CN105045778B - A kind of Chinese homonym mistake auto-collation - Google Patents
A kind of Chinese homonym mistake auto-collation Download PDFInfo
- Publication number
- CN105045778B CN105045778B CN201510354692.4A CN201510354692A CN105045778B CN 105045778 B CN105045778 B CN 105045778B CN 201510354692 A CN201510354692 A CN 201510354692A CN 105045778 B CN105045778 B CN 105045778B
- Authority
- CN
- China
- Prior art keywords
- homonym
- word
- chinese
- adjacent
- mistake
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Machine Translation (AREA)
Abstract
The invention discloses a kind of Chinese homonym mistake auto-collation, the homonym that this method firstly generates Chinese word obscures collection, then left adjacent binary is counted by the training of extensive Web language materials, right adjacent binary, adjacent ternary model, obscure collection using homonym and probability Estimation algorithm obtains local adjacent NGram models, then the combined method of Weight is utilized, context support of the homonym of concentration in sentence is obscured by the word and its corresponding homonym that calculate in sentence, judge whether homonym mistake, and homonym mistake is marked and amending advice list is provided, so as to realize the automatic Proofreading of Chinese homonym.The homonym mistake auto-collation that the present invention is provided, faster system response, precision meet practical application request, and validity and accuracy are high.
Description
Technical field
The present invention relates to the natural language processing in artificial intelligence computer field, more particularly to automatic proofreading for Chinese texts
Field.
Background technology
With the information processing technology and the high speed development of internet, traditional text work almost all is taken by computer
In generation, e-text, blog, the microblogging such as e-book, electronic newspaper, Email, office document etc. all turn into people's daily life
A part, but in text mistake it is also more and more, this brings very big challenge to proof-reading.Traditional artificial school
, intensity low to efficiency is big, cycle length obviously can not meet the demand of text proofreading.
Text automatic Proofreading is one of main application of natural language processing, is also the problem of natural language understanding.Chinese
It is to be input to by input method in computer, as increasing people using spelling input method inputs Chinese character, and Pinyin Input
Method can be with input word and phrase, therefore occurs increasing homonym mistake in the text, and homonym mistake is to belong to true word
The category of mistake.The auto-collation of the true word mistake of Chinese has problems with：
1) word for occurring true word mistake is also the word in dictionary, and this is the difficult point of automatic proofreading for Chinese texts.
2) true word mistake can disturb the syntax and semantics of whole sentence, consequently found that true word mistake need many knowledge with
Resource.
3) Sparse is a main obstacle of the automatic Proofreading of true word mistake；
4) homonym automatic Proofreading includes automatic errordetecting and automatic error-correcting, and automatic errordetecting is to find that the homonym in sentence is wrong
By mistake, and automatic error-correcting is to be proofreaded that there is provided amending advice to the mistake in sentence.And many methods are all to look into automatically at present
Wrong and two stages of automatic error-correcting separate.
For above-mentioned Railway Project, the present invention proposes and realizes the automatic errordetecting of Chinese homonym mistake and automatic school
To method.
The content of the invention
Goal of the invention：In order to overcome the deficiencies in the prior art, the present invention provides a kind of Chinese homonym mistake certainly
Dynamic proofreading method, is a kind of method for integrating automatic errordetecting and automatic Proofreading.
Technical scheme：
In order to solve the above technical problems, the present invention provides a kind of Chinese homonym mistake auto-collation, it is based on same
Sound word obscures the adjacent NGram models combination determining method of the part of collection and Weight and carries out Chinese homonym mistake automatic Proofreading, should
Method comprises the following steps：
1) by phonetic transcriptions of Chinese characters, the homonym for setting up Chinese word obscures collection；
2) set up that the second from left is first, right binary and ternary local adjacent NGram models, based on step 1) obtained homonym mixes
Confuse collection, carries out probability Estimation to the local adjacent NGram models by Probabilistic estimation, is trained by large-scale corpus
To local adjacent NGram models；
3) it is based on step 2) obtained adjacent NGram models of part, using the combined method of Weight, by calculating sentence
In word and its corresponding homonym obscure context support of the homonym of concentration in sentence, judge whether unisonance
Word mistake, and homonym mistake is marked and amending advice list is provided.
