# US 2010/0063798 A1 · Error-detecting apparatus and methods for a chinese article

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

- Number: US 2010/0063798 A1
- Document ID: US20100063798A1
- Status: Published application
- Kind: A1
- Application number: 12/324,590
- Filed: 2008-11-26
- Published: 2010-03-11
- Earliest claimed: 2008-09-09
- Assignees: Institute for Information Industry
- Inventors: Tsun KU; Shih-Hung WU; Wen-Nan WANG; Wen-Tai HSIEH; Ta-Hung HUNG
- Web page: https://patentgrep.com/patent/US20100063798A1

## CPC

- G06F 40/232: Orthographic correction, e.g. spell checking or vowelisation
- G06F 40/53: Processing of non-Latin text (kana-to-kanji conversion G06F40/129; vowelisation G06F40/232)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The invention discloses an error-detecting method for a Chinese article, handling a Chinese sentence including a first erroneous Chinese character string in a first location. The method includes subdividing the first erroneous Chinese character string into a plurality of first subgroups, wherein each of the first subgroups consists of two consecutive and non-consecutive Chinese characters out of the first erroneous Chinese character string. The method further includes providing a database containing a plurality of first correct Chinese character strings and a plurality of corresponding first correct indices, wherein the first correct indices consist of two consecutive and non-consecutive Chinese characters out of the first correct Chinese character strings. The method further includes acquiring one of the first correct indices according to the first subgroup, and one of the first correct Chinese character strings according to the acquired first correct index. The method further includes generating a best candidate sentence according to the acquired first correct Chinese character string, and showing the Chinese sentence and the best candidate sentence on a display device.

## Claims

### Claim 1

An error-detecting method for a Chinese article, applicable for handling a Chinese sentence comprising a first erroneous Chinese character string in a first location, comprising:
subdividing the first erroneous Chinese character string into a plurality of first subgroups, wherein each of the first subgroups consists of two consecutive and non-consecutive Chinese characters out of the first erroneous Chinese character string;
providing a database comprising a plurality of first correct Chinese character strings and a plurality of corresponding first correct indices, wherein the first correct indices consist of two consecutive and non-consecutive Chinese characters out of the first correct Chinese character strings;
acquiring one of the first correct indices according to the first subgroups, and one of the first correct Chinese character strings according to the acquired first correct index;
generating a best candidate sentence according to the acquired first correct Chinese character string; and
showing the Chinese sentence and the best candidate sentence on a display device.

### Claim 2 (depends on 1)

The error-detecting method for a Chinese article as claimed in claim 1, the best candidate sentence is generated by replacing the first erroneous Chinese character string of the Chinese sentence with the acquired first correct Chinese character string.

### Claim 3 (depends on 1)

The error-detecting method for a Chinese article as claimed in claim 1, wherein the Chinese sentence further comprises a second erroneous Chinese character string in a second location.

### Claim 4 (depends on 3)

The error-detecting method for a Chinese article as claimed in claim 3, wherein the database further comprises a plurality of second erroneous indices and a plurality of corresponding second correct Chinese character strings.

### Claim 5 (depends on 4)

The error-detecting method for a Chinese article as claimed in claim 4, further comprising setting the acquired first correct Chinese character string as a first candidate term and generating a second candidate term according to the second erroneous indices.

### Claim 6 (depends on 5)

The error-detecting method for a Chinese article as claimed in claim 5, further comprising determining whether the second erroneous Chinese character string is the same as one of the second erroneous indices.

### Claim 7 (depends on 6)

The error-detecting method for a Chinese article as claimed in claim 6, farther comprising:
acquiring the second erroneous index that is the same as the second erroneous Chinese character string, and one of the second correct Chinese character strings according to the acquired second erroneous index; and
setting the acquired second correct Chinese character string as the second candidate term.

### Claim 8 (depends on 7)

The error-detecting method for a Chinese article as claimed in claim 7, further comprising generating a plurality of candidate sentences by replacing the first location and the second location of the Chinese sentence with the first erroneous Chinese character string, the second erroneous Chinese character sting, the first candidate term and the second candidate term.

### Claim 9 (depends on 8)

The error-detecting method for a Chinese article as claimed in claim 8, further comprising generating the best candidate sentence by scoring the candidate sentences according to a candidate-sentence scoring mechanism.

### Claim 10 (depends on 9)

The error-detecting method for a Chinese article as claimed in claim 9, wherein the candidate-sentence scoring mechanism scores the candidate sentences according to the following four factors:
at least one language model;
sentence similarity between the Chinese sentence and the candidate sentences;
pronunciation similarity between the Chinese sentence and the candidate sentences; and
Chinese character similarity between the Chinese sentence and the candidate sentences.

### Claim 11 (depends on 6)

The error-detecting method for a Chinese article as claimed in claim 6, wherein the database farther comprises a plurality of contexts.

### Claim 12 (depends on 11)

The error-detecting method for a Chinese article as claimed in claim 11, further comprising:
acquiring the second erroneous index that is the same as the second erroneous Chinese character string, and one of the contexts according to the acquired second erroneous index; and
determining whether the Chinese sentence comprises the acquired context.

