CN107832447A - User feedback error correction method, device and its equipment for mobile terminal - Google Patents

User feedback error correction method, device and its equipment for mobile terminal Download PDF

Info

Publication number
CN107832447A
CN107832447A CN201711173999.XA CN201711173999A CN107832447A CN 107832447 A CN107832447 A CN 107832447A CN 201711173999 A CN201711173999 A CN 201711173999A CN 107832447 A CN107832447 A CN 107832447A
Authority
CN
China
Prior art keywords
fragment
participle
original
user
error correction
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.)
Pending
Application number
CN201711173999.XA
Other languages
Chinese (zh)
Inventor
肖求根
詹金波
郑利群
邓卓彬
付志宏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Baidu Netcom Science and Technology Co Ltd
Original Assignee
Beijing Baidu Netcom Science and Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Baidu Netcom Science and Technology Co Ltd filed Critical Beijing Baidu Netcom Science and Technology Co Ltd
Priority to CN201711173999.XA priority Critical patent/CN107832447A/en
Publication of CN107832447A publication Critical patent/CN107832447A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/232—Orthographic correction, e.g. spell checking or vowelisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Machine Translation (AREA)

Abstract

The present invention proposes a kind of user feedback error correction method, device and its equipment for mobile terminal, wherein, method includes：Obtain user and the error correction report request for including the original participle fragment of one or more corresponding with report information is sent by mobile terminal；The fisrt feature information of original participle fragment is carried out calculating the confidence level for obtaining original participle fragment, obtaining candidate corresponding with original participle fragment according to phrase substitution table when confidence level is less than predetermined threshold value segments fragment；The second feature information that fragment is segmented to original participle fragment and corresponding candidate carries out calculating the score value for obtaining candidate and segmenting fragment；The score value that fragment is segmented to candidate carries out decoding process, and participle fragment will be decoded when decoding participle fragment corresponding to decoded result and meeting default intervention condition as original participle fragment corresponding to target participle fragment replacement.Thus, user carries out correction process by way of mobile terminal feedback, and the accuracy of error correction is improved while reducing running cost.