It is preferred that, the step 1) include：Using phonetic transcriptions of Chinese characters table and Chinese dictionary, generation homonym obscures collection：
Wherein WiIt is a Chinese word,It is WiHomonym.
It is preferred that, the step 1) in homonym obscure collection source include two parts：Automatic identification part and artificial school
To part；
Wherein automatic identification part comprises the following steps：
Step 11) Chinese dictionary is read, the Chinese word in Chinese dictionary is read into Chinese word structure；
Step 12) phonetic transcriptions of Chinese characters of phonetic transcriptions of Chinese characters table is read in into phonetic transcriptions of Chinese characters structure；
Step 13) combine step 11) obtained Chinese word and step 12) obtained phonetic transcriptions of Chinese characters, by Chinese dictionary
Chinese word is converted into phonetic, is put into generation homonym dictionary configuration, i.e. homonym in unisonance word structure and obscures collection；
Wherein artificial check and correction part includes：To step 13) obtained homonym obscures collection and manually proofreaded, renewal unisonance
Word obscures collection；
The structure that the homonym obscures collection is as follows：
Wherein WiIt is a word,It is WiHomonym.
It is preferred that, the step 2) comprise the following steps：
Step 21) extensive Web language materials are based on, set up the local adjacent of left adjacent binary, right adjacent binary and adjacent ternary
Connect NGram models：Participle is carried out to the sentence in language material, such as carrying out participle to sentence L obtains L=W1W2..Wi-1WiWi+1…
Wn, for word WiFor,
Left adjacent binary is：LeftBiGram(Wi)=Wi-1Wi；
Right adjacent binary is：RightBiGram(Wi)=WiWi+1；
Abutting ternary is：TriGram(Wi)=Wi-1WiWi+1；
Step 22) collection CSet (W are obscured based on extensive Web language materials and homonymi), it is all that statistics homonym obscures concentration
Left adjacent binary and its co-occurrence frequency of word, right adjacent binary and its co-occurrence frequency and adjacent ternary and its co-occurrence frequency, its
In It is WiHomonym；
Step 23) collection CSet (W are obscured based on homonymi), to the office of left adjacent binary, right adjacent binary and adjacent ternary
Portion's adjoining NGram models carry out probability Estimation, so as to generate the adjacent NGram models of the part comprising probabilistic estimated value；Wherein
The probability Estimation of left adjacent binary is：
The probability Estimation of right adjacent binary is：
The probability Estimation of adjacent ternary is：
Wherein Wk∈CSet(Wi), Count (Wi-1Wi) represent Wi-1WiThe co-occurrence frequency in language material, Count (WiWi+1)
Represent WiWi+1The co-occurrence frequency in language material, Count (Wi-1WiWi+1) represent Wi-1WiWi+1The co-occurrence frequency in language material,RepresentThe co-occurrence frequency in language material,RepresentIn language material
The co-occurrence frequency,RepresentThe co-occurrence frequency in language material.
