CN115410215A

CN115410215A - Loss assessment and claim settlement method, device, equipment and storage medium

Info

Publication number: CN115410215A
Application number: CN202210987666.5A
Authority: CN
Inventors: 肖南平
Original assignee: Ping An Property and Casualty Insurance Company of China Ltd
Current assignee: Ping An Property and Casualty Insurance Company of China Ltd
Priority date: 2022-08-17
Filing date: 2022-08-17
Publication date: 2022-11-29

Abstract

The application relates to artificial intelligence and provides a method, a device, equipment and a storage medium for loss assessment and claim settlement. The method includes the steps of obtaining an initial label, initial label information and a claim form image of a loss assessment object, conducting image recognition on the claim form image, determining keyword label information according to a keyword label and a text keyword generated after the image recognition, respectively calculating a general coefficient and a characteristic coefficient of a preset loss assessment item based on the initial label, the initial label information, the keyword label information, a preset general coefficient model and a preset characteristic coefficient model, respectively adjusting and correcting the general coefficient based on the characteristic coefficient and a preset loss assessment standard library to obtain a loss assessment coefficient of the preset loss assessment item, calculating claim data of the loss assessment object according to a plurality of preset loss assessment items and a plurality of loss assessment coefficients, and improving accuracy of human injury loss assessment. In addition, the application also relates to a block chain technology, and the claim data can be stored in the block chain.

Description

Loss assessment and claim settlement method, device, equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, a device, and a storage medium for loss assessment and claim settlement.

Background

In the current human injury damage assessment scheme, as a large amount of data needs to be analyzed, the process of human injury damage assessment is complex, and the accuracy of human injury damage assessment is poor. Therefore, how to improve the accuracy of human injury damage assessment becomes a problem which needs to be solved urgently.

Disclosure of Invention

In view of the above, there is a need to provide a method, an apparatus, a device and a storage medium for damage assessment, which can solve the technical problem of how to improve the accuracy of damage assessment.

In one aspect, the present application provides a damage assessment and claim method, where the damage assessment and claim method includes:

the method comprises the steps of obtaining an initial label of a loss assessment object, initial label information corresponding to the initial label and a loss assessment form image, identifying the loss assessment form image by using a pre-trained character identification model, obtaining a keyword label and a text keyword of the loss assessment form image, generating keyword label information corresponding to the keyword label based on the corresponding label position and text position of the keyword label and the text keyword in the loss assessment form image, inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset loss assessment item and a characteristic coefficient of the preset loss assessment item, adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset loss assessment item, correcting the target coefficient based on a preset loss assessment standard library to obtain a loss assessment coefficient of the preset loss assessment item, and calculating loss assessment data of the loss assessment object according to the preset loss assessment items and the loss assessment coefficients corresponding to each preset loss assessment item.

According to an optional embodiment of the present application, the pre-trained character recognition model includes a cyclic sequence generation network and a decoding network, and the recognizing the claim form image by using the pre-trained character recognition model to obtain the keyword tag and the text keyword of the claim form image includes:

the method comprises the steps of positioning the position of text information in a claim form image based on pixel values of pixels in the claim form image to obtain a character position, performing feature extraction on the text information based on the character position to obtain a feature sequence, inputting the feature sequence into a cyclic sequence generation network to obtain a cyclic sequence, decoding the cyclic sequence based on a decoding network to obtain character information of each loss-assessment image, determining a first preset keyword in the character information as a keyword tag, and determining a second preset keyword in the character information as the text keyword.

According to an optional embodiment of the present application, the generating of the keyword tag information corresponding to the keyword tag based on the corresponding tag position and text position of the keyword tag and the text keyword in the claim form image includes:

detecting whether a label rectangular area corresponding to the keyword label is intersected with a text rectangular area corresponding to the text keyword according to the label position and the text position, if the label rectangular area is intersected with the text rectangular area, calculating an intersection area of the label rectangular area and each intersected text rectangular area, calculating a first area ratio of the intersection area on the label rectangular area, calculating a second area ratio of the intersection area on the text rectangular area, selecting the text rectangular area with the first area ratio and the second area ratio both larger than a preset threshold value as a target rectangular area, and generating keyword label information corresponding to the keyword label based on the number of the target rectangular areas and the text keyword corresponding to the target rectangular area.

