# US 12,585,876 B2 · Method of training POS tagging model, computer-readable recording medium and POS tagging method

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

- Number: US 12,585,876 B2
- Document ID: US12585876B2
- Status: Granted
- Kind: B2
- Application number: 18/205,615
- Filed: 2023-06-05
- Published: 2024-02-01
- Granted: 2026-03-24
- Earliest claimed: 2022-07-28
- Assignees: Hyundai Motor Company; Kia Corporation
- Applicants: Hyundai Motor Company; Kia Corporation
- Inventors: Cheoneum Park; Juae Kim; Cheongjae Lee; Soo Jong Do
- Web page: https://patentgrep.com/patent/US12585876B2

## CPC

- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/205: Parsing
- G06F 40/216: using statistical methods
- G06F 40/268: Morphological analysis
- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method of training a part-of-speech (POS) tagging model includes: separating an input sentence into units of syllables to generate an input sequence; encoding, using at least one encoder included in a part-of-speech (POS) tagging model, the input sequence; generating, based on the encoded input sequence and using a first discriminator included in the POS tagging model, a POS tagging result; and generating, based on the encoded input sequence and using a second discriminator included in the POS tagging model, a spacing result.

## Claims

### Claim 1

A method comprising:
separating an input sentence into units of syllables to generate an input sequence;
encoding, using at least one encoder included in a part-of-speech (POS) tagging model, the input sequence, wherein the encoding the input sequence comprises:
performing global encoding by inputting the input sequence to a first encoder; and
performing encoding based on a word order by inputting an output of the first encoder to a second encoder;
generating, based on the encoded input sequence and using a first discriminator included in the POS tagging model, a POS tagging result; and
generating, based on the encoded input sequence and using a second discriminator included in the POS tagging model, a spacing result.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
calculating a first loss value on the POS tagging result; and
calculating a second loss value on the spacing result.

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
calculating a total loss value by summing the first loss value and the second loss value; and
adjusting parameters of the POS tagging model based on the total loss value.

### Claim 4 (depends on 1)

The method of claim 1, wherein the at least one encoder included in the POS tagging model comprises the first encoder and the second encoder.

### Claim 5 (depends on 1)

The method of claim 1, wherein the first encoder includes a pre-trained language model.

### Claim 6 (depends on 1)

The method of claim 1, wherein the second encoder includes a bidirectional simple recurrent unit (SRU).

### Claim 7 (depends on 1)

The method of claim 1, wherein the first discriminator includes a conditional random field (CRF) layer.

### Claim 8 (depends on 1)

The method of claim 1, wherein the second discriminator includes a conditional random field (CRF) layer.

### Claim 9

An apparatus comprising:
one or more processors; and
memory storing instructions that, when executed by the one or more processors, cause the apparatus to:
separate an input sentence into units of syllables to generate an input sequence;
encode, using at least one encoder included in a part-of-speech (POS) tagging model, the input sequence by:
performing global encoding by inputting the input sequence to a first encoder; and
performing encoding based on a word order by inputting an output of the first encoder to a second encoder;
generate, based on the encoded input sequence and using a first discriminator included in the POS tagging model, a POS tagging result; and
generate, based on the encoded input sequence and using a second discriminator included in the POS tagging model, a spacing result.

### Claim 10 (depends on 9)

The apparatus of claim 9, wherein the at least one encoder included in the POS tagging model comprises the first encoder and the second encoder.

### Claim 11 (depends on 9)

The apparatus of claim 9, wherein the first encoder includes a pre-trained language model.

### Claim 12 (depends on 9)

The apparatus of claim 9, wherein the second encoder includes a bidirectional simple recurrent unit (SRU).

### Claim 13 (depends on 9)

The apparatus of claim 9, wherein the first discriminator includes a conditional random field (CRF) layer.

### Claim 14 (depends on 9)

The apparatus of claim 9, wherein the second discriminator includes a conditional random field (CRF) layer.

### Claim 15

A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause:
separating an input sentence into units of syllables to generate an input sequence;
encoding, using at least one encoder included in a part-of-speech (POS) tagging model, the input sequence by:
performing global encoding by inputting the input sequence to a first encoder; and
performing encoding based on a word order by inputting an output of the first encoder to a second encoder;
generating, based on the encoded input sequence and using a first discriminator included in the POS tagging model, a POS tagging result; and
generating, based on the encoded input sequence and using a second discriminator included in the POS tagging model, a spacing result.

### Claim 16 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the instructions, when executed by the one or more processors, further cause:
calculating a first loss value on the POS tagging result; and
calculating a second loss value on the spacing result.

### Claim 17 (depends on 16)

The non-transitory computer-readable medium of claim 16, wherein the instructions, when executed by the one or more processors, further cause:
calculating a total loss value by summing the first loss value and the second loss value; and
adjusting parameters of the POS tagging model based on the total loss value.

### Claim 18 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the at least one encoder included in the POS tagging model comprises the first encoder and the second encoder.

### Claim 19 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the first encoder includes a pre-trained language model, and
wherein the second encoder includes a bidirectional simple recurrent unit (SRU).

### Claim 20 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein each of the first discriminator and the second discriminator includes a conditional random field (CRF) layer.