It is preferred that, the step 3) comprise the following steps：
Step 31) to the sentence S progress participles of application this method, travel through the word W in the sentence S after participlei, judge whether
There is homonym and obscure collection CSet (Wi), to wherein exist homonym obscure collection word carry out step 32) processing, until sentence
In word be traversed；
Step 32) if WiThere is CSet (Wi), based on step 2) obtained adjacent NGram models of part, utilize Weight
Combined method, context of the homonym of concentration in sentence is obscured by the word and its corresponding homonym that calculate in sentence
Support, judges whether homonym mistake, and homonym mistake is marked and amending advice list is provided, specific bag
Include：
Step 32-1) by combining the word W in scoring function Score calculating sentences SiContext support in sentence
For
Score(Wi)=α1*Pleft(Wi|Wi-1)+α2*Pright(Wi|Wi+1)+α3*Ptri(Wi|Wi-1Wi+1)(4)；
Wherein α1+α2+α3=1, α1>0,α2>0,α3>0, α1、α2、α3Left adjacent binary, right adjacent binary, neighbour are represented respectively
Connect the weight of ternary；
Step 32-2) by combining the word W in scoring function Score calculating sentences SiCorresponding homonym obscures concentration
Each homonymContext support in sentence is
Wherein：
Step 32-3) to WiAnd its in CSet (Wi) in the context support Score of each homonym arranged
Sequence；
Step 32-4) if Score (Wi)=0, then to WiMarked erroneous, and list in Score sequencing tables'sIt is used as amending advice list；Otherwise step 32-5 is turned to)；
Step 32-5) if Score (Wi)>0, andTo Wi
Marked erroneous, is listed file names with It is correspondingAs amending advice list, W is otherwise markediFor
Correct word, wherein β is the wrong probability into its homonym of a word.
It is used as preferred, above-mentioned steps 32-1) and step 32-2) in, the weight α of left adjacent binary1=0.25, right adjoining
The weight α of binary2=0.25, the weight α of adjacent ternary3=0.5.
It is preferred that, the step 32-5) in word it is wrong into probability values≤0.01 of its homonym.
Beneficial effect：The present invention proposes a kind of auto-collation of Chinese homonym mistake, and this method uses unisonance
Word obscures collection and left adjacent binary, right adjacent binary and the adjacent triple combination method of Weight are carried out to the homonym in sentence
Judge, recognize homonym mistake, and provide the amending advice of homonym mistake, integrate automatic errordetecting and automatic Proofreading.
Experiment shows that the method recall rate for the homonym mistake automatic Proofreading that the present invention is provided reaches 81.2%, and precision reaches 75.6%,
Faster system response, precision meet practical application request, and validity and accuracy are high, with higher practicality.
Brief description of the drawings
Fig. 1 homonym mistake automatic Proofreading flow charts.
Embodiment
The present invention is further described with reference to the accompanying drawings and examples.
A kind of Chinese homonym mistake auto-collation that the present invention is provided obscures collection and Weight based on homonym
Local adjacent NGram models combination determining method carries out Chinese homonym mistake automatic Proofreading, and this method comprises the following steps： 1)、
Set up homonym and obscure collection, to Chinese word by phonetic transcriptions of Chinese characters, the homonym for setting up Chinese word obscures collection.
As shown in figure 1, using phonetic transcriptions of Chinese characters table and Chinese dictionary, generation homonym obscures collection：
Wherein WiIt is a Chinese word,It is WiHomonym.
In the present embodiment homonym obscure collection source include two parts：Automatic identification part and artificial check and correction part；
Wherein automatic identification part comprises the following steps：
Step 11) Chinese dictionary is read, the Chinese word in Chinese dictionary is read into Chinese word structure；
Step 12) phonetic transcriptions of Chinese characters of phonetic transcriptions of Chinese characters table is read in into phonetic transcriptions of Chinese characters structure；
Step 13) combine step 11) obtained Chinese word and step 12) obtained phonetic transcriptions of Chinese characters, by Chinese dictionary
Chinese word is converted into phonetic, is put into generation homonym dictionary configuration, i.e. homonym in unisonance word structure and obscures collection；
Wherein artificial check and correction part includes：To step 13) obtained homonym obscures collection and manually proofreaded, renewal unisonance
Word obscures collection.