Description

User feedback error correction method, device and its equipment for mobile terminal
Technical field
The present invention relates to technical field of information processing, more particularly to a kind of user feedback error correction side for mobile terminal Method, device and its equipment.
Background technology
Artificial intelligence (Artificial Intelligence), english abbreviation AI.It is research, develop for simulating, Extension and the extension intelligent theory of people, method, a new technological sciences of technology and application system.Artificial intelligence is to calculate One branch of machine science, it attempts to understand essence of intelligence, and produce it is a kind of it is new can be in a manner of human intelligence be similar The intelligence machine made a response, the research in the field include robot, speech recognition, image recognition, natural language processing and specially Family's system etc..
At present, content distribution is the important battlefield of mobile Internet company, and high-quality content tends to bring user to stop Stay the lifting of duration and brand reputation.Wherein, hard defects of the wrong word as content, any one may be appeared in certain probability In piece article.In correlation technique, by that without fixed report mode, the increase of model learning difficulty can be caused, do not sought unity of standard, increased Add error correction subsequent operation cost, and the problems such as training pattern may not be suitable for mobile terminal.
The content of the invention
The purpose of the present invention is intended to one of technical problem at least solving in correlation technique to a certain extent.
Therefore, first purpose of the present invention is to propose a kind of user feedback error correction method for mobile terminal, use In solving not seek unity of standard in the prior art, increase error correction subsequent operation cost, and training pattern may not be suitable for moving The problems such as dynamic terminal.
Second object of the present invention is to propose another user feedback error correction method for being used for mobile terminal.
Third object of the present invention is to propose a kind of user feedback error correction device for mobile terminal.
Fourth object of the present invention is to propose another user feedback error correction device for being used for mobile terminal.
The 5th purpose of the present invention is to propose a kind of computer equipment.
The 6th purpose of the present invention is to propose a kind of non-transitorycomputer readable storage medium.
The 7th purpose of the present invention is to propose a kind of computer program product.
For the above-mentioned purpose, first aspect present invention embodiment proposes a kind of user feedback error correction for mobile terminal Method, it the described method comprises the following steps：Obtain the error correction that user is sent by mobile terminal and report request, wherein, it is described to ask Ask including：Text message belonging to report information, and the original participle fragment of one or more corresponding with the report information, Wherein, the mobile terminal text message is normalized according to the participle database being locally stored and word segmentation processing, It is determined that the original participle fragment of one or more corresponding with the report information；Extract one or more of original participle fragments Fisrt feature information, using preset model to the fisrt feature information carry out calculate obtain it is described it is original participle fragment put Reliability, if judging to know that the confidence level is less than predetermined threshold value, obtained and the original according to the phrase substitution table pre-established Begin one or more candidates' participle fragments corresponding to participle fragment；Piece is segmented according to the original participle fragment and corresponding candidate Section extraction second feature information, the second feature information is carried out using preset model to calculate acquisition candidate's participle fragment Score value；The score value for segmenting fragment to the candidate using default decoding algorithm carries out decoding process, if judging to know and solve Decoding participle fragment meets default intervention condition corresponding to code result, then the decoding is segmented into fragment segments piece as target Original participle fragment corresponding to section replacement.
The user feedback error correction method for mobile terminal of the embodiment of the present invention, passes through mobile terminal by obtaining user The error correction report request of transmission, wherein, request includes：Text message belonging to report information, and it is corresponding with report information One or more original participle fragments, wherein, mobile terminal is returned according to the participle database being locally stored to text message One change and word segmentation processing, it is determined that the original participle fragment of one or more corresponding with report information, then extraction is one or more The fisrt feature information of original participle fragment, and application preset model to fisrt feature information calculate and obtains original participle piece The confidence level of section, obtained and original according to the phrase substitution table pre-established when judging to know that the confidence level is less than predetermined threshold value Begin one or more candidates' participle fragments corresponding to participle fragment, so as to segment piece according to original participle fragment and corresponding candidate Section extraction second feature information, and application preset model to second feature information carry out calculate obtain candidate segment fragment divide Value, the score value for finally segmenting fragment to candidate using default decoding algorithm carry out decoding process, tied judging to know with decoding Participle fragment is decoded corresponding to fruit to meet decoding is segmented into fragment as target participle fragment replacement pair during default intervention condition The original participle fragment answered.Thus, user carries out correction process by way of mobile terminal feedback, and in mobile terminal sheet Ground determines original participle fragment corresponding with report information, unified standard, the accurate of error correction is improved while reducing running cost Property, meet user's request.
For the above-mentioned purpose, second aspect of the present invention embodiment proposes another user feedback for being used for mobile terminal and entangled Wrong method, the described method comprises the following steps：Selection operation of the user to report information in text message is obtained, according to the shifting The text message is normalized the participle database of dynamic terminal storage and word segmentation processing, and the text is shown to the user The participle fragment interface of this information；Obtain user in the participle fragment interface pair one corresponding with the report information or The selection operation of multiple original participle fragments, and obtain the user and error correction report functional entrance in current application interface is touched Hair operation, and then error correction report request is sent to server, wherein, the request includes：The text message, and with it is described One or more original participle fragments corresponding to report information；Being used for of obtaining that the server sends replaces the original participle The target participle fragment of fragment, and it is shown to the user.
The user feedback error correction method for mobile terminal of the embodiment of the present invention, by obtaining user in text message The selection operation of report information, according to mobile terminal store participle database text message is normalized and participle at Reason, the participle fragment interface of text message is shown to user, it is pair corresponding with report information in participle fragment interface to obtain user It is one or more it is original participle fragments selection operation, and obtain user in current application interface error correction report functional entrance Trigger action, and then to server send error correction report request, wherein, request includes：Text message, and and report information The corresponding original participle fragment of one or more, so as to obtain the target point for being used to replace original participle fragment of server transmission Word fragment, and it is shown to user.Thus, user carries out correction process by way of mobile terminal feedback, reduces running cost While improve error correction accuracy, meet user's request.
For the above-mentioned purpose, third aspect present invention embodiment proposes a kind of user feedback error correction for mobile terminal Device, described device include：Acquisition module, obtain the error correction that user is sent by mobile terminal and report request, wherein, it is described to ask Ask including：Text message belonging to report information, and the original participle fragment of one or more corresponding with the report information, Wherein, the mobile terminal text message is normalized according to the participle database being locally stored and word segmentation processing, It is determined that the original participle fragment of one or more corresponding with the report information；First processing module, it is one for extracting Or the fisrt feature information of multiple original participle fragments, the fisrt feature information is carried out using preset model to calculate acquisition institute The confidence level of original participle fragment is stated, if judging to know that the confidence level is less than predetermined threshold value, according to the phrase pre-established Substitution table obtains one or more candidates participle fragments corresponding with the original participle fragment；Computing module is extracted, for root Snippet extraction second feature information is segmented according to the original participle fragment and corresponding candidate, using preset model to described second Characteristic information, which calculate, obtains the score value that the candidate segments fragment；Second processing module, calculated for the default decoding of application The score value that method segments fragment to the candidate carries out decoding process, if judging to know decoding participle fragment corresponding with decoded result Meet default intervention condition, then the decoding is segmented into fragment as original participle piece corresponding to target participle fragment replacement Section.
The user feedback error correction device for mobile terminal of the embodiment of the present invention, passes through mobile terminal by obtaining user The error correction report request of transmission, wherein, request includes：Text message belonging to report information, and it is corresponding with report information One or more original participle fragments, wherein, mobile terminal is returned according to the participle database being locally stored to text message One change and word segmentation processing, it is determined that the original participle fragment of one or more corresponding with report information, then extraction is one or more The fisrt feature information of original participle fragment, and application preset model to fisrt feature information calculate and obtains original participle piece The confidence level of section, obtained and original according to the phrase substitution table pre-established when judging to know that the confidence level is less than predetermined threshold value Begin one or more candidates' participle fragments corresponding to participle fragment, so as to segment piece according to original participle fragment and corresponding candidate Section extraction second feature information, and application preset model to second feature information carry out calculate obtain candidate segment fragment divide Value, the score value for finally segmenting fragment to candidate using default decoding algorithm carry out decoding process, tied judging to know with decoding Participle fragment is decoded corresponding to fruit to meet decoding is segmented into fragment as target participle fragment replacement pair during default intervention condition The original participle fragment answered.Thus, user carries out correction process by way of mobile terminal feedback, and in mobile terminal sheet Ground determines original participle fragment corresponding with report information, unified standard, the accurate of error correction is improved while reducing running cost Property, meet user's request.
