CN112001381B - Intelligent pre-filling bill auditing method and device - Google Patents

Intelligent pre-filling bill auditing method and device Download PDF

Info

Publication number
CN112001381B
CN112001381B CN202010963046.9A CN202010963046A CN112001381B CN 112001381 B CN112001381 B CN 112001381B CN 202010963046 A CN202010963046 A CN 202010963046A CN 112001381 B CN112001381 B CN 112001381B
Authority
CN
China
Prior art keywords
information
client
text
clients
corrected
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
Application number
CN202010963046.9A
Other languages
Chinese (zh)
Other versions
CN112001381A (en
Inventor
黄文强
季蕴青
胡路苹
胡玮
黄雅楠
胡传杰
浮晨琪
李蚌蚌
徐晨敏
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.)
Bank of China Ltd
Original Assignee
Bank of China 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 Bank of China Ltd filed Critical Bank of China Ltd
Priority to CN202010963046.9A priority Critical patent/CN112001381B/en
Publication of CN112001381A publication Critical patent/CN112001381A/en
Application granted granted Critical
Publication of CN112001381B publication Critical patent/CN112001381B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/02—Banking, e.g. interest calculation or account maintenance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/24—Aligning, centring, orientation detection or correction of the image
    • G06V10/243—Aligning, centring, orientation detection or correction of the image by compensating for image skew or non-uniform image deformations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition

Abstract

The invention provides an intelligent pre-filling bill auditing method and device, wherein the method comprises the following steps: acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information; carrying out structural correction on the client information of the target bill; and carrying out text recognition on the corrected client information, and auditing a text recognition result based on column information of the target bill. Based on the invention, the examination of the bill can be completed in the queuing time of the client, thereby reminding the client of modifying in time, preventing errors from occurring in service handling and improving the efficiency of service handling.