2), set up that the second from left is first, right binary and ternary local adjacent NGram models, based on step 1) obtained homonym
Obscure collection, probability Estimation is carried out to the local adjacent NGram models by Probabilistic estimation, trained by large-scale corpus
Obtain local adjacent NGram models.Specially：
Step 21) extensive Web language materials are based on, set up the local adjacent of left adjacent binary, right adjacent binary and adjacent ternary
Connect NGram models：Participle is carried out to the sentence in language material, such as carrying out participle to sentence L obtains L=W1W2..Wi-1WiWi+1…
Wn, for word WiFor,
Left adjacent binary is：LeftBiGram(Wi)=Wi-1Wi；
Right adjacent binary is：RightBiGram(Wi)=WiWi+1；
Abutting ternary is：TriGram(Wi)=Wi-1WiWi+1；
Step 22) collection CSet (W are obscured based on extensive Web language materials and homonymi), it is all that statistics homonym obscures concentration
Left adjacent binary and its co-occurrence frequency of word, right adjacent binary and its co-occurrence frequency and adjacent ternary and its co-occurrence frequency, its
In It is WiHomonym；
Step 23) collection CSet (W are obscured based on homonymi), to the office of left adjacent binary, right adjacent binary and adjacent ternary
Portion's adjoining NGram models carry out probability Estimation, so as to generate the adjacent NGram models of the part comprising probabilistic estimated value；Wherein
The probability Estimation of left adjacent binary is：
The probability Estimation of right adjacent binary is：
The probability Estimation of adjacent ternary is：
Wherein Wk∈CSet(Wi), Count (Wi-1Wi) represent Wi-1WiThe co-occurrence frequency in language material, Count (WiWi+1)
Represent WiWi+1The co-occurrence frequency in language material, Count (Wi-1WiWi+1) represent Wi-1WiWi+1The co-occurrence frequency in language material,Represent Wi- The co-occurrence frequency in language material,RepresentWi+1In language material
The co-occurrence frequency,RepresentThe co-occurrence frequency in language material.
3), based on step 2) obtained adjacent NGram models of part, using the combined method of Weight, by calculating sentence
Word and its corresponding homonym in son obscure context support of the homonym of concentration in sentence, judge whether same
Sound word mistake, and homonym mistake is marked and amending advice list is provided.As shown in figure 1, being specially：
Step 31) to the sentence S progress participles of application this method, travel through the word W in the sentence S after participlei, judge whether
There is homonym and obscure collection CSet (Wi), to wherein exist homonym obscure collection word carry out step 32) processing, until sentence
In word be traversed；
Step 32) if WiThere is CSet (Wi), based on step 2) obtained adjacent NGram models of part, utilize Weight
Combined method, context of the homonym of concentration in sentence is obscured by the word and its corresponding homonym that calculate in sentence
Support, judges whether homonym mistake, and homonym mistake is marked and amending advice list is provided, specific bag
Include：
Step 32-1) by combining the word W in scoring function Score calculating sentences SiContext support in sentence
For
Score(Wi)=α1*Pleft(Wi|Wi-1)+α2*Pright(Wi|Wi+1)+α3*Ptri(Wi|Wi-1Wi+1)(4)；
Wherein α1+α2+α3=1, α1>0,α2>0,α3>0, α1、α2、α3Left adjacent binary, right adjacent binary, neighbour are represented respectively
Connect the weight of ternary；In the present embodiment, α1=α2=0.25, α3=0.5, naturally it is also possible to suitably adjusted according to actual needs
It is whole.
Step 32-2) by combining the word W in scoring function Score calculating sentences SiCorresponding homonym obscures concentration
Each homonymContext support in sentence is
Wherein
Step 32-3) to WiAnd its in CSet (Wi) in the context support Score of each homonym arranged
Sequence；
Step 32-4) if Score (Wi)=0, then to WiMarked erroneous, and list in Score sequencing tables'sIt is used as amending advice list；Otherwise step 32-5 is turned to)；
Step 32-5) if Score (Wi)>0, and To Wi
Marked erroneous, is listed file names with It is correspondingAs amending advice list, W is otherwise markediFor
Correct word, wherein β is the wrong probability into its homonym of a word, usual β≤0.01, β=0.01 in the present embodiment.