According to an optional embodiment of the present application, the generating, based on the number of the target rectangular areas and the text keywords corresponding to the target rectangular areas, keyword tag information corresponding to the keyword tag includes:

if the number of the target rectangular areas is single, determining the text keywords corresponding to the target rectangular areas as keyword label information corresponding to the keyword labels, or if the number of the target rectangular areas is multiple, performing weighting and operation on each first area ratio and the corresponding second area ratio to obtain a final score value of each target rectangular area, and selecting the text keywords corresponding to the target rectangular area with the largest final score value as the keyword label information corresponding to the keyword labels.

According to an optional embodiment of the present application, before inputting the initial label, the initial label information, the keyword label, and the keyword label information into a preset feature coefficient model, the method further includes:

the method comprises the steps of obtaining a preset antagonistic neural network, obtaining position information to which a damage assessment object belongs and training data corresponding to the position information, selecting preset feature data from the training data based on preset feature keywords, and training the preset antagonistic neural network based on the preset feature data to obtain a preset feature coefficient model.

According to an optional embodiment of the present application, the general coefficient model includes a plurality of preset general disability levels and a general correspondence corresponding to each preset general disability level, the general coefficient includes a general disability coefficient, a general base number and a general month number, the inputting of the initial tag, the initial tag information, the keyword tag and the keyword tag information into the preset general coefficient model calculates a general coefficient of a preset damage item, including:

selecting a damage label from the initial label and the keyword label based on a preset damage keyword, taking a numerical value corresponding to the damage label in the claims form image as an initial damage grade of the damage assessment object, determining a general proportion coefficient corresponding to a preset general damage grade which is the same as the initial damage grade in the general coefficient model as the general damage coefficient, determining a general corresponding relation corresponding to the preset general damage grade in the general coefficient model as a target corresponding relation, identifying an initial base label corresponding to a general base label in the target corresponding relation and an initial month label corresponding to a general month label in the target corresponding relation from the initial label and the keyword label, calculating the general month number based on the initial base value of the initial base label and a general month number calculation mode in the target corresponding relation, and calculating the month number based on the initial month number of the initial base label and the general month number calculation mode in the target corresponding relation.

According to an optional embodiment of the present application, the adjusting the general coefficient based on the feature coefficient to obtain the target coefficient of the preset damage assessment item includes:

detecting whether the characteristic coefficients correspond to each general coefficient, if the characteristic coefficients do not correspond to the general coefficients, determining the general coefficients as the target coefficients, or if at least one characteristic coefficient corresponds to the general coefficient, replacing the general coefficients with the corresponding characteristic coefficients, and determining the replaced general coefficients as the target coefficients.

In another aspect, the present invention further provides a damage assessment and claim settlement apparatus, where the apparatus includes:

the system comprises an acquisition unit, a database unit and a database unit, wherein the acquisition unit is used for acquiring an initial label of a damage assessment object, initial label information corresponding to the initial label and a claim form image, and recognizing the claim form image by using a pre-trained character recognition model to obtain a keyword label and a text keyword of the claim form image; the generating unit is used for generating keyword tag information corresponding to the keyword tag based on the corresponding tag position and text position of the keyword tag and the text keyword in the claim form image; the input unit is used for inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset loss assessment item and a characteristic coefficient of the preset loss assessment item; the adjusting unit is used for adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and correcting the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item; and the calculation unit is used for calculating the claim settlement data of the damage assessment object according to the preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

In another aspect, the present application further provides an electronic device, including:

a memory storing computer readable instructions; and

a processor executing computer readable instructions stored in the memory to implement the damage settlement method.

In another aspect, the present application also provides a computer-readable storage medium having computer-readable instructions stored therein, which are executed by a processor in an electronic device to implement the damage settlement method.