For the above-mentioned purpose, fourth aspect present invention embodiment proposes another user feedback for being used for mobile terminal and entangled Misloading is put, and described device includes：Display module is obtained, for obtaining selection operation of the user to report information in text message, The text message is normalized the participle database stored according to the mobile terminal and word segmentation processing, to the user Show the participle fragment interface of the text message；Sending module is obtained, for obtaining user in the participle fragment interface Selection operation pair the original participle fragment of one or more corresponding with the report information, and the user is obtained to currently should The trigger action of functional entrance is reported with error correction in interface, and then error correction report request is sent to server, wherein, the request Including：The text message, and the original participle fragment of one or more corresponding with the report information；Display module, use Fragment is segmented in the target for being used to replace the original participle fragment for obtaining the server transmission, and is shown to the use Family.
The user feedback error correction device for mobile terminal of the embodiment of the present invention, by obtaining user in text message The selection operation of report information, according to mobile terminal store participle database text message is normalized and participle at Reason, the participle fragment interface of text message is shown to user, it is pair corresponding with report information in participle fragment interface to obtain user It is one or more it is original participle fragments selection operation, and obtain user in current application interface error correction report functional entrance Trigger action, and then to server send error correction report request, wherein, request includes：Text message, and and report information The corresponding original participle fragment of one or more, so as to obtain the target point for being used to replace original participle fragment of server transmission Word fragment, and it is shown to user.Thus, user carries out correction process by way of mobile terminal feedback, reduces running cost While improve error correction accuracy, meet user's request.
For the above-mentioned purpose, fifth aspect present invention embodiment proposes a kind of computer equipment, including memory, processing Device and storage on a memory and the computer program that can run on a processor, during the computing device described program, reality The now user feedback error correction method for mobile terminal as described in first aspect embodiment and second aspect embodiment.
To achieve these goals, fourth aspect present invention embodiment proposes a kind of computer-readable storage of non-transitory Medium, when the instruction in the storage medium is performed by processor, enabling perform first aspect embodiment and second The user feedback error correction method for mobile terminal described in aspect embodiment.
To achieve these goals, fifth aspect present invention embodiment proposes a kind of computer program product, when described When instruction processing unit in computer program product performs, the use described in execution first aspect embodiment and second aspect embodiment In the user feedback error correction method of mobile terminal.
The additional aspect of the present invention and advantage will be set forth in part in the description, and will partly become from the following description Obtain substantially, or recognized by the practice of the present invention.
Brief description of the drawings
Of the invention above-mentioned and/or additional aspect and advantage will become from the following description of the accompanying drawings of embodiments Substantially and it is readily appreciated that, wherein：
Fig. 1 is that mobile terminal according to an embodiment of the invention selects word exemplary plot；
Fig. 2 is the flow signal of the user feedback error correction method according to an embodiment of the invention for mobile terminal Figure；
Fig. 3 is that the search term of collection user input according to an embodiment of the invention and the mapping wantonly searched between title are believed The exemplary plot of breath；
Fig. 4 is that the search term of collection user input according to an embodiment of the invention and the error correction that search engine provides are believed The exemplary plot of breath；
Fig. 5 is according to an embodiment of the invention the exemplary plot that correct sequence recalls to be completed by decoding process；
Fig. 6 is the exemplary plot of phonetic mapping table according to an embodiment of the invention；
Fig. 7 is the flow signal of the user feedback error correction method in accordance with another embodiment of the present invention for mobile terminal Figure；
Fig. 8 is illustrated according to the flow of the user feedback error correction method for mobile terminal of another embodiment of the invention Figure；
Fig. 9 is applied customization layer in the user feedback error correction method according to an embodiment of the invention for mobile terminal Schematic diagram；
Figure 10 is shown according to the flow of the user feedback error correction method for mobile terminal of further embodiment of the present invention It is intended to；
Figure 11 is the exemplary plot of the user feedback error correction of mobile terminal according to an embodiment of the invention；
Figure 12 is the exemplary plot of the user feedback error correction of mobile terminal in accordance with another embodiment of the present invention；
Figure 13 is the structural representation of the user feedback error correction device according to an embodiment of the invention for mobile terminal Figure；
Figure 14 is that the structure of the user feedback error correction device in accordance with another embodiment of the present invention for mobile terminal is shown It is intended to；
Figure 15 is shown according to the structure of the user feedback error correction device for mobile terminal of another embodiment of the invention It is intended to；
Figure 16 is the structural representation of computer equipment according to an embodiment of the invention.
Embodiment
Embodiments of the invention are described below in detail, the example of the embodiment is shown in the drawings, wherein from beginning to end Same or similar label represents same or similar element or the element with same or like function.Below with reference to attached The embodiment of figure description is exemplary, it is intended to for explaining the present invention, and is not considered as limiting the invention.
Below with reference to the accompanying drawings describe the embodiment of the present invention for the user feedback error correction method of mobile terminal, device and its Equipment.
Generally, Chinese error correction has many types, such as wrong word mistake, syntax error, semantic error correction etc..Wherein, it is wrong Malapropism error correction is most easily understood by, and Chinese character commonly used word has 5000, about 560,000 everyday words, and these words and word are even correct shape Formula, ill-formalness have that multiword, wrongly written character, hiatus etc. are more changeable, and search space is impossible to exhaust again.In correlation technique, Yong Hu Mobile terminal carries out participle and selects word, as shown in figure 1, after user selects word, clicks on report and following two situations occur：For example use Family selects single word " Cheng " to be reported；User selects multiple words " anti-Cheng " for another example, either " meets anti-Cheng " or " anti-Cheng is high Peak ", reported, avoid " anti-Cheng is high " and " Cheng is high " two kinds of forms.Thus, do not seek unity of standard, increase error correction is subsequently grasped Make cost, and the problems such as training pattern may not be suitable for mobile terminal.
For this problem, the embodiments of the invention provide a kind of user feedback error correction method for mobile terminal, uses Family carries out correction process by way of mobile terminal feedback, and locally determines original corresponding with report information in mobile terminal Begin participle fragment, unified standard, the accuracy of error correction is improved while reducing running cost, meets user's request.It is specific as follows：
Fig. 2 is the flow signal of the user feedback error correction method according to an embodiment of the invention for mobile terminal Figure.As shown in Fig. 2 the user feedback error correction method for being used for mobile terminal includes：
Step 101, obtain the error correction that user is sent by mobile terminal and report request, wherein, request includes：Report information Affiliated text message, and the original participle fragment of one or more corresponding with report information, wherein, mobile terminal is according to this Text message is normalized the participle database of ground storage and word segmentation processing, it is determined that one corresponding with report information or more Individual original participle fragment.
Among practical application, user can be sent as desired by modes such as long-press, clicks by mobile terminal to be entangled Mistake report request.Can wherein it is possible to carry out parsing to error correction report request by modes such as default related algorithm or models To obtain the text message and the original participle fragment of one or more corresponding with report information belonging to report information.
For example, user is transmitted error correction report request by long-press " being violent ", can get report information " hair The text message of 9 words belonging to hurricane " is " general manager be violent Leading Speaches ", and original participle fragment corresponding to " being violent " " being violent " etc..
Specifically, mobile terminal can be carried out after locally taking a variety of modes that text message is normalized Participle, for example can remove one or more kinds of normalization in afterbody carriage-return character, either traditional and simplified characters, capital and small letter and full half-angle etc. Processing.
Further, the text message after normalized is segmented, it is to be understood that the form of participle has very It is a variety of, it can need to be segmented according to practical application.As a kind of example, text message is " general manager be violent Leading Speaches " It can be divided into " general manager ", " being violent ", " important " and " speech "；As another example, text message is " general manager is violent again Talk " it can be divided into " total ", " manager ", " hair ", " hurricane ", " important " and " speech " etc..
Further, by taking the participle of the first example as an example, an original participle fragment corresponding with report information is " hair Hurricane "；By taking the participle of second of example as an example, an original participle fragment corresponding with report information is " hair " and " hurricane ".
Wherein, in order to further meet error correction demand, mobile terminal can also carry out phonetic notation to text message, by initial consonant rhythm Mother is mapped to corresponding original participle piece fragment position.
Step 102, the fisrt feature information of the one or more original participle fragments of extraction, it is special to first using preset model Reference breath carries out calculating the confidence level for obtaining original participle fragment, if judging to know that confidence level is less than predetermined threshold value, according to pre- The phrase substitution table first established obtains one or more candidates corresponding with original participle fragment and segments fragment.
It is understood that before the error correction report request that user sends is obtained, training in advance and the mould for storing correlation Type can carry out the searching and positioning of false segments.
Specifically, original participle fragment has a variety of fisrt feature information, is illustrated below：
The first example, the frequency of occurrence of original participle fragment and context in language material.
Second of example, the change frequency of original participle fragment and context in application scenarios are searched for.
The semantic similarity of the third example, original participle fragment and context.Such as the distance of term vector.
Can be needed to select according to practical application it is above-mentioned in one or more as fisrt feature information, and application is pre- If model carries out calculating the confidence level for obtaining original participle fragment to fisrt feature information.