Description

Intelligent pre-filling bill auditing method and device
Technical Field
The invention relates to the technical field of banking business processing, in particular to an intelligent pre-filling bill auditing method and device.
Background
The clients need to fill in paper receipts before the banks transact business, and when the clients transact business, the staff needs to audit the contents in the application form, and once errors occur, the clients need to refill, which seriously affects the speed of business transacting.
Disclosure of Invention
In view of the above, in order to solve the above problems, the present invention provides an intelligent pre-filling bill auditing method and apparatus, and the technical scheme is as follows:
an intelligent pre-form auditing method, the method comprising:
acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information;
carrying out structural correction on the client information of the target bill;
and carrying out text recognition on the corrected client information, and auditing a text recognition result based on the column information of the target bill.
Preferably, the performing structural correction on the client information of the target document includes:
acquiring the identity information of the clients and calling the identity information of a plurality of predetermined clients to be corrected;
if the identity information of the client belongs to one of the identity information of the plurality of corrected clients, a structural correction model corresponding to the identity information of the client is called, wherein the structural correction model is obtained by training the structural correction model by taking client information filled in by the client in a history receipt as a training sample, taking a mark of a prediction result of the training sample, which is close to the training sample, of the structural correction model to be trained as a target, and taking the mark of the training sample as a standard structure;
and carrying out structural correction on the client information of the target bill through the structural correction model.
Preferably, the determining process of the plurality of clients to be corrected includes:
determining a plurality of candidate clients to be screened;
calculating the filling accuracy of each candidate client based on the client information filled in the history document by each candidate client and the corresponding history text recognition result;
and determining the candidate clients with the filling accuracy smaller than a preset accuracy threshold value from the plurality of candidate clients as corrected clients.
Preferably, the text recognition of the corrected client information includes:
recognizing corrected customer information based on an optical character recognition technology to obtain a first recognition result;
acquiring a text information panorama of the client, wherein the text information panorama comprises a second identification result of client information filled in a history bill of the client;
and acquiring a weighted proportion of the first identification result and the second identification result, and processing the first identification result and the second identification result based on the weighted proportion, wherein the weighted proportion is calculated based on a genetic algorithm.
Preferably, the auditing the text recognition result based on the column information of the target document includes:
and acquiring a logic rule corresponding to the column information of the target bill through an expert system, and traversing a text recognition result based on the logic rule.
Preferably, the auditing the text recognition result based on the column information of the target document further includes:
and identifying the rationality of the text identification result by using a naive Bayesian model.
An intelligent pre-form auditing apparatus, the apparatus comprising:
the information acquisition module is used for acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information;
the correcting module is used for carrying out structural correction on the client information of the target bill;
and the identification and verification module is used for carrying out text identification on the corrected client information and verifying a text identification result based on the column information of the target bill.
Preferably, the correction module is specifically configured to:
acquiring the identity information of the clients and calling the identity information of a plurality of predetermined clients to be corrected; if the identity information of the client belongs to one of the identity information of the plurality of corrected clients, a structural correction model corresponding to the identity information of the client is called, wherein the structural correction model is obtained by training the structural correction model by taking client information filled in by the client in a history receipt as a training sample, taking a mark of a prediction result of the training sample, which is close to the training sample, of the structural correction model to be trained as a target, and taking the mark of the training sample as a standard structure; and carrying out structural correction on the client information of the target bill through the structural correction model.
Preferably, the identification and audit module is configured to perform text identification on corrected client information, and is specifically configured to:
recognizing corrected customer information based on an optical character recognition technology to obtain a first recognition result; acquiring a text information panorama of the client, wherein the text information panorama comprises a second identification result of client information filled in a history bill of the client; and acquiring a weighted proportion of the first identification result and the second identification result, and processing the first identification result and the second identification result based on the weighted proportion, wherein the weighted proportion is calculated based on a genetic algorithm.
Preferably, the recognition and verification module is configured to verify a text recognition result based on field information of the target document, and is specifically configured to:
and acquiring a logic rule corresponding to the column information of the target bill through an expert system, and traversing a text recognition result based on the logic rule.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides an intelligent pre-filling bill auditing method and device, which can identify client information filled in a bill after the client fills in the bill, ensure the text identification effect through structural correction, and finally audit the text identification result of the client information based on the column information of the bill. Based on the invention, the examination of the bill can be completed in the queuing time of the client, thereby reminding the client of modifying in time, preventing errors from occurring in service handling and improving the efficiency of service handling.