According to the technical scheme, the method for correcting the damage assessment object based on the keyword comprises the steps that the claim form image is identified, the information of the damage assessment object can be comprehensively obtained, the text keywords and the numerical keywords in the claim form image information are extracted, redundant information can be removed, the keyword tag information corresponding to the keyword tag is determined based on the tag position and the text position, the keyword tag information corresponds to the keyword tag information, the general coefficient and the characteristic coefficient can be rapidly calculated according to the corresponding relation of the keyword tag and the keyword tag information, the characteristic coefficient is adjusted to obtain the target coefficient corresponding to the preset damage assessment item, the preset characteristic data is more specific information than the preset general data, therefore, the use of the characteristic coefficient model trained by the characteristic data can be improved, the characteristic coefficient output by the characteristic coefficient is more accurate, the general coefficient can be adjusted, the accuracy of the target coefficient can be improved, the target coefficient is corrected based on the preset damage assessment standard library, and the accuracy of the damage assessment result output by the damage assessment standard library is ensured.

Drawings

FIG. 1 is a flow chart of a preferred embodiment of the method of claims damage assessment of the present application.

FIG. 2 is a functional block diagram of the damage assessment method according to the present application.

Fig. 3 is a schematic structural diagram of an electronic device according to a preferred embodiment of the damage assessment method.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in detail below with reference to the accompanying drawings and specific embodiments.

Fig. 1 is a flow chart of a damage-assessment method according to a preferred embodiment of the present application. The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs.

The damage settlement method can acquire and process related data based on an artificial intelligence technology. Among them, artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.

The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

The loss assessment method is applied to one or more electronic devices, which are devices capable of automatically performing numerical calculation and/or information processing according to computer readable instructions set in advance or stored in advance, and the hardware of the electronic devices includes, but is not limited to, a microprocessor, an Application Specific Integrated Circuit (ASIC), a programmable gate array (FPGA), a Digital Signal Processor (DSP), an embedded device, and the like.

The electronic device may be any electronic product capable of performing human-computer interaction with a user, for example, a personal computer, a tablet computer, a smart phone, a Personal Digital Assistant (PDA), a game machine, an interactive network television (IPTV), an intelligent wearable device, and the like.

The electronic device may include a network device and/or a user device. Wherein the network device includes, but is not limited to, a single network electronic device, an electronic device group consisting of a plurality of network electronic devices, or a cloud computing (cloud computing) based cloud consisting of a large number of hosts or network electronic devices.

The network in which the electronic device is located includes, but is not limited to: the internet, a wide area network, a metropolitan area network, a local area network, a Virtual Private Network (VPN), and the like.

101, obtaining an initial tag of a damage assessment object, initial tag information corresponding to the initial tag, and a claim form image, and identifying the claim form image by using a pre-trained character recognition model to obtain a keyword tag and a text keyword of the claim form image.

In at least one embodiment of the present application, the initial tag refers to partial text data of the damage assessment object, the initial tag information refers to keyword information corresponding to the initial tag, and the claim form image refers to an image that includes information required for claim settlement of the damage assessment object. For example: the initial tags include, but are not limited to: the general area where the damage assessment object is located, the social security card number of the damage assessment object, and when the initial tag is the general area where the damage assessment object is located, the initial tag information corresponding to the initial tag may be, for example, shanghai, and the claim settlement form image includes past disability identification information, labor contract performance, household registration nature, gender, date of birth, and the like of the damage assessment object.

In at least one embodiment of the application, the electronic device provides an information input interface for a user to input the initial tag information and import the claim form image through the information input interface. The information input interface comprises an input tag, an input button corresponding to the input tag, an information input field corresponding to the input tag, an image uploading button and an image import field corresponding to the image uploading button, the electronic equipment receives a first trigger instruction generated by pressing the input button by a user, receives initial tag information input by the user in the information input field of the input tag corresponding to the initial tag according to the first trigger instruction, receives a second trigger instruction generated by pressing the image uploading button by the user, and receives a claims form image uploaded by the user in the image import field according to the second trigger instruction.

In at least one embodiment of the application, the electronic device identifies the claim form image by using a pre-trained character recognition model, and obtains a keyword tag and a text keyword of the claim form image.

In at least one embodiment of the present application, the text recognition model refers to a model that identifies text information and numerical information in the claim form image.