It is understood that when knowing that confidence level is more than predetermined threshold value, can directly recognize without follow-up correction process It is set to user's report etc. by mistake.When knowing that confidence level is less than predetermined threshold value, can be obtained according to the phrase substitution table pre-established One or more candidates corresponding with original participle fragment are taken to segment fragment.Wherein, predetermined threshold value can be according to practical application need Carry out selection setting.That is confidence level is lower, represents that current error correction report request is more effective.
Wherein it is possible to using statistical information knots such as the active acts of revision for retrieving end in a search engine from user Matched moulds type carries out erroneous point judgement.Specifically, find a rational phrase substitution table (phrase table, PT table) with up to Only possible common type of error is covered to a less form, form is replaced for whether being sent out in text message according to this part Raw mistake is differentiated.
Wherein, phrase substitution table is pre-set, the change conversion frequency predominantly between fragment, for recording a fragment The number of itself and other fragments is snapped to, can be obtained by a variety of modes, as a kind of example, collection user exists Active modification information in search engine to search term, the search term of collection user's input and the mapping wantonly searched between title are believed Breath, the search term and the error correction information of search engine offer of collection user's input, according to active modification information, map information and entangles Wrong information establishes the phrase substitution table.
Specifically, user's active modification information to search term, such as in Baidu search in a search engine, pin are gathered The big data for actively changing user search term behavior counts, and obtains segmenting the change frequency between fragment.Such as：User Continuous input " blue or green Hua Da ", " Tsing-Hua University ", we obtain " blue or green China->Tsing-Hua University ", " blue or green Hua Da->Tsing-Hua University ", it is " big Learn->University ".
Specifically, the map information for gathering the search term of user's input and wantonly searching between title, such as in Baidu search, The search term of user's input aligns mapping with wantonly searching for the fragment between title, the change frequency between obtained participle fragment, example As shown in Figure 3.
Specifically, the error correction information that the search term of user's input provides with search engine is gathered, such as in Baidu search, User feedback data between the search term of user's input and Baidu active error correction is alignd.User is adopted into error correction result, with original The mistake that begins input is alignd, and excavation obtains segmenting fragment alignment information.Such as shown in Fig. 4：The error correction of insertion prompting form, when User clicks the error correction result of recommendation, i.e., can carry out alignment excavation.
Step 103, snippet extraction second feature information is segmented according to original participle fragment and corresponding candidate, using default Model carries out calculating the score value for obtaining candidate and segmenting fragment to second feature information.
Step 104, the score value for segmenting fragment to candidate using default decoding algorithm carries out decoding process, if judging to know Decoding participle fragment corresponding with decoded result meets default intervention condition, then segments piece using decoding participle fragment as target Original participle fragment corresponding to section replacement.
Specifically, segmenting snippet extraction second feature information according to original participle fragment and corresponding candidate has many kinds, For example candidate segments the qualitative character of fragment, the qualitative character of original participle fragment, original participle fragment and candidate and segments fragment Assemblage characteristic, active user historical behavior feature in one or more combinations of features.
Wherein, candidate segments the qualitative character of fragment：The main frequency for including candidate segment in itself, the context of the fragment Feature, characteristic statisticses language material source are long text, and whether proper name feature；The qualitative character of original participle fragment：Main bag Including candidate and segment the frequency of fragment in itself, the contextual feature of the fragment, characteristic statisticses language material source is long text, and whether Proper name feature；Original participle fragment and candidate segment the assemblage characteristic of fragment：Candidate segments frequency of the fragment with original participle fragment Secondary ratio, candidate segment historical behavior aspect ratio of the fragment with original participle fragment, in the contextual feature ratio of text information, two Semantic similarity degree of person etc.；The historical behavior feature of active user：The language material of statistical nature comes from phrase substitution table.
For example original participle fragment ori has tri- candidates of A, B, C to segment fragment and itself, in original participle fragment and candidate Pair { ori, A } { ori, B } { ori, C } pair { ori, ori } is formed between participle fragment, for each pair, by A spy Reference breath subtracts corresponding ori feature, the Characterizations as A.
Further, second feature information is carried out using preset model calculating the score value for obtaining candidate and segmenting fragment, with And the default decoding algorithm of application segments the score value progress decoding process of fragment to candidate.That is, segment fragment in candidate The height of score value be not optimal selection, it is also necessary to pass through such as viterbi algorithm (viterbi), beam search (beam Search), or greed searches for decoding algorithms such as (greedy search) and the score value of candidate's participle fragment is carried out at decoding Reason, decoding participle fragment meets that default intervention condition carries out final choice according to corresponding to decoded result.
Wherein, after different fragments are recalled, many fragment results can be obtained, there are multiple combinations may.I.e. a variety of The candidate of section, fragment candidate network is formed, will when decoding participle fragment meets default intervention condition corresponding to decoded result Decoding participle fragment is as original participle fragment corresponding to target participle fragment replacement.As a kind of example, as shown in figure 5, logical The correct sequence of decoding process completion is crossed to recall " this master worker is faster and betterly dry ".
In summary, the user feedback error correction method for mobile terminal of the embodiment of the present invention, is led to by obtaining user The error correction report request of mobile terminal transmission is crossed, wherein, request includes：Text message belonging to report information, and with report One or more original participle fragments corresponding to information, wherein, mobile terminal is according to the participle database being locally stored to text Information is normalized and word segmentation processing, it is determined that the original participle fragment of one or more corresponding with report information, is then extracted The fisrt feature information of one or more original participle fragments, and calculating acquisition is carried out to fisrt feature information using preset model The confidence level of original participle fragment, is replaced when judging to know that the confidence level is less than predetermined threshold value according to the phrase pre-established Table obtains one or more candidates corresponding with original participle fragment and segments fragments, so as to according to the original fragment and corresponding of segmenting Candidate segments snippet extraction second feature information, and application preset model to second feature information calculate and obtains candidate point The score value of word fragment, the score value for finally segmenting fragment to candidate using default decoding algorithm carry out decoding process, are judging to obtain Know that decoding participle fragment corresponding with decoded result meets to segment using decoding participle fragment as target during default intervention condition Original participle fragment corresponding to fragment replacement.Thus, user carries out correction process, Yi Ji by way of mobile terminal feedback Mobile terminal locally determines original participle fragment corresponding with report information, unified standard, is improved while reducing running cost The accuracy of error correction, meets user's request.
Based on above-described embodiment, it is to be understood that carried segmenting fragment according to original participle fragment and corresponding candidate Before taking second feature information, in addition to：Obtained and original participle fragment corresponding one according to the phonetic substitution table pre-established Individual or multiple candidates segment fragment.
Wherein, phonetic substitution table is from the phonetic notation pinyin string of text message, by mixing the double sides deleted of the initial and the final Method recalls candidate and segments fragment, obtains error correction candidate and segments fragment, as a kind of example, segments statistical result by language material, takes HFS, phonetic notation is carried out, inverted index is carried out by phonetic.Wherein, recalled to expand, part can be carried out to the initial and the final Deletion is indexed.Such as " China ", phonetic notation are " zhonghua ", generating key-value (keyword) according to the initial and the final is " zhonghua ", " zhhua ", " onghua ", " zhongua ", " zhongh " } _ -->{ " China " }.
As another example, by recalling result from spelling input method, it is accustomed to according to the conventional key entry of user, to work as The initial and the final sequential system of preceding word is recalled, and " zhonghua " " zhongh ", " zhhua " obtains the candidate of spelling input method Word list.It can also introduce to obscure sound and be enlarged and recall result according to application, such as the mapping table shown in Fig. 6.
Thus, one or more candidates point corresponding with original participle fragment are obtained according to the phonetic substitution table pre-established Word fragment, further reduces running cost, improves the accuracy and efficiency of error correction.
Fig. 7 is the flow signal of the user feedback error correction method in accordance with another embodiment of the present invention for mobile terminal Figure.As shown in fig. 7, after the default decoding algorithm of application segments the score value progress decoding process of fragment to candidate, in addition to：
Step 201, if judging to know that decoding participle fragment corresponding with decoded result is unsatisfactory for default intervention condition, Original participle fragment corresponding to fragment replacement is segmented by Manual definition's target.
Step 202, if judging to know that decoding participle fragment corresponding with decoded result meets default replacement blacklist, It is defined as invalid report, without correction process.
Specifically, fragment intervention is mainly the intervention for completing to replace for false segments and synonym fragment is replaced, and is also propped up Manual errors fragment intervention operation is held, for example by recording whether user feedback error correction is correct, the fragment that can obtain mistake is replaced Change result；Excavated for another example by synonym, invalid error correction replacement can be obtained；It is wrong again such as by human-edited Fragment, which is replaced, by mistake intervenes, and can complete the further tuning of result (for example, the badcase of PM feedbacks intervenes).Thus, further carry The high accuracy and efficiency of error correction, meets application demand of the user under different scenes.
Based on above-described embodiment, by the user feedback error correction method for mobile terminal of the present invention, use can be screened Whether the report at family is that malice is reported, the suggestion function of correct word or word form can be provided after report, for that can not give Go out suggestion, if effective report, error pattern prompting form can be retained as output.Specifically illustrated with reference to Fig. 8 It is described as follows：