In addition, based on the invention, the user information in the bill is already identified by the text, and the repeated input of staff is not needed when the business is handled, thereby further improving the business handling efficiency.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for auditing an intelligent pre-filling order according to an embodiment of the present invention;
FIG. 2 is a partial method flowchart of an intelligent pre-filling order auditing method provided by an embodiment of the invention;
fig. 3 is a schematic structural diagram of an intelligent pre-filling bill auditing device according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
The embodiment of the invention provides an intelligent pre-filling bill auditing method, a flow chart of the method is shown in figure 1, and the method comprises the following steps:
s10, acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information.
In the embodiment of the invention, in the queuing process of the business transacted by the client, the client can fill out the relevant paper bill according to the business transacted by the client, and after the client fills up the paper bill, the client verifies the identity by means of swiping an identity card or a bank card and starts the pre-filling bill auditing mechanism of the invention.
The paper bill contains column information such as 'name', 'gender', 'identity card number', and the client fills relevant client information based on the column information, such as 'Zhang San' in a filling area corresponding to the 'name'.
S20, carrying out structural correction on the client information of the target bill.
In the embodiment of the invention, the client information of the target bill can be structurally corrected based on the existing algorithm. In the specific implementation process, in order to improve the pertinence of correction, a proprietary structure interleaving mechanism can be added for clients. Step S20 may be performed as follows, and the method flowchart is shown in fig. 2:
s201, acquiring identity information of clients and calling the identity information of a plurality of predetermined clients to be corrected.
In the embodiment of the invention, the identity information of the client can be an identity card number, a client number and other identifiers capable of uniquely marking the client. The clients to be corrected can be specified by the manager, the existing business information can be analyzed, and the clients with difficult recognition, namely the clients to be corrected, are determined according to the historical text recognition results and the modification conditions of the staff.
Specifically, the determining process of the plurality of clients to be corrected includes the following steps:
determining a plurality of candidate clients to be screened; calculating the filling accuracy of each candidate client based on the client information filled in the history document by each candidate client and the corresponding history text recognition result; and determining the candidate clients with the filling accuracy smaller than a preset accuracy threshold value from the plurality of candidate clients as corrected clients.
In the embodiment of the invention, a manager can determine candidate clients in clients handling the service or handling the specified service in a specified period, and for each candidate client, client information filled in a history bill and related history text recognition results are called when the client handles the history service.
And identifying correct client information in the historical text identification result, and taking the text quantity ratio of the client information filled in the historical document as the filling accuracy of the subsequent client.
And for the candidate clients with filling accuracy greater than or equal to the preset accuracy threshold, the candidate clients are not used as correction clients, so that the clients can not be subjected to structural correction during subsequent business handling.
S202, if the identity information of the client belongs to one of the identity information of the plurality of corrected clients, a structural correction model corresponding to the identity information of the client is called, the structural correction model is obtained by taking client information filled in a history document by the client as a training sample, a prediction result of the training sample by the structural correction model to be trained is close to a label of the training sample as a target, and the label of the training sample is a standard structure.
In the embodiment of the invention, a corresponding structure correction model is established for each client to be corrected, so that the correction pertinence can be ensured, and the accuracy of the subsequent text recognition can be improved.
When the structure correction model is built, the neural network model can be used as a to-be-trained structure correction model based, a three-layer neural network is built, client information filled in a history receipt of a to-be-corrected client is used as a training sample, a standard structure of the client information is marked for the client information, the marked training sample is input into the to-be-trained structure correction model, and parameter weights of all layers of the neural network are adjusted.
Of course, a test sample may be prepared, and the trained model may be tested to verify the accuracy of its prediction to obtain an effective model.
S203, carrying out structural correction on the client information of the target bill through a structural correction model.
In the embodiment of the invention, the client information of the target bill is input into the structural correction model, and the structural correction model can output the correction result of the client information.
It should be noted that, the text structure is corrected, and no adjustment is made to the content thereof.
S30, text recognition is carried out on the corrected client information, and the text recognition result is checked based on the column information of the target bill.
In the embodiment of the invention, the corrected client information can be subjected to text recognition by adopting an optical character recognition technology, and the result of text recognition is more accurate because the client information is corrected.