In at least one embodiment of the present application, the keyword tag includes a first keyword tag and a second keyword tag, the text keyword includes a text keyword and a numerical keyword, the first keyword tag indicates a tag corresponding to the text keyword, the second keyword tag indicates a tag corresponding to the numerical keyword, the text keyword indicates specific information of a fixed loss object corresponding to the first keyword tag, and the numerical keyword indicates specific information of a number of the fixed loss object corresponding to the second keyword tag. For example, when the first keyword tag is gender, the text keyword is male, when the second keyword tag is birth date, the numerical keyword is 1987-12-28, when the second keyword tag is age, the numerical keyword is 35, and so on.

In at least one embodiment of the present application, the text recognition model includes a cyclic sequence generation network and a decoding network, and the electronic device uses a pre-trained text recognition model to recognize the claim form image, so as to obtain a keyword tag and a text keyword of the claim form image, including:

the electronic equipment positions the position of text information in the claim form image based on the pixel value of pixel point in the claim form image to obtain a text position, further, the electronic equipment performs feature extraction on the text information based on the text position to obtain a feature sequence, further, the electronic equipment inputs the feature sequence into a cyclic sequence generation network to obtain a cyclic sequence, further, the electronic equipment decodes the cyclic sequence based on the decoding network to obtain the text information of each loss-assessment image, further, the electronic equipment determines a first preset keyword in the text information as the keyword tag, and determines a second preset keyword in the text information as the text keyword.

In this embodiment, the first preset keyword includes, but is not limited to: gender, occupation category, age, birth date, monthly average wage, and the like, and the second preset keywords include, but are not limited to: male, female, technical, 35, 1987-12-28 and 8000, and so on.

Specifically, the electronic device performs feature extraction on the text information based on a feature extraction network to obtain the feature sequence, where the feature extraction network may be a rescet 50 network, the cyclic sequence generation network may be a bidirectional long-short term memory neural network, and the decoding network may be a neural network-based time sequence class classification algorithm (CTC).

Specifically, the electronic device locates the position of the text information in the claim form image based on the pixel values of the pixel points in the claim form image, and obtains the text position, including:

the electronic equipment acquires pixel positions of all pixel points in the claim form image, the electronic equipment traverses pixel values of all pixel points in the claim form image, and further the electronic equipment determines a plurality of pixel points corresponding to the pixel values larger than a preset value as text information and determines the pixel positions of the corresponding pixel points as the text positions.

Specifically, the cyclic sequence generation network includes a convolutional layer and a pooling layer, and the electronic device inputs the feature sequence into the cyclic sequence generation network to obtain a cyclic sequence, including:

the electronic device divides the feature sequence to obtain a plurality of division features, further performs convolution operation on the division features based on the convolution layer to obtain convolution results, further performs pooling processing on the convolution results based on the pooling layer to obtain feature vectors corresponding to each division feature, and further performs splicing on the feature vectors to obtain the cyclic sequence.

The segmentation process is prior art, and is not described herein in detail.

Specifically, the electronic device decodes the cyclic sequence based on the decoding network to obtain the text information of each damaged image, including:

the electronic equipment obtains a characteristic sequence corresponding to each characteristic vector in the cyclic sequence, maps each characteristic vector in the cyclic sequence based on a pre-constructed dictionary to obtain a characteristic character corresponding to each characteristic vector, and splices a plurality of characteristic characters according to the characteristic sequence to obtain the text information of each fixed loss image.

And the pre-constructed dictionary stores the corresponding relation between each feature vector and each feature character, and the feature characters are spliced according to the corresponding feature sequence to obtain the text information of each damaged image.

In this embodiment, the accuracy of the text information can be ensured by combining a plurality of feature characters according to the feature sequence.

102, generating keyword tag information corresponding to the keyword tag based on the tag position and the text position of the keyword tag and the text keyword in the claim form image.

In at least one embodiment of the present application, the generating, by the electronic device, keyword tag information corresponding to the keyword tag based on a tag position and a text position of the keyword tag and the text keyword in the claim form image includes:

the electronic equipment detects whether a label rectangular area corresponding to the keyword label is intersected with a text rectangular area corresponding to the text keyword according to the label position and the text position, if the label rectangular area is intersected with the text rectangular area, the electronic equipment calculates an intersection area of the label rectangular area and each intersected text rectangular area, further, the electronic equipment calculates a first area ratio of the intersection area on the label rectangular area and calculates a second area ratio of the intersection area on the text rectangular area, further, the electronic equipment selects the text rectangular area with the first area ratio and the second area ratio larger than a preset threshold value as a target rectangular area, and generates keyword label information corresponding to the keyword label based on the number of the target rectangular areas and the text keyword corresponding to the target rectangular area.

