data.