In addition, when the text recognition result is checked, the standard information corresponding to the column information of the target document can be searched through the identity information of the client, the text recognition result of the standard information and the client information corresponding to the column information is further compared, if the standard information and the text recognition result are the same, the client information filled in by the column information is represented to be correct by the client, otherwise, the problem exists.
Once the problem of the client information corresponding to a certain column information is found, the column information can be further marked and displayed to the client to prompt the client to modify. If the customer selects the modification, the relevant customer data is updated directly.
In a specific implementation process, in order to improve the accuracy of text recognition by the optical character recognition technology, in the embodiment of the present invention, the text recognition on corrected client information may include the following steps:
recognizing corrected customer information based on an optical character recognition technology to obtain a first recognition result; acquiring a text information panorama of the client, wherein the text information panorama comprises a second identification result of client information filled in a history bill of the client; and obtaining a weighted proportion of the first identification result and the second identification result, and processing the first identification result and the second identification result based on the weighted proportion, wherein the weighted proportion is calculated based on a genetic algorithm.
In the embodiment of the present invention, for the corrected client information, the first recognition result output by the optical character recognition technology includes the recognized text and the possible probability of the text, for example, the possible probability of the a text is a, the possible probability of the B text is B, and the possible probability of the C text is C, where a+b+c=1 in the first recognition result.
In addition, for the client information filled in the history bill by the client, the second recognition result output in the text recognition process comprises the recognized text and the use probability of the text, and the recognized text in the second recognition result is accurate text. For example, in the second recognition result, D times appear for the A text, e times appear for the B text, f times appear for the C text, and g times appear for the D text, the using probability of the A text is D/(d+e+f+g), the using probability of the B text is e/(d+e+f+g), the using probability of the C text is f/(d+e+f+g), and the using probability of the D text is g/(d+e+f+g).
Thus, for the a text in the first recognition result, the possible probability is a and the use probability is d/(d+e+f+g); the possible probability of the text B is B, and the use probability is e/(d+e+f+g); the possible probability of C text is C, and the use probability is f/(d+e+f+g).
Further, based on the weighted proportion of the first recognition result and the second recognition result, the possible probability a of the A text, the possible probability B of the B text, the possible probability e/(d+e+f+g), the possible probability C of the C text, and the possible probability f/(d+e+f+g) are weighted in this order.
Taking the weighted proportion of 2:3 as an example, the weighted result of the A text is a 0.4+d/(d+e+f+g) 0.6, the weighted result of the B text is b 0.4+e/(d+e+f+g) 0.6, and the weighted result of the C text is c 0.4+f/(d+e+f+g) 0.6. And taking the text with the largest weighted result value as a final text recognition result.
The weighted proportion is calculated by a genetic algorithm, an initial value is given to the weighted proportion, then an evaluation function is established, the evaluation function is the ratio of the error between the weighted recognition result and the true value and the ratio between the weighted recognition result and the non-weighted error, whether the precision requirement can be met or not is judged by the evaluation function, if the precision requirement can not be met, a new weighted proportion is obtained according to the functions of inheritance, variation and the like of the genetic algorithm, and therefore an effective weighted proportion is obtained, and a more accurate text recognition result is obtained.
In a specific implementation process, in order to further verify logic of document filling, in the embodiment of the present invention, the "auditing the text recognition result based on the field information of the target document" may include the following steps:
and acquiring a logic rule corresponding to the column information of the target bill by using an expert system, and traversing the text recognition result based on the logic rule.
In the embodiment of the invention, whether the filling logic of the column is problematic is judged by an expert system, for example, a client has a class of users and cannot open a class of users again, or a client selects an A product and cannot handle a B product, and the like. The expert system consists of a knowledge base and an inference engine, wherein an administrator in the knowledge base sets logic rules in the banking business handling process, and the logic rules can be continuously enriched and continuously optimized according to problems encountered in the subsequent business handling process; the inference engine uses a tree diagram method to continuously traverse, infers whether the client information input by the final client meets the requirement of service handling, displays the result to the client, and the client can modify the client information in advance.
Furthermore, in order to detect whether the service pre-filling information is reasonable, if not, prompt the customer to modify the customer information in time, in the embodiment of the invention, the step of checking the text recognition result based on the column information of the target document can be adopted as follows:
the rationality of the text recognition result is recognized using a naive bayes model.
In the embodiment of the invention, a naive Bayesian model establishment method is as follows:
1. let x= { A1, a 2..an } be the customer personal information case, including information of customer personal asset, use case of customer personal product, customer personality, etc.
2. There is a category set c= (Y1, Y2), Y1 indicates that the traffic is reasonable, and Y2 indicates that the traffic is unreasonable.
3. The probabilities of P (Y1|x) and P (Y2|x), i.e., the business handled under the customer's personal conditions, being both reasonable and unreasonable are calculated.
The key point is now to find P (Y1|x) and P (Y2|x), the method is as follows:
1. finding a set of x in the past client information, wherein the set is called a training sample, namely, the reasonable corresponding relation between the past client information and the business is obtained, namely, the corresponding relation between the existing x= { A1, A2..an } and C= (Y1, Y2)
2. And (5) counting to obtain the conditional probability estimation of each characteristic attribute under each category. I.e.
P (a1|y1) P (a2|y1) … P (an|y1) and P (a1|y2) P (a2|y2) … P (an|y2) are probabilities of each attribute under reasonable traffic conditions and each attribute under unreasonable conditions. If the history data shows that the client handles the large inventory is reasonable, the probability is that the client asset exceeds the reasonable proportion of one million attributes to all the large inventory.
3. There are following derivations according to the bayesian theorem:
p (Yi|x) =P (x|Yi) P (Yi)/P (x), because the denominator is constant for all classes, we only need a numerator to get the corresponding probability
P(x|Yi)*P(Yi)＝P(A1|Yi)*P(A2|Yi)*…*P(Am|Yi)*P(Yi)
4、P(Y1|x)＝P(x|Y1)P(Y1)＝P(A1|Y1)*P(A2|Y1)*…*P(Am|Y1)*P(Y1)；
P(Y2|x)＝P(x|Y2)P(Y2)＝P(A1|Y2)*P(A2|Y2)*…*P(Am|Y2)*P(Y2)。
5. When P (y1|x) reaches the threshold value m (the setting of m is set according to the requirements of the client), the service is a reasonable service, i.e. the text recognition result is reasonable.
The intelligent pre-filling bill auditing method provided by the embodiment of the invention can identify the client information filled in the bill after the client fills in the bill, ensure the text identification effect through structural correction, and finally audit the text identification result of the client information based on the column information of the bill. Based on the invention, the examination of the bill can be completed in the queuing time of the client, thereby reminding the client of modifying in time, preventing errors from occurring in service handling and improving the efficiency of service handling.
Based on the intelligent pre-filling bill auditing method provided by the embodiment, the embodiment of the invention correspondingly provides a device for executing the intelligent pre-filling bill auditing method, and the structure schematic diagram of the device is shown in fig. 3, and the device comprises:
the information acquisition module 10 is used for acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information;
the correction module 20 is used for carrying out structural correction on the client information of the target bill;
the identification and verification module 30 is configured to perform text identification on the corrected customer information, and verify the text identification result based on the field information of the target document.
Optionally, the correction module 20 is specifically configured to: acquiring identity information of clients, and calling the identity information of a plurality of predetermined clients to be corrected; if the identity information of the client belongs to one of the identity information of the plurality of corrected clients, a structural correction model corresponding to the identity information of the client is called, the structural correction model is obtained by taking the client information filled in a history document by the client as a training sample, taking the prediction result of the training sample by the structural correction model to be trained as a target, and training the structural correction model to be trained, wherein the mark of the training sample is a standard structure; and carrying out structural correction on the client information of the target bill through the structural correction model.
Optionally, the process of determining a plurality of clients to be corrected by the correction module 20 includes:
determining a plurality of candidate clients to be screened; calculating the filling accuracy of each candidate client based on the client information filled in the history document by each candidate client and the corresponding history text recognition result; and determining the candidate clients with the filling accuracy smaller than a preset accuracy threshold value from the plurality of candidate clients as corrected clients.
Optionally, the recognition auditing module 30 is configured to perform text recognition on the corrected client information, and specifically is configured to:
recognizing corrected customer information based on an optical character recognition technology to obtain a first recognition result; acquiring a text information panorama of the client, wherein the text information panorama comprises a second identification result of client information filled in a history bill of the client; and obtaining a weighted proportion of the first identification result and the second identification result, and processing the first identification result and the second identification result based on the weighted proportion, wherein the weighted proportion is calculated based on a genetic algorithm.
Optionally, the recognition and auditing module 30 is configured to audit the text recognition result based on the field information of the target document, and specifically is configured to:
and acquiring a logic rule corresponding to the column information of the target bill by using an expert system, and traversing the text recognition result based on the logic rule.
Optionally, the recognition and auditing module 30 for auditing the text recognition result based on the field information of the target document is further configured to:
the rationality of the text recognition result is recognized using a naive bayes model.
The intelligent pre-filling bill auditing device provided by the embodiment of the invention can identify the client information filled in the bill after the client fills in the bill, ensure the text identification effect through structural correction, and finally audit the text identification result of the client information based on the column information of the bill. Based on the invention, the examination of the bill can be completed in the queuing time of the client, thereby reminding the client of modifying in time, preventing errors from occurring in service handling and improving the efficiency of service handling.
The above describes in detail an intelligent pre-form auditing method and device provided by the invention, and specific examples are applied to illustrate the principle and implementation of the invention, and the above examples are only used for helping to understand the method and core idea of the invention; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in accordance with the ideas of the present invention, the present description should not be construed as limiting the present invention in view of the above.