Experiment：
Repeatedly open test is lived through, experiment is used in the testing material of 10,000 row sentences, manual construction language material sentence
At homonym error 6 00, the parameter given using in embodiment is experiment parameter.Experiment shows that the homonym that the present invention is provided is wrong
The method recall rate of automatic Proofreading reaches 81.2% by mistake, and precision reaches 75.6%.This precision has exceeded prior art, reaches
The demand of practical application, with higher validity and accuracy.
It is only presently preferred embodiments of the present invention to implement row above, does not constitute restriction to the present invention, relevant staff is not
Deviate in the range of the technology of the present invention thought, any modification, equivalent substitution and improvements carried out etc. all fall within the guarantor of the present invention
In the range of shield.
Claims (5)
1. a kind of Chinese homonym mistake auto-collation, it is characterised in that the office of collection and Weight is obscured based on homonym
Adjoining NGram model combination determining methods in portion carry out Chinese homonym mistake automatic Proofreading, and this method comprises the following steps：
1) by phonetic transcriptions of Chinese characters, the homonym for setting up Chinese word obscures collection；
2) set up that the second from left is first, right binary and ternary local adjacent NGram models, based on step 1) obtained homonym obscures
Collection, carries out probability Estimation to the local adjacent NGram models by Probabilistic estimation, is obtained by large-scale corpus training
Local adjacent NGram models；
3) it is based on step 2) obtained adjacent NGram models of part, using the combined method of Weight, by calculating in sentence
Word and its corresponding homonym obscure context support of the homonym of concentration in sentence, judge whether that homonym is wrong
Miss, and homonym mistake is marked and amending advice list is provided；
Wherein described step 2) comprise the following steps：
Step 21) extensive Web language materials are based on, the part for setting up left adjacent binary, right adjacent binary and adjacent ternary is adjacent
NGram models：Participle is carried out to the sentence in language material, such as carrying out participle to sentence L obtains L=W1W2..Wi-1WiWi+1…Wn, it is right
In word WiFor,
Left adjacent binary is：LeftBiGram(Wi)=Wi-1Wi；
Right adjacent binary is：RightBiGram(Wi)=WiWi+1；
Abutting ternary is：TriGram(Wi)=Wi-1WiWi+1；
Step 22) collection CSet (W are obscured based on extensive Web language materials and homonymi), statistics homonym, which is obscured, concentrates all words
Left adjacent binary and its co-occurrence frequency, right adjacent binary and its co-occurrence frequency and adjacent ternary and its co-occurrence frequency, wherein It is WiHomonym；
Step 23) collection CSet (W are obscured based on homonymi), to the local adjacent of left adjacent binary, right adjacent binary and adjacent ternary
Connect NGram models and carry out probability Estimation, so as to generate the adjacent NGram models of the part comprising probabilistic estimated value；Wherein left adjoining
The probability Estimation of binary is：
The probability Estimation of right adjacent binary is：
The probability Estimation of adjacent ternary is：
WhereinCount(Wi-1Wi) represent Wi-1WiThe co-occurrence frequency in language material, Count (WiWi+1) represent
WiWi+1The co-occurrence frequency in language material, Count (Wi-1WiWi+1) represent Wi-1WiWi+1The co-occurrence frequency in language material,RepresentThe co-occurrence frequency in language material,RepresentIn language material