As shown in figure 8, carry out determining whether true false segments using the characteristic information of original participle fragment after starting, Progress phrase substitution table is needed to recall, then according to the acquisition of phonetic substitution table and original participle fragment corresponding one pre-established Individual or multiple candidates segment fragment, and then carrying out fragment marking, fragment decoding and fragment intervention, specific implementation process can join Above-described embodiment is seen, in addition, the applied customization layer in the user feedback error correction method for mobile terminal of the present invention can be as Shown in Fig. 9, it is allowed to different applied customization layers, the shared weight for possessing multiplexing value, it is allowed to which sub-function module is flexibly risen Level.Such as cutting word or phonetic notation module etc. are changed, disparate modules can directly plug collocation, combination investigation, increase iteratively faster Ability.
Thus, the positional information that user annotation is false segments is increased, the fragment is fragment containing wrong word by user annotation, And it is wrong other that pt_recall_layer＆ed_recall_layer＆self_recall_layer, which will only occur in user annotation, Word slice fragment position, further increase the accuracy and efficiency of error correction.
Figure 10 is shown according to the flow of the user feedback error correction method for mobile terminal of further embodiment of the present invention It is intended to.As shown in Figure 10, the user feedback error correction method for being used for mobile terminal includes：
Step 301, selection operation of the user to report information in text message, the participle stored according to mobile terminal are obtained Text message is normalized database and word segmentation processing, and the participle fragment interface of text message is shown to user.
Step 302, user's pair original participle of one or more corresponding with report information in participle fragment interface is obtained The selection operation of fragment, and the trigger action that user reports error correction in current application interface functional entrance is obtained, and then to clothes Business device sends error correction report request, wherein, request includes：Text message, and it is corresponding with report information one or more former Begin participle fragment.
Specifically, after user can be chosen by the operation such as long-press, click to report information, mobile terminal can root Take a variety of modes to be segmented after text message is normalized according to the participle database of storage, for example can be The one or more kinds of normalized rear lines removed in afterbody carriage-return character, either traditional and simplified characters, capital and small letter and full half-angle etc. show Show the participle fragment interface of text message.
As a kind of example, report information in user's long-press text message " long holidays draw to an end Fujian Ying FanCheng peaks " " Cheng " carries out selection operation, and so as to the participle database that is stored according to mobile terminal, to text message, " long holidays draw to an end Fujian Ying FanCheng peaks " are normalized and word segmentation processing, and the participle fragment interface of text message is shown to user, specifically such as Shown in Figure 11, so as to obtain user's pair original point of one or more corresponding with report information in participle fragment interface The selection operation of word fragment, and obtain the trigger action that user reports error correction in current application interface functional entrance, Jin Erxiang Server sends error correction report request.Wherein, as in Figure 11, " Cheng " after selection can prompt user's quilt by modes such as discolorations Choose, with user-friendly.
As another example, and user's long-press text message " 11 long holidays on National Day already close to coda, the friend to travel outdoors Friend starts to return the trip of Cheng successively." in " Cheng " carry out selection operation, so as to the participle database pair stored according to mobile terminal Text message " 11 long holidays on National Day, the friend to travel outdoors started to return the trip of Cheng successively already close to coda " be normalized and Word segmentation processing, and to the participle fragment interface of user's display text message, specific as shown in figure 12, user can be by segmenting piece A participle fragment " returning Cheng " original to one corresponding with report information " Cheng " carries out selection operation and then to server in segment limit face Send error correction report request.Wherein, as in Figure 12, the original participle fragment after selection can prompt user by modes such as discolorations It is selected, with user-friendly.
Specifically, user's pair original participle fragment of one or more corresponding with report information in participle fragment interface After selection operation, a variety of forms can be used to send error correction report request to server in the current interface of mobile terminal, than " transmission speech ", " news is unreal ", " personal attack " and " wrong word report " one or more option is such as provided to user Selection；There is provided the modes such as input frame for another example is supplied to user to be judged to input corresponding information according to itself.Can be according to reality Border carries out selection setting using needs.
Step 303, obtain the target for being used to replace original participle fragment that server is sent and segment fragment, and be shown to use Family.
Specifically, server can provide the target participle for replacing original participle fragment after corresponding processing is carried out Piece, mobile terminal segment the original participle fragment of fragment replacement according to the target of offer and are shown to user.
In summary, the user feedback error correction method for mobile terminal of the embodiment of the present invention, by obtaining user couple The selection operation of report information in text message, text message is normalized according to the participle database that mobile terminal stores And word segmentation processing, the participle fragment interface of text message is shown to user, obtain user in participle fragment interface pair with report The selection operation of one or more original participle fragments corresponding to information, and obtain user and error correction in current application interface is reported The trigger action of functional entrance, and then error correction report request is sent to server, wherein, request includes：Text message, Yi Jiyu One or more original participle fragments corresponding to report information, so as to obtain server transmission be used for replace original participle fragment Target participle fragment, and be shown to user.Thus, user carries out correction process by way of mobile terminal feedback, reduces The accuracy of error correction is improved while running cost, meets user's request.
In order to realize above-described embodiment, the present invention also proposes a kind of user feedback error correction device for mobile terminal, schemes 13 be the structural representation of the user feedback error correction device according to an embodiment of the invention for mobile terminal.Such as Figure 13 institutes Show, the user feedback error correction device for being used for mobile terminal includes：Acquisition module 11, the first determining module 12, first processing mould Block 13, extraction computing module 14 and Second processing module 15.
Wherein, request is reported in acquisition module 11, the error correction for obtaining user's transmission, wherein, request includes：Report information Affiliated text message, and positional information of the report information in text message.
First determining module 12, for being segmented after text message normalized, according to positional information determine with One or more original participle fragments corresponding to report information.
First processing module 13, for extracting the fisrt feature information of one or more original participle fragments, using default Model carries out calculating the confidence level for obtaining original participle fragment to fisrt feature information, if judging to know that confidence level is less than default threshold Value, then one or more candidates corresponding with original participle fragment are obtained according to the phrase substitution table pre-established and segment fragment.
Computing module 14 is extracted, for segmenting snippet extraction second feature letter according to original participle fragment and corresponding candidate Breath, second feature information is carried out using preset model to calculate the score value for obtaining candidate and segmenting fragment.
Second processing module 15, the score value for segmenting fragment to candidate for the default decoding algorithm of application are carried out at decoding Reason, if judging to know that decoding participle fragment corresponding with decoded result meets default intervention condition, decoding is segmented into fragment As original participle fragment corresponding to target participle fragment replacement.
Wherein, in one embodiment of the invention, the fisrt feature information of original participle fragment, including：Original participle The frequency of occurrence of fragment and context in language material；And/or original participle fragment and context changing in application scenarios are searched for The dynamic frequency；And/or the semantic similarity of original participle fragment and context.
Wherein, in one embodiment of the invention, obtained and original participle according to the phrase substitution table pre-established Before one or more candidates segment fragment corresponding to fragment, in addition to：User is gathered in a search engine to the master of search term Dynamic modification information；The search term of collection user's input and the map information wantonly searched between title；Gather the search term of user's input The error correction information provided with search engine；Phrase substitution table is established according to active modification information, map information and error correction information.
Wherein, in one embodiment of the invention, after being segmented after to text message normalized, also wrap Include：Phonetic notation is carried out to text message, and the initial and the final is mapped to corresponding participle piece fragment position；According to original participle fragment Before snippet extraction second feature information being segmented with corresponding candidate, in addition to：Obtained according to the phonetic substitution table pre-established One or more candidates corresponding with original participle fragment segment fragment.
Wherein, in one embodiment of the invention, snippet extraction is segmented according to original participle fragment and corresponding candidate Second feature information, including：Candidate segments the qualitative character of fragment, the qualitative character of original participle fragment, original participle fragment One or more combinations of features in the assemblage characteristic of fragment, the historical behavior feature of active user are segmented with candidate.
It should be noted that the explanation of the foregoing user feedback error correction method embodiment to for mobile terminal is also fitted For the user feedback error correction device for mobile terminal of the embodiment, here is omitted.