It should be noted that, in the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described as different from other embodiments, and identical and similar parts between the embodiments are all enough to be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
It is further noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include, or is intended to include, elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. An intelligent pre-filling order auditing method, which is characterized by comprising the following steps:
acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information;
carrying out structural correction on the client information of the target bill;
text recognition is carried out on the corrected client information, and the text recognition result is checked based on the column information of the target bill;
the structure correction of the client information of the target bill comprises the following steps:
acquiring the identity information of the clients and calling the identity information of a plurality of predetermined clients to be corrected;
if the identity information of the client belongs to one of the identity information of the plurality of corrected clients, a structural correction model corresponding to the identity information of the client is called;
and carrying out structural correction on the client information of the target bill through the structural correction model.
2. The method according to claim 1, wherein the structural correction model is obtained by training a structural correction model to be trained by taking customer information filled in a history document by the customer as a training sample and taking a label of a predicted result of the training sample by the structural correction model to be trained approaching to the training sample as a target, and the label of the training sample is a standard structure.
3. The method of claim 2, wherein the determining of the plurality of clients to be corrected comprises:
determining a plurality of candidate clients to be screened;
calculating the filling accuracy of each candidate client based on the client information filled in the history document by each candidate client and the corresponding history text recognition result;
and determining the candidate clients with the filling accuracy smaller than a preset accuracy threshold value from the plurality of candidate clients as corrected clients.
4. The method of claim 1, wherein said text recognition of corrected customer information comprises:
recognizing corrected customer information based on an optical character recognition technology to obtain a first recognition result;
acquiring a text information panorama of the client, wherein the text information panorama comprises a second identification result of client information filled in a history bill of the client;
and acquiring a weighted proportion of the first identification result and the second identification result, and processing the first identification result and the second identification result based on the weighted proportion, wherein the weighted proportion is calculated based on a genetic algorithm.
5. The method of claim 1, wherein the auditing the text recognition result based on the field information of the target document comprises:
and acquiring a logic rule corresponding to the column information of the target bill through an expert system, and traversing a text recognition result based on the logic rule.
6. The method of claim 5, wherein the auditing the text recognition result based on the field information of the target document further comprises:
and identifying the rationality of the text identification result by using a naive Bayesian model.
7. An intelligent pre-form auditing apparatus, the apparatus comprising:
the information acquisition module is used for acquiring information of a target bill, wherein the information of the target bill comprises column information and client information filled by clients aiming at the column information;
the correcting module is used for carrying out structural correction on the client information of the target bill;
the identification and verification module is used for carrying out text identification on the corrected client information and verifying a text identification result based on the column information of the target bill;
wherein, the correction module is specifically configured to:
acquiring the identity information of the clients and calling the identity information of a plurality of predetermined clients to be corrected; if the identity information of the client belongs to one of the identity information of the plurality of corrected clients, a structural correction model corresponding to the identity information of the client is called; and carrying out structural correction on the client information of the target bill through the structural correction model.
8. The apparatus of claim 7, wherein the structural correction model is obtained by training a structural correction model to be trained by taking customer information filled in a history document by the customer as a training sample and taking a label of a predicted result of the training sample by the structural correction model to be trained approaching to the training sample as a target, and the label of the training sample is a standard structure.
9. The device according to claim 8, wherein the recognition auditing module is configured to perform text recognition on the corrected customer information, specifically configured to:
recognizing corrected customer information based on an optical character recognition technology to obtain a first recognition result; acquiring a text information panorama of the client, wherein the text information panorama comprises a second identification result of client information filled in a history bill of the client; and acquiring a weighted proportion of the first identification result and the second identification result, and processing the first identification result and the second identification result based on the weighted proportion, wherein the weighted proportion is calculated based on a genetic algorithm.
10. The apparatus of claim 8, wherein the recognition and auditing module configured to audit a text recognition result based on field information of the target document is specifically configured to:
and acquiring a logic rule corresponding to the column information of the target bill through an expert system, and traversing a text recognition result based on the logic rule.
CN202010963046.9A 2020-09-14 2020-09-14 Intelligent pre-filling bill auditing method and device Active CN112001381B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010963046.9A CN112001381B (en) 2020-09-14 2020-09-14 Intelligent pre-filling bill auditing method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010963046.9A CN112001381B (en) 2020-09-14 2020-09-14 Intelligent pre-filling bill auditing method and device