The co-occurrence frequency,RepresentThe co-occurrence frequency in language material；
Wherein described step 3) comprise the following steps：
Step 31) to the sentence S progress participles of application this method, travel through the word W in the sentence S after participlei, judge whether
Homonym obscures collection CSet (Wi), to wherein exist homonym obscure collection word carry out step 32) processing, until in sentence
Word has been traversed；
Step 32) if WiThere is CSet (Wi), based on step 2) obtained adjacent NGram models of part, utilize the group of Weight
Conjunction method, obscures context of the homonym of concentration in sentence by the word and its corresponding homonym that calculate in sentence and supports
Degree, judges whether homonym mistake, and homonym mistake is marked and amending advice list is provided, and specifically includes：
Step 32-1) by combining the word W in scoring function Score calculating sentences SiContext support in sentence is
Score(Wi)=α1*Pleft(Wi|Wi-1)+α2*Pright(Wi|Wi+1)+α3*Ptri(Wi|Wi-1Wi+1) (4)；
Wherein α1+α2+α3=1, α1>0,α2>0,α3>0, α1、α2、α3Left adjacent binary, right adjacent binary, adjacent three are represented respectively
The weight of member；
Step 32-2) by combining the word W in scoring function Score calculating sentences SiCorresponding homonym obscures each of concentration
Individual homonymContext support in sentence is
Wherein
Step 32-3) to WiAnd its in CSet (Wi) in the context support Score of each homonym be ranked up；
Step 32-4) if Score (Wi)=0, then to WiMarked erroneous, and list in Score sequencing tables'sIt is used as amending advice list；Otherwise step 32-5 is turned to)；
Step 32-5) if Score (Wi)>0, andTo WiMark is wrong
By mistake, list file names withIt is correspondingAs amending advice list, W is otherwise markediTo be correct
Word, wherein β are the wrong probability into its homonym of a word.
2. Chinese homonym mistake auto-collation according to claim 1, it is characterised in that the step 1) include：
Using phonetic transcriptions of Chinese characters table and Chinese dictionary, generation homonym obscures collection：
Wherein WiIt is a Chinese word,It is WiHomonym.
3. Chinese homonym mistake auto-collation according to claim 1, it is characterised in that the step 1) in it is same
The source that sound word obscures collection includes two parts：Automatic identification part and artificial check and correction part；
Wherein automatic identification part comprises the following steps：
Step 11) Chinese dictionary is read, the Chinese word in Chinese dictionary is read into Chinese word structure；
Step 12) phonetic transcriptions of Chinese characters of phonetic transcriptions of Chinese characters table is read in into phonetic transcriptions of Chinese characters structure；
Step 13) combine step 11) obtained Chinese word and step 12) obtained phonetic transcriptions of Chinese characters, by the Chinese in Chinese dictionary
Word is converted into phonetic, is put into generation homonym dictionary configuration, i.e. homonym in unisonance word structure and obscures collection；
Wherein artificial check and correction part includes：To step 13) obtained homonym obscures collection and manually proofreaded, and updates homonym mixed
Confuse collection；
The structure that the homonym obscures collection is as follows：
Wherein WiIt is a word,It is WiHomonym.
4. Chinese homonym mistake auto-collation according to claim 1, it is characterised in that：The step 32-1)
With step 32-2) in, the weight α of left adjacent binary1=0.25, the weight α of right adjacent binary2=0.25, the weight of adjacent ternary
α3=0.5.