Specifically, the positioning manner of the tag position includes:

the electronic equipment acquires a minimum abscissa value corresponding to a pixel point in the rectangular tag region, acquires a minimum ordinate value corresponding to the pixel point in the rectangular tag region, acquires a maximum abscissa value corresponding to the pixel point in the rectangular tag region, and acquires a maximum ordinate value corresponding to the pixel point in the rectangular tag region, and the electronic equipment takes a first abscissa interval formed by the minimum abscissa value and the maximum abscissa value and a first ordinate interval formed by the minimum ordinate value and the maximum ordinate value as the tag position.

Specifically, the detecting, by the electronic device, whether a tag rectangular region corresponding to the keyword tag intersects with a text rectangular region corresponding to the text keyword according to the tag position and the text position includes:

the electronic equipment identifies whether the first abscissa interval and the second abscissa interval are overlapped, identifies whether the first ordinate interval and the second ordinate interval are overlapped, and determines that the label rectangular area is intersected with the text rectangular area if the first abscissa interval and the second abscissa interval are overlapped and the first ordinate interval and the second ordinate interval are overlapped.

Specifically, the generating, by the electronic device, keyword tag information corresponding to the keyword tag based on the number of the target rectangular areas and the text keyword corresponding to the target rectangular area includes:

if the number of the target rectangular areas is single, the electronic equipment determines the text keywords corresponding to the target rectangular areas as the keyword label information corresponding to the keyword labels, or if the number of the target rectangular areas is multiple, the electronic equipment performs weighting and operation on each first area ratio and the corresponding second area ratio to obtain the final score value of each target rectangular area, and selects the text keywords corresponding to the target rectangular area with the largest final score value as the keyword label information corresponding to the keyword labels.

In this embodiment, when the number of the target rectangular areas is multiple, the text keyword corresponding to the target rectangular area corresponding to the largest final score value is selected as the keyword tag information corresponding to the keyword tag, and since the final score value is generated through weighting operation, the keyword tag information most matched with the keyword tag can be reasonably selected.

103, inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset damage assessment item and a characteristic coefficient of the preset damage assessment item.

In at least one embodiment of the present application, the general coefficient model is configured to calculate a general coefficient, the feature coefficient model is configured to calculate a feature coefficient, and an intersection exists between the preset general data and the preset feature data. For example, the common coefficients include: the general base 8600, the general month number 18 and the general disability coefficient 0.8, and the characteristic coefficients comprise a characteristic base 8700 and a characteristic disability coefficient 0.8.

In at least one embodiment of the present application, before inputting the initial label, the initial label information, the keyword label, and the keyword label information to a preset feature coefficient model, the method further includes:

the electronic equipment acquires a preset antagonistic neural network, acquires the position information to which the damage assessment object belongs and training data corresponding to the position information, selects preset characteristic data from the training data based on preset characteristic keywords, and further trains the preset antagonistic neural network based on the preset characteristic data to obtain the preset characteristic coefficient model.

The training data corresponding to the position information refers to claim settlement information of a plurality of damage assessment objects in the position information, and the training data comprises the preset general data and the preset characteristic data. The preset feature keywords comprise the occupation category of the damage assessment object, the type of the labor contract and the like. For example, the preset feature keyword may be 2022106 road and bridge engineering technician, fixed term labor contract, or the like.

In at least one embodiment of the present application, the general coefficient model includes a plurality of preset general disability levels, a general proportionality coefficient corresponding to each preset general disability level, and a general correspondence corresponding to each preset general disability level, the general correspondence corresponding to each preset general disability level includes a general base number tag and a calculation method of a general base number corresponding to the general base number tag, and a calculation method of a general month number tag and a general month number corresponding to the general month number tag, and the general coefficient includes a general disability coefficient, a general base number, and a general month number.