In summary, the user feedback error correction device for mobile terminal of the embodiment of the present invention, is led to by obtaining user The error correction report request of mobile terminal transmission is crossed, wherein, request includes：Text message belonging to report information, and with report One or more original participle fragments corresponding to information, wherein, mobile terminal is according to the participle database being locally stored to text Information is normalized and word segmentation processing, it is determined that the original participle fragment of one or more corresponding with report information, is then extracted The fisrt feature information of one or more original participle fragments, and calculating acquisition is carried out to fisrt feature information using preset model The confidence level of original participle fragment, is replaced when judging to know that the confidence level is less than predetermined threshold value according to the phrase pre-established Table obtains one or more candidates corresponding with original participle fragment and segments fragments, so as to according to the original fragment and corresponding of segmenting Candidate segments snippet extraction second feature information, and application preset model to second feature information calculate and obtains candidate point The score value of word fragment, the score value for finally segmenting fragment to candidate using default decoding algorithm carry out decoding process, are judging to obtain Know that decoding participle fragment corresponding with decoded result meets to segment using decoding participle fragment as target during default intervention condition Original participle fragment corresponding to fragment replacement.Thus, user carries out correction process, Yi Ji by way of mobile terminal feedback Mobile terminal locally determines original participle fragment corresponding with report information, unified standard, is improved while reducing running cost The accuracy of error correction, meets user's request.
Figure 14 is that the structure of the user feedback error correction device in accordance with another embodiment of the present invention for mobile terminal is shown It is intended to.As shown in figure 14, on the basis of Figure 13, in addition to：The determining module 17 of replacement module 16 and second.
Wherein, replacement module 16, if for judging to know that decoding participle fragment corresponding with decoded result is unsatisfactory for presetting Intervention condition, then pass through Manual definition's target segment fragment replace corresponding to original participle fragment.
Second determining module 17, if for judging to know that decoding participle fragment corresponding with decoded result meets default replace Change blacklist, it is determined that be invalid report, without correction process.
Thus, further meet the individual demand of user, further increase the accuracy and efficiency of error correction.
Figure 15 is shown according to the structure of the user feedback error correction device for mobile terminal of another embodiment of the invention It is intended to.As shown in figure 15, the user feedback error correction device for being used for mobile terminal includes：Obtain display module 21, obtain transmission Module 22 and display module 23
Wherein, display module 21 is obtained, for obtaining selection operation of the user to report information in text message, according to shifting Text message is normalized the participle database of dynamic terminal storage and word segmentation processing, and the text message is shown to user Segment fragment interface.
Sending module 22 is obtained, for obtaining user pair one corresponding with report information or more in participle fragment interface The selection operation of individual original participle fragment, and obtain user and report that error correction in current application interface the triggering of functional entrance is grasped Make, and then error correction report request is sent to server, wherein, request includes：Text message, and corresponding with report information one Individual or multiple original participle fragments.
Display module 23, the target for being used to replace original participle fragment for obtaining server transmission segment fragment, and It is shown to user.
In summary, the user feedback error correction device for mobile terminal of the embodiment of the present invention, by obtaining user couple The selection operation of report information in text message, text message is normalized according to the participle database that mobile terminal stores And word segmentation processing, the participle fragment interface of text message is shown to user, obtain user in participle fragment interface pair with report The selection operation of one or more original participle fragments corresponding to information, and obtain user and error correction in current application interface is reported The trigger action of functional entrance, and then error correction report request is sent to server, wherein, request includes：Text message, Yi Jiyu One or more original participle fragments corresponding to report information, so as to obtain server transmission be used for replace original participle fragment Target participle fragment, and be shown to user.Thus, user carries out correction process by way of mobile terminal feedback, reduces The accuracy of error correction is improved while running cost, meets user's request.
The present invention proposes a kind of computer equipment, and Figure 16 is the structure of computer equipment according to an embodiment of the invention Schematic diagram.As shown in figure 16, memory 31, processor 32 and it is stored in the meter that can be run on memory 31 and on processor 32 Calculation machine program.
Processor 32 realizes that the user feedback for mobile terminal provided in above-described embodiment is entangled when performing described program Wrong method.
Further, computer equipment also includes：
Communication interface 33, for the communication between memory 31 and processor 32.
Memory 31, for depositing the computer program that can be run on processor 32.
Memory 31 may include high-speed RAM memory, it is also possible to also including nonvolatile memory (non-volatile Memory), a for example, at least magnetic disk storage.
Processor 32, the user feedback for mobile terminal described in above-described embodiment is realized during for performing described program Error correction method.
If memory 31, processor 32 and the independent realization of communication interface 33, communication interface 33, memory 31 and processing Device 32 can be connected with each other by bus and complete mutual communication.The bus can be industry standard architecture (Industry Standard Architecture, referred to as ISA) bus, external equipment interconnection (Peripheral Component, referred to as PCI) bus or extended industry-standard architecture (Extended Industry Standard Architecture, referred to as EISA) bus etc..The bus can be divided into address bus, data/address bus, controlling bus etc.. For ease of representing, only represented in Figure 16 with a thick line, it is not intended that an only bus or a type of bus.
Optionally, in specific implementation, if memory 31, processor 32 and communication interface 33, are integrated in chip piece Upper realization, then memory 31, processor 32 and communication interface 33 can complete mutual communication by internal interface.
Processor 32 is probably a central processing unit (Central Processing Unit, referred to as CPU), or Specific integrated circuit (Application Specific Integrated Circuit, referred to as ASIC), or by with It is set to the one or more integrated circuits for implementing the embodiment of the present invention.
In order to realize above-described embodiment, the present invention also proposes a kind of non-transitorycomputer readable storage medium, when described When instruction in storage medium is performed by processor, enabling perform the use for mobile terminal described in above-described embodiment Family feedback error correction method.
In order to realize above-described embodiment, the present invention also proposes a kind of computer program product, when the computer program produces When instruction processing unit in product performs, the user feedback error correction method for mobile terminal described in above-described embodiment is performed.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show The description of example " or " some examples " etc. means specific features, structure, material or the spy for combining the embodiment or example description Point is contained at least one embodiment or example of the present invention.In this manual, to the schematic representation of above-mentioned term not Identical embodiment or example must be directed to.Moreover, specific features, structure, material or the feature of description can be with office Combined in an appropriate manner in one or more embodiments or example.In addition, in the case of not conflicting, the skill of this area Art personnel can be tied the different embodiments or example and the feature of different embodiments or example described in this specification Close and combine.
In addition, term " first ", " second " are only used for describing purpose, and it is not intended that instruction or hint relative importance Or the implicit quantity for indicating indicated technical characteristic.Thus, define " first ", the feature of " second " can be expressed or Implicitly include at least one this feature.In the description of the invention, " multiple " are meant that at least two, such as two, three It is individual etc., unless otherwise specifically defined.
Any process or method described otherwise above description in flow chart or herein is construed as, and represents to include Module, fragment or the portion of the code of the executable instruction of one or more the step of being used to realize custom logic function or process Point, and the scope of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discuss suitable Sequence, including according to involved function by it is basic simultaneously in the way of or in the opposite order, carry out perform function, this should be of the invention Embodiment person of ordinary skill in the field understood.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for Instruction execution system, device or equipment (such as computer based system including the system of processor or other can be held from instruction The system of row system, device or equipment instruction fetch and execute instruction) use, or combine these instruction execution systems, device or set It is standby and use.For the purpose of this specification, " computer-readable medium " can any can be included, store, communicate, propagate or pass Defeated program is for instruction execution system, device or equipment or the dress used with reference to these instruction execution systems, device or equipment Put.The more specifically example (non-exhaustive list) of computer-readable medium includes following：Electricity with one or more wiring Connecting portion (electronic installation), portable computer diskette box (magnetic device), random access memory (RAM), read-only storage (ROM), erasable edit read-only storage (EPROM or flash memory), fiber device, and portable optic disk is read-only deposits Reservoir (CDROM).In addition, computer-readable medium, which can even is that, to print the paper of described program thereon or other are suitable Medium, because can then enter edlin, interpretation or if necessary with it for example by carrying out optical scanner to paper or other media His suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each several part of the present invention can be realized with hardware, software, firmware or combinations thereof.Above-mentioned In embodiment, software that multiple steps or method can be performed in memory and by suitable instruction execution system with storage Or firmware is realized.Such as, if realized with hardware with another embodiment, following skill well known in the art can be used Any one of art or their combination are realized：With the logic gates for realizing logic function to data-signal from Logic circuit is dissipated, the application specific integrated circuit with suitable combinational logic gate circuit, programmable gate array (PGA), scene can compile Journey gate array (FPGA) etc..
Those skilled in the art are appreciated that to realize all or part of step that above-described embodiment method carries Suddenly it is that by program the hardware of correlation can be instructed to complete, described program can be stored in a kind of computer-readable storage medium In matter, the program upon execution, including one or a combination set of the step of embodiment of the method.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing module, can also That unit is individually physically present, can also two or more units be integrated in a module.Above-mentioned integrated mould Block can both be realized in the form of hardware, can also be realized in the form of software function module.The integrated module is such as Fruit is realized in the form of software function module and as independent production marketing or in use, can also be stored in a computer In read/write memory medium.
Storage medium mentioned above can be read-only storage, disk or CD etc..Although have been shown and retouch above Embodiments of the invention are stated, it is to be understood that above-described embodiment is exemplary, it is impossible to be interpreted as the limit to the present invention System, one of ordinary skill in the art can be changed to above-described embodiment, change, replace and become within the scope of the invention Type.