5. Chinese homonym mistake auto-collation according to claim 1, it is characterised in that：The step 32-5)
In word it is wrong into probability values≤0.01 of its homonym.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510354692.4A CN105045778B (en) | 2015-06-24 | 2015-06-24 | A kind of Chinese homonym mistake auto-collation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510354692.4A CN105045778B (en) | 2015-06-24 | 2015-06-24 | A kind of Chinese homonym mistake auto-collation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105045778A CN105045778A (en) | 2015-11-11 |
| CN105045778B true CN105045778B (en) | 2017-10-17 |
Family
ID=54452334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510354692.4A Active CN105045778B (en) | 2015-06-24 | 2015-06-24 | A kind of Chinese homonym mistake auto-collation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105045778B (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105573979B (en) * | 2015-12-10 | 2018-05-22 | 江苏科技大学 | A kind of wrongly written character word knowledge generation method that collection is obscured based on Chinese character |
| CN106528616B (en) * | 2016-09-30 | 2019-12-17 | 厦门快商通科技股份有限公司 | Language error correction method and system in human-computer interaction process |
| CN109388252B (en) * | 2017-08-14 | 2022-10-04 | 北京搜狗科技发展有限公司 | Input method and device |
| CN110083819B (en) * | 2018-01-26 | 2024-02-09 | 北京京东尚科信息技术有限公司 | Spelling error correction method, device, medium and electronic equipment |
| CN108563634A (en) * | 2018-03-29 | 2018-09-21 | 广州视源电子科技股份有限公司 | Recognition methods, system, computer equipment and the storage medium of word misspelling |
| CN108519973A (en) * | 2018-03-29 | 2018-09-11 | 广州视源电子科技股份有限公司 | Detection method, system, computer equipment and the storage medium of word spelling |
| CN108563632A (en) * | 2018-03-29 | 2018-09-21 | 广州视源电子科技股份有限公司 | Modification method, system, computer equipment and the storage medium of word misspelling |
| CN108491392A (en) * | 2018-03-29 | 2018-09-04 | 广州视源电子科技股份有限公司 | Modification method, system, computer equipment and the storage medium of word misspelling |
| CN108845984B (en) * | 2018-05-22 | 2022-04-22 | 广州视源电子科技股份有限公司 | Wrongly written character detection method and device, computer readable storage medium and terminal equipment |
| CN108984515B (en) * | 2018-05-22 | 2022-09-06 | 广州视源电子科技股份有限公司 | Wrongly written character detection method and device, computer readable storage medium and terminal equipment |
| CN108874770B (en) * | 2018-05-22 | 2022-04-22 | 广州视源电子科技股份有限公司 | Wrongly written character detection method and device, computer readable storage medium and terminal equipment |
| CN108829665B (en) * | 2018-05-22 | 2022-05-31 | 广州视源电子科技股份有限公司 | Wrongly written character detection method and device, computer readable storage medium and terminal equipment |
| CN110600011B (en) * | 2018-06-12 | 2022-04-01 | 中国移动通信有限公司研究院 | Voice recognition method and device and computer readable storage medium |
| CN110619119B (en) * | 2019-07-23 | 2022-06-10 | 平安科技（深圳）有限公司 | Intelligent text editing method and device and computer readable storage medium |
| CN110717021B (en) * | 2019-09-17 | 2023-08-29 | 平安科技（深圳）有限公司 | Input text acquisition and related device in artificial intelligence interview |
| CN110851599B (en) * | 2019-11-01 | 2023-04-28 | 中山大学 | Automatic scoring method for Chinese composition and teaching assistance system |
| CN110991166B (en) * | 2019-12-03 | 2021-07-30 | 中国标准化研究院 | Chinese wrongly-written character recognition method and system based on pattern matching |
| CN111161739B (en) * | 2019-12-28 | 2023-01-17 | 科大讯飞股份有限公司 | Speech recognition method and related product |
| CN111312209A (en) * | 2020-02-21 | 2020-06-19 | 北京声智科技有限公司 | Text-to-speech conversion processing method and device and electronic equipment |
| CN111709228B (en) * | 2020-06-22 | 2023-11-21 | 中国标准化研究院 | Automatic identification method for word repetition errors |
| CN112668328A (en) * | 2020-12-25 | 2021-04-16 | 广东南方新媒体科技有限公司 | Media intelligent proofreading algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102375807A (en) * | 2010-08-27 | 2012-03-14 | 汉王科技股份有限公司 | Method and device for proofing characters |