In at least one embodiment of the present application, the step of inputting, by the electronic device, the initial tag information, the keyword tag, and the keyword tag information into a preset general coefficient model to calculate a general coefficient of a preset damage assessment item includes:

the electronic equipment selects a disability label from the initial label and the keyword label based on a preset disability keyword, takes a numerical value corresponding to the disability label in the claims form image as an initial disability grade of the damage assessment object, further determines a general proportion coefficient corresponding to a preset general disability grade identical to the initial disability grade in the general coefficient model as the general disability coefficient, determines a general corresponding relation corresponding to the preset general disability grade in the general coefficient model as a target corresponding relation, further identifies an initial base label corresponding to a general base label in the target corresponding relation and an initial month label corresponding to a general month label in the target corresponding relation from the initial label and the keyword label, and further calculates the general base based on an initial base numerical value of the initial base label and a general base calculation mode in the target corresponding relation, and calculates the general month number based on the general month calculation mode in the initial base numerical value and the target month corresponding relation.

Wherein the predetermined impairment term includes, but is not limited to: disposable industrial injury claim and repayment and disposable disability employment fund.

For example, the initial tag and the initial tag information corresponding to the initial tag are: the working address is as follows: shanghai and social security card number: 000888xxxx, wherein the keyword tag and the keyword tag information corresponding to the keyword tag are average payroll per month: 8000, gender: male, date of birth: 1987-12-28, age: 35, contract type: no fixed term labor contract, and so on.

The initial cardinal number label comprises monthly average wages, the initial monthly number label comprises disability grades, gender, birth date and age, the preset disability keywords can be the disability grades, the preset general disability grades comprise first-grade disability to tenth-grade disability, and the range of the proportional coefficient corresponding to each preset general disability grade is [0,1]. For example, the initial damage level is three-level damage, and the proportionality coefficient for the three-level damage is 0.7.

In this embodiment, when the initial tag includes the initial radix tag, the initial radix value is initial tag information corresponding to the initial tag, and when the keyword tag includes the initial radix tag, the initial radix value is keyword tag information corresponding to the keyword tag.

The calculation mode of the universal cardinality comprises the following steps:

determining a reference base value according to the region where the damage assessment object is located, determining a first wage threshold value and a second wage threshold value according to the reference base value, wherein the first wage threshold value is larger than the second wage threshold value, comparing the first wage threshold value with the initial base value, determining the first wage threshold value as the universal base number if the initial base value is larger than or equal to the first wage threshold value, determining the second wage threshold value as the universal base number if the initial base value is smaller than or equal to the second wage threshold value, and determining the initial base value as the universal base number if the initial base value corresponding to the initial base number label is between the first wage threshold value and the second wage threshold value.

For example, the reference base value may be an average payroll of the employees in a general area where the damage assessment object is located, the first payroll threshold may be 300% of the average payroll of the employees, and the second payroll threshold may be 60% of the average payroll of the employees.

The calculation mode of the general month number comprises the following steps:

and constructing a month number mapping table based on a preset damage assessment item, a preset birth date, a preset age, a preset disability level and a preset month number, mapping the initial month number label based on the month number mapping table, and determining the preset month number corresponding to the initial month number label as the universal month number.

In this embodiment, the common coefficient can be accurately calculated from the specific information of the damage assessment target by the common radix calculation method and the common month calculation method.

In at least one embodiment of the present application, the feature coefficient model includes a tag base number value corresponding to the feature base number tag, a feature month number tag, a tag month number value corresponding to the feature month number tag, a feature coefficient tag, and a feature scaling coefficient corresponding to the feature coefficient tag, where the feature coefficient includes a feature base number, a feature month number, and a feature disability coefficient.

In at least one embodiment of the present application, the step of inputting, by the electronic device, the initial tag information, the keyword tag, and the keyword tag information into a preset feature coefficient model to calculate a feature coefficient of the preset impairment term includes:

the electronic equipment acquires a radix label type corresponding to the characteristic radix label, and detects whether a label exists in the initial label and the keyword label and corresponds to the characteristic radix label based on the radix label type.