Publications (2)

Publication Number Publication Date
CN112001381A CN112001381A (en) 2020-11-27
CN112001381B true CN112001381B (en) 2023-09-19

Family

ID=73469888

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010963046.9A Active CN112001381B (en) 2020-09-14 2020-09-14 Intelligent pre-filling bill auditing method and device

Country Status (1)

Country Link
CN (1) CN112001381B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109214362A (en) * 2018-10-23 2019-01-15 金蝶软件（中国）有限公司 Bill processing method and relevant device
CN110291527A (en) * 2018-01-19 2019-09-27 深圳前海达闼云端智能科技有限公司 Information processing method, system, cloud processing equipment and computer program product
CN110738602A (en) * 2019-09-12 2020-01-31 北京三快在线科技有限公司 Image processing method and device, electronic equipment and readable storage medium
CN110765996A (en) * 2019-10-21 2020-02-07 北京百度网讯科技有限公司 Text information processing method and device
CN110866495A (en) * 2019-11-14 2020-03-06 杭州睿琪软件有限公司 Bill image recognition method, bill image recognition device, bill image recognition equipment, training method and storage medium
CN111612009A (en) * 2020-05-21 2020-09-01 腾讯科技（深圳）有限公司 Text recognition method, device, equipment and storage medium

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110291527A (en) * 2018-01-19 2019-09-27 深圳前海达闼云端智能科技有限公司 Information processing method, system, cloud processing equipment and computer program product
CN109214362A (en) * 2018-10-23 2019-01-15 金蝶软件（中国）有限公司 Bill processing method and relevant device
CN110738602A (en) * 2019-09-12 2020-01-31 北京三快在线科技有限公司 Image processing method and device, electronic equipment and readable storage medium
CN110765996A (en) * 2019-10-21 2020-02-07 北京百度网讯科技有限公司 Text information processing method and device
CN110866495A (en) * 2019-11-14 2020-03-06 杭州睿琪软件有限公司 Bill image recognition method, bill image recognition device, bill image recognition equipment, training method and storage medium
CN111612009A (en) * 2020-05-21 2020-09-01 腾讯科技（深圳）有限公司 Text recognition method, device, equipment and storage medium

Also Published As

Publication number Publication date
CN112001381A (en) 2020-11-27

Similar Documents

Publication Publication Date Title
US11816544B2 (en) Composite machine learning system for label prediction and training data collection
US11681924B2 (en) Training neural networks using a variational information bottleneck
CN108629593B (en) Fraud transaction identification method, system and storage medium based on deep learning
US11443170B2 (en) Semi-supervised training of neural networks
US11651247B2 (en) Method for verifying lack of bias of deep learning AI systems
CN108596630B (en) Fraud transaction identification method, system and storage medium based on deep learning
EP4094253A1 (en) System, apparatus and methods for providing an intent suggestion to a user in a text-based conversational experience with user feedback
CN108550065B (en) Comment data processing method, device and equipment
EP3933702A1 (en) Latent feature dimensionality bounds for robust machine learning on high dimensional datasets
CN103617146B (en) A kind of machine learning method and device based on hardware resource consumption
CN110162939B (en) Man-machine identification method, equipment and medium
CN111626827A (en) Method, device, equipment and medium for recommending articles based on sequence recommendation model
CN110717037B (en) Method and device for classifying users
CN112001381B (en) Intelligent pre-filling bill auditing method and device
CN114119191A (en) Wind control method, overdue prediction method, model training method and related equipment
CN112463964B (en) Text classification and model training method, device, equipment and storage medium
CN111667339B (en) Defamation malicious user detection method based on improved recurrent neural network
CN114186646A (en) Block chain abnormal transaction identification method and device, storage medium and electronic equipment
CN114462417A (en) Comment text processing method applied to big data and storage medium
WO2024098282A1 (en) Geometric problem-solving method and apparatus, and device and storage medium
US11941531B1 (en) Attention-based prototypical learning
JP2023542146A (en) Method for performing address matching and related electronic devices
CN115705412A (en) Object identification method and device, computing equipment and storage medium
CN116644233A (en) Course recommendation method, device, equipment and storage medium of online learning platform
CN113886779A (en) Method for identifying person identity, storage medium and computer program product

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
GR01 Patent grant
GR01 Patent grant