| US8364487B2 (en) * | 2008-10-21 | 2013-01-29 | Microsoft Corporation | Speech recognition system with display information |
| CN104166462A (en) * | 2013-05-17 | 2014-11-26 | 北京搜狗科技发展有限公司 | Input method and system for characters |
-
2015
- 2015-06-24 CN CN201510354692.4A patent/CN105045778B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8364487B2 (en) * | 2008-10-21 | 2013-01-29 | Microsoft Corporation | Speech recognition system with display information |
| CN102375807A (en) * | 2010-08-27 | 2012-03-14 | 汉王科技股份有限公司 | Method and device for proofing characters |
| CN104166462A (en) * | 2013-05-17 | 2014-11-26 | 北京搜狗科技发展有限公司 | Input method and system for characters |
Non-Patent Citations (2)
| Title |
|---|
| 基于决策列表的中文同音词自动识别与校对;石敏 等;《电子设计工程》;20150531;第23卷(第9期);第39-41页 * |
| 汉字种子混淆集的构建方法研究;施恒利 等;《计算机科学》;20140831;第41卷(第8期);第230-231页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105045778A (en) | 2015-11-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105045778B (en) | A kind of Chinese homonym mistake auto-collation | |
| CN104991889B (en) | A kind of non-multi-character word error auto-collation based on fuzzy participle | |
| Shaalan et al. | Arabic word generation and modelling for spell checking. | |
| CN103970765A (en) | Error correcting model training method and device, and text correcting method and device | |
| CN105824800B (en) | A kind of true word mistake auto-collation of Chinese | |
| Zhang et al. | HANSpeller++: A unified framework for Chinese spelling correction | |
| CN105512110B (en) | A kind of wrongly written character word construction of knowledge base method based on fuzzy matching with statistics | |
| KR101633556B1 (en) | Apparatus for grammatical error correction and method using the same | |
| Ljubešić et al. | Predicting the level of text standardness in user-generated content | |
| Richter et al. | Korektor–a system for contextual spell-checking and diacritics completion | |
| CN108280065B (en) | Foreign text evaluation method and device | |
| CN104346326A (en) | Method and device for determining emotional characteristics of emotional texts | |
| CN106528533A (en) | Dynamic sentiment word and special adjunct word-based text sentiment analysis method | |
| JP6626917B2 (en) | Readability evaluation method and system based on English syllable calculation method | |
| Schneider et al. | Comparing rule-based and SMT-based spelling normalisation for English historical texts | |
| CN106202037B (en) | Vietnamese phrase tree constructing method based on chunking | |
| CN112115701B (en) | News reading text readability evaluation method and system | |
| CN107894977A (en) | With reference to the Vietnamese part of speech labeling method of conversion of parts of speech part of speech disambiguation model and dictionary | |
| Geyken et al. | On-the-fly Generation of Dictionary Articles for the DWDS Website | |
| Schottmüller et al. | Issues in translating verb-particle constructions from german to english | |
| CN111027314A (en) | Character attribute extraction method based on language fragment | |
| US10755594B2 (en) | Method and system for analyzing a piece of text | |
| Ji et al. | Analysis and repair of name tagger errors | |
| Lu et al. | Language model for Mongolian polyphone proofreading | |
| CN104978311B (en) | A kind of Vietnamese segmenting method based on condition random field |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20151111 Assignee: JIANGSU KEDA HUIFENG SCIENCE AND TECHNOLOGY Co.,Ltd. Assignor: JIANGSU University OF SCIENCE AND TECHNOLOGY Contract record no.: X2020980007325 Denomination of invention: An automatic correction method for Chinese homonym errors Granted publication date: 20171017 License type: Common License Record date: 20201029 |
|
| EC01 | Cancellation of recordation of patent licensing contract | ||
| EC01 | Cancellation of recordation of patent licensing contract |
Assignee: JIANGSU KEDA HUIFENG SCIENCE AND TECHNOLOGY Co.,Ltd. Assignor: JIANGSU University OF SCIENCE AND TECHNOLOGY Contract record no.: X2020980007325 Date of cancellation: 20201223 |