### Claim 13 (depends on 12)

The error-detecting method for a Chinese article as claimed in claim 12, further comprising:
acquiring one of the second correct Chinese character strings when the Chinese sentence comprises the acquired context; and
setting the acquired second correct Chinese character string as the second candidate term.

### Claim 14

An error-detecting apparatus for a Chinese article, applicable for handling a Chinese sentence comprising a first erroneous Chinese character string in a first location, comprising:
an article-dividing module subdividing the first erroneous Chinese character string into a plurality of first subgroups, wherein each of the first subgroups consists of two consecutive and non-consecutive Chinese characters out of the first erroneous Chinese character string;
a database comprising a plurality of first correct Chinese character strings and a plurality of corresponding first correct indices, wherein the first correct indices consist of two consecutive and non-consecutive Chinese characters out of the first correct Chinese character strings;
a candidate term generation module acquiring one of the first correct indices according to the first subgroups, and one of the first correct Chinese character strings according to the acquired first correct index;
is a candidate sentence generating and scoring module generating a best candidate sentence according to the acquired first correct Chinese character string; and
a display device showing the Chinese sentence and the best candidate sentence.

### Claim 15 (depends on 14)

The error-detecting apparatus for a Chinese article as claimed in claim 14, wherein the best candidate sentence is generated by replacing the first erroneous Chinese character string of the Chinese sentence with the acquired first correct Chinese character string.

### Claim 16 (depends on 14)

The error-detecting apparatus for a Chinese article as claimed in claim 14, wherein the Chinese sentence further comprises a second erroneous Chinese character string in a second location.

### Claim 17 (depends on 16)

The error-detecting apparatus for a Chinese article as claimed in claim 16, wherein the database farther comprises a plurality of second erroneous indices and a plurality of corresponding second correct Chinese character strings.

### Claim 18 (depends on 17)

The error-detecting apparatus for a Chinese article as claimed in claim 17, wherein the candidate term generation module further sets the acquired first correct Chinese character suing as a first candidate term and generates a second candidate term according to the second erroneous indices.

### Claim 19 (depends on 18)

The error-detecting apparatus for a Chinese article as claimed in claim 18, wherein the candidate term generation module further determines whether the second erroneous Chinese character sting is the same as one of the second erroneous indices.

### Claim 20 (depends on 19)

The error-detecting apparatus for a Chinese article as claimed in claim 19, wherein the candidate term generation module further performs the following procedures:
acquiring the second erroneous index that is the same as the second erroneous Chinese character string, and one of the second correct Chinese character strings according to the acquired second erroneous index; and
setting the acquired second correct Chinese character string as the second candidate term.

### Claim 21 (depends on 20)

The error-detecting apparatus for a Chinese article as claimed in claim 20, wherein the candidate sentence generating and scoring module further generates a plurality of candidate sentences by replacing the first location and the second location of the Chinese sentence with the first erroneous Chinese character string, the second erroneous Chinese character string, the first candidate term and the second candidate term.

### Claim 22 (depends on 21)

The error-detecting apparatus for a Chinese article as claimed in claim 21, wherein the candidate sentence generating and scoring module further generates the best candidate sentence by scoring the candidate sentences according to a candidate-sentence scoring mechanism.

### Claim 23 (depends on 22)

The error-detecting apparatus for a Chinese article as claimed in claim 22, wherein the candidate-sentence scoring mechanism scores the candidate sentences according to the following four factors:
at least one language model;
sentence similarity between the Chinese sentence and the candidate sentences;
pronunciation similarity between the Chinese sentence and the candidate sentences; and
Chinese character similarity between the Chinese sentence and the candidate sentences.

### Claim 24 (depends on 19)

The error-detecting apparatus for a Chinese article as claimed in claim 19, wherein the database farther comprises a plurality of contexts.

### Claim 25 (depends on 24)

The error-detecting apparatus for a Chinese article as claimed in claim 24, wherein the candidate term generation module further performs the following procedures:
acquiring the second erroneous index that is the same as the second erroneous Chinese character string, and one of the contexts according to the acquired second erroneous index; and
determining whether the Chinese sentence comprises the acquired context.

### Claim 26 (depends on 25)

The error-detecting apparatus for a Chinese article as claimed in claim 25, wherein the candidate term generation module further performs the following procedures:
acquiring one of the second correct Chinese character strings when the Chinese sentence comprises the acquired context; and
setting the acquired second correct Chinese character string as the second candidate term.

### Claim 27

A storage medium for storing an error-detecting program for a Chinese article, wherein the error-detecting program comprises a plurality of program codes to be loaded onto a computer system so that an error-detecting method for the Chinese article is executed by the computer system, and the error-detecting method is applicable for handling a Chinese sentence comprising a first erroneous Chinese character string in a first location, comprising:
subdividing the first erroneous Chinese character string into a plurality of first subgroups, wherein each of the first subgroups consists of two consecutive and non-consecutive Chinese characters out of the first erroneous Chinese character string;
storing a plurality of first correct Chinese character strings and a plurality of corresponding first correct indices, wherein the first correct indices consist of two consecutive and non-consecutive Chinese characters out of the first correct Chinese character strings;
acquiring one of the first correct indices according to the first subgroups, and one of the first correct Chinese character strings according to the acquired first correct index;
generating a best candidate sentence according to the acquired first correct Chinese character string; and
showing the Chinese sentence and the best candidate sentence.