Claims (12)

1. a kind of user feedback error correction method for mobile terminal, it is characterised in that comprise the following steps：
Obtain the error correction that user is sent by mobile terminal and report request, wherein, the request includes：Text belonging to report information This information, and the original participle fragment of one or more corresponding with the report information, wherein, the mobile terminal is according to this The text message is normalized the participle database of ground storage and word segmentation processing, it is determined that corresponding with the report information One or more original participle fragments；
The fisrt feature information of one or more of original participle fragments is extracted, the fisrt feature is believed using preset model Breath carries out calculating the confidence level for obtaining the original participle fragment, if judging to know that the confidence level is less than predetermined threshold value, root One or more candidates' participle fragments corresponding with the original participle fragment are obtained according to the phrase substitution table that pre-establishes；
Snippet extraction second feature information is segmented according to the original participle fragment and corresponding candidate, using preset model to institute Second feature information is stated to carry out calculating the score value for obtaining candidate's participle fragment；
The score value for segmenting fragment to the candidate using default decoding algorithm carries out decoding process, is tied if judging to know with decoding Decoding participle fragment meets default intervention condition corresponding to fruit, then the decoding is segmented into fragment replaces as target participle fragment Original participle fragment corresponding to changing.
2. the method as described in claim 1, it is characterised in that the fisrt feature information of the original participle fragment, including：
The frequency of occurrence of original the participle fragment and context in language material；And/or
The change frequency of original the participle fragment and context in application scenarios are searched for；And/or
The semantic similarity of the original participle fragment and context.
3. the method as described in claim 1, it is characterised in that the phrase substitution table acquisition pre-established in the basis and institute Before one or more candidates corresponding to stating original participle fragment segment fragment, in addition to：
Gather user's active modification information to search term in a search engine；
The search term of collection user's input and the map information wantonly searched between title；
Gather the error correction information that the search term of user's input provides with search engine；
The phrase substitution table is established according to the active modification information, the map information and the error correction information.
4. the method as described in claim 1, it is characterised in that the mobile terminal carries out phonetic notation to the text message, and The initial and the final is mapped to corresponding participle piece fragment position；
It is described according to it is described it is original participle fragment and corresponding candidate segment snippet extraction second feature information before, also wrap Include：
One or more candidates' participle pieces corresponding with the original participle fragment are obtained according to the phonetic substitution table that pre-establishes Section.
5. the method as described in claim 1, it is characterised in that described according to the original participle fragment and corresponding candidate point Word snippet extraction second feature information, including：
The candidate segment the qualitative character of fragment, the qualitative character of the original participle fragment, the original participle fragment and One or more combinations of features in the assemblage characteristic of candidate's participle fragment, the historical behavior feature of active user.
6. the method as described in claim 1, it is characterised in that segmented in the default decoding algorithm of application to the candidate After the score value of fragment carries out decoding process, in addition to：
If judgement knows that decoding participle fragment corresponding with decoded result is unsatisfactory for default intervention condition, pass through Manual definition Original participle fragment corresponding to target participle fragment replacement；
If judgement knows that decoding participle fragment corresponding with decoded result meets default replacement blacklist, it is determined that is invalid act Report, without correction process.
7. a kind of user feedback error correction method for mobile terminal, it is characterised in that comprise the following steps：
Obtain selection operation of the user to report information in text message, the participle database pair stored according to the mobile terminal The text message is normalized and word segmentation processing, and the participle fragment interface of the text message is shown to the user；
Obtain user's pair original participle fragment of one or more corresponding with the report information in the participle fragment interface Selection operation, and obtain the trigger action that the user reports error correction in current application interface functional entrance, and then to clothes Business device sends error correction report request, wherein, the request includes：The text message, and it is corresponding with the report information One or more original participle fragments；
Obtain the target for being used to replace the original participle fragment that the server is sent and segment fragment, and be shown to the use Family.
A kind of 8. user feedback error correction device for mobile terminal, it is characterised in that including：
Acquisition module, obtain the error correction that user is sent by mobile terminal and report request, wherein, the request includes：Offence reporting letter Text message belonging to breath, and the original participle fragment of one or more corresponding with the report information, wherein, the movement The text message is normalized according to the participle database being locally stored for terminal and word segmentation processing, it is determined that with the report One or more original participle fragments corresponding to information；
First processing module, for extracting the fisrt feature information of one or more of original participle fragments, using default mould Type carries out calculating the confidence level for obtaining the original participle fragment to the fisrt feature information, if judging to know the confidence level Less than predetermined threshold value, then obtained according to the phrase substitution table pre-established corresponding one or more with the original participle fragment Candidate segments fragment；
Computing module is extracted, for segmenting snippet extraction second feature letter according to the original participle fragment and corresponding candidate Breath, the second feature information is carried out using preset model to calculate the score value for obtaining the candidate and segmenting fragment；
Second processing module, the score value for segmenting fragment to the candidate for the default decoding algorithm of application carry out decoding process, If judgement knows that decoding participle fragment corresponding with decoded result meets default intervention condition, the decoding is segmented into fragment As original participle fragment corresponding to target participle fragment replacement.
A kind of 9. user feedback error correction device for mobile terminal, it is characterised in that including：
Display module is obtained, for obtaining selection operation of the user to report information in text message, according to the mobile terminal The text message is normalized the participle database of storage and word segmentation processing, and the text message is shown to the user Participle fragment interface；
Obtain sending module, for obtain user in the participle fragment interface pair one corresponding with the report information or The selection operation of multiple original participle fragments, and obtain the user and error correction report functional entrance in current application interface is touched Hair operation, and then error correction report request is sent to server, wherein, the request includes：The text message, and with it is described One or more original participle fragments corresponding to report information；
Display module, fragment is segmented for obtaining the target for being used to replace the original participle fragment that the server is sent, And it is shown to the user.
10. a kind of computer equipment, it is characterised in that on a memory and can handled including memory, processor and storage The computer program run on device, during the computing device described program, realize the use as described in any in claim 1-7 In the user feedback error correction method of mobile terminal.
11. a kind of non-transitorycomputer readable storage medium, is stored thereon with computer program, it is characterised in that the program The user feedback error correction method for mobile terminal as described in any in claim 1-7 is realized when being executed by processor.
12. a kind of computer program product, it is characterised in that when the instruction in the computer program product is by computing device When, perform the user feedback error correction method for mobile terminal as described in any in claim 1-7.
CN201711173999.XA 2017-11-22 2017-11-22 User feedback error correction method, device and its equipment for mobile terminal Pending CN107832447A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711173999.XA CN107832447A (en) 2017-11-22 2017-11-22 User feedback error correction method, device and its equipment for mobile terminal

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711173999.XA CN107832447A (en) 2017-11-22 2017-11-22 User feedback error correction method, device and its equipment for mobile terminal

Publications (1)

Publication Number Publication Date
CN107832447A true CN107832447A (en) 2018-03-23

Family

ID=61652364

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711173999.XA Pending CN107832447A (en) 2017-11-22 2017-11-22 User feedback error correction method, device and its equipment for mobile terminal

Country Status (1)

Country Link
CN (1) CN107832447A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110751234A (en) * 2019-10-09 2020-02-04 科大讯飞股份有限公司 OCR recognition error correction method, device and equipment
CN111259897A (en) * 2018-12-03 2020-06-09 杭州翼心信息科技有限公司 Knowledge-aware text recognition method and system
CN112733529A (en) * 2019-10-28 2021-04-30 阿里巴巴集团控股有限公司 Text error correction method and device
CN113033186A (en) * 2021-05-31 2021-06-25 江苏联著实业股份有限公司 Error correction early warning method and system based on event analysis
CN115037988A (en) * 2021-03-05 2022-09-09 北京字节跳动网络技术有限公司 Page display method, device and equipment
CN111259897B (en) * 2018-12-03 2024-05-31 杭州翼心信息科技有限公司 Knowledge-aware text recognition method and system

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103942223A (en) * 2013-01-23 2014-07-23 北京百度网讯科技有限公司 Method and system for conducting online error correction on language model
CN104915264A (en) * 2015-05-29 2015-09-16 北京搜狗科技发展有限公司 Input error-correction method and device
CN105933096A (en) * 2016-06-30 2016-09-07 广东小天才科技有限公司 Information processing method and device based on error correction feedback
US9602133B1 (en) * 2015-01-27 2017-03-21 Microsemi Storage Solutions (U.S.), Inc. System and method for boost floor mitigation
CN106534548A (en) * 2016-11-17 2017-03-22 科大讯飞股份有限公司 Voice error correction method and device
CN106528845A (en) * 2016-11-22 2017-03-22 北京百度网讯科技有限公司 Artificial intelligence-based searching error correction method and apparatus

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103942223A (en) * 2013-01-23 2014-07-23 北京百度网讯科技有限公司 Method and system for conducting online error correction on language model
US9602133B1 (en) * 2015-01-27 2017-03-21 Microsemi Storage Solutions (U.S.), Inc. System and method for boost floor mitigation
CN104915264A (en) * 2015-05-29 2015-09-16 北京搜狗科技发展有限公司 Input error-correction method and device
CN105933096A (en) * 2016-06-30 2016-09-07 广东小天才科技有限公司 Information processing method and device based on error correction feedback
CN106534548A (en) * 2016-11-17 2017-03-22 科大讯飞股份有限公司 Voice error correction method and device
CN106528845A (en) * 2016-11-22 2017-03-22 北京百度网讯科技有限公司 Artificial intelligence-based searching error correction method and apparatus

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111259897A (en) * 2018-12-03 2020-06-09 杭州翼心信息科技有限公司 Knowledge-aware text recognition method and system
CN111259897B (en) * 2018-12-03 2024-05-31 杭州翼心信息科技有限公司 Knowledge-aware text recognition method and system
CN110751234A (en) * 2019-10-09 2020-02-04 科大讯飞股份有限公司 OCR recognition error correction method, device and equipment
CN110751234B (en) * 2019-10-09 2024-04-16 科大讯飞股份有限公司 OCR (optical character recognition) error correction method, device and equipment
CN112733529A (en) * 2019-10-28 2021-04-30 阿里巴巴集团控股有限公司 Text error correction method and device
CN112733529B (en) * 2019-10-28 2023-09-29 阿里巴巴集团控股有限公司 Text error correction method and device
CN115037988A (en) * 2021-03-05 2022-09-09 北京字节跳动网络技术有限公司 Page display method, device and equipment
CN115037988B (en) * 2021-03-05 2024-05-14 北京字节跳动网络技术有限公司 Page display method, device and equipment
CN113033186A (en) * 2021-05-31 2021-06-25 江苏联著实业股份有限公司 Error correction early warning method and system based on event analysis

Similar Documents

Publication Publication Date Title
US11790006B2 (en) Natural language question answering systems
US20220382752A1 (en) Mapping Natural Language To Queries Using A Query Grammar
CN109670163B (en) Information identification method, information recommendation method, template construction method and computing device
CN110019732B (en) Intelligent question answering method and related device
CN107977357A (en) Error correction method, device and its equipment based on user feedback
WO2020108063A1 (en) Feature word determining method, apparatus, and server
CN107832447A (en) User feedback error correction method, device and its equipment for mobile terminal
CN108304372A (en) Entity extraction method and apparatus, computer equipment and storage medium
CN108228571B (en) Method and device for generating couplet, storage medium and terminal equipment
CN111460170B (en) Word recognition method, device, terminal equipment and storage medium
CN112347767B (en) Text processing method, device and equipment
CN112115232A (en) Data error correction method and device and server
CN103488752A (en) POI (point of interest) searching method
CN108763202A (en) Method, apparatus, equipment and the readable storage medium storing program for executing of the sensitive text of identification
CN110674301A (en) Emotional tendency prediction method, device and system and storage medium
CN110209781A (en) A kind of text handling method, device and relevant device
CN103150409A (en) Method and system for recommending user search word
CN114706894A (en) Information processing method, apparatus, device, storage medium, and program product
CN113779987A (en) Event co-reference disambiguation method and system based on self-attention enhanced semantics
CN113822059A (en) Chinese sensitive text recognition method and device, storage medium and equipment
WO2023103914A1 (en) Text sentiment analysis method and device, and computer-readable storage medium
CN110874408B (en) Model training method, text recognition device and computing equipment
WO2021098491A1 (en) Knowledge graph generating method, apparatus, and terminal, and storage medium
CN108304367A (en) Segmenting method and device
CN114595696A (en) Entity disambiguation method, entity disambiguation apparatus, storage medium, and electronic device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20180323