In this embodiment, if a tag exists in the initial tag or the keyword tag and corresponds to the feature radix tag, the electronic device determines a tag radix value corresponding to the feature radix tag as the feature radix, and the electronic device obtains a month tag type corresponding to the feature month tag and detects whether the initial tag and the keyword tag have a tag corresponding to the feature month tag based on the month tag type.

In this embodiment, if a tag exists in the initial tag or the keyword tag and corresponds to the feature month number tag, the electronic device determines a tag month number value corresponding to the feature month number tag as the feature month number, and the electronic device obtains a coefficient tag type corresponding to the feature coefficient tag and detects whether the tag exists in the initial tag and the keyword tag and corresponds to the feature coefficient tag based on the coefficient tag type.

In this embodiment, if there is a label in the initial label or the keyword label corresponding to the feature coefficient label, the electronic device determines a label coefficient value corresponding to the feature coefficient label as the feature disability coefficient.

Wherein the feature cardinality labels include, but are not limited to: disability level, monthly average wages, household property, whether a labor contract is made, work address, property of employment unit, type and category of job when a labor contract is made, the characteristic monthly tags include, but are not limited to: disability level, age, gender, category of occupation, household nature, work address, whether a work contract has been made and whether a work relationship with employment units has been preserved so far, including, but not limited to: disability level, household registration nature.

With the above embodiment, since the feature base number label, the feature month number label, and the feature coefficient label relate to more specific information on the damage assessment target, the feature coefficient can be more accurate than the general coefficient.

And 104, adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and correcting the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item.

In at least one embodiment of the present application, the electronic device detects whether the feature coefficient corresponds to each general coefficient, and if the feature coefficient does not correspond to each general coefficient, the electronic device determines the general coefficient as the target coefficient, or if at least one feature coefficient corresponds to the general coefficient, the electronic device replaces the general coefficient with a corresponding feature coefficient, and determines the replaced general coefficient as the target coefficient.

For example, the feature base number corresponds to the general base number, the feature scale coefficient corresponds to the general scale coefficient, and the general disability coefficient corresponds to the feature disability coefficient. For example, when the general-purpose base 8600, the general-purpose month number 18, and the general-purpose disability coefficient are 0.8, and the characteristic base 8700 and the characteristic disability coefficient are 0.8, the general-purpose base 8600 corresponds to the characteristic base 8700, and the general-purpose disability coefficient 0.8 corresponds to the characteristic disability coefficient 0.8.

Through the embodiment, when at least one characteristic coefficient corresponds to the general coefficient, the general coefficient is replaced by the corresponding characteristic coefficient to obtain the target coefficient, and the accuracy of the target coefficient can be improved because the characteristic coefficient is more accurate than the general coefficient.

In at least one embodiment of the present application, the library of predetermined damage criteria includes, but is not limited to: a national standard library for disability, a standard library for indemnification and a disability determination library. The disabled country standard library is used for storing disabled judgment standards corresponding to all geographic areas respectively, the indemnity standard library is used for storing indemnity standards issued according to regional maintenance national statistics bureaus, and the disabled country standard library is used for storing corresponding relations among the geographic areas, the age lower limit, the age upper limit and the indemnity evaluation algorithm.

In at least one embodiment of the present application, the electronic device corrects the target coefficient based on a preset damage standard library, and obtaining a damage coefficient of the preset damage item includes:

the electronic device inputs the initial tag, the initial tag information, the keyword tag and the keyword tag information into the preset loss standard library, and determines a loss tag associated with each preset loss item according to a plurality of preset loss items matched with a loss object, further, the electronic device determines a matching parameter of the loss object on the loss tag associated with each preset loss item according to information data stored in the preset loss standard library, further, the electronic device compares the target coefficient with the matching parameter, if the difference between the target coefficient and the matching parameter is less than or equal to a configuration value, the electronic device determines the target coefficient as the loss coefficient, or if the difference between the target coefficient and the matching parameter is greater than the configuration value, the electronic device determines the matching parameter as the loss coefficient.

For example, when a predetermined damage-assessment item is a one-time industrial damage compensation, the damage label of the one-time industrial damage compensation includes a compensation base value, a disability coefficient, and a compensation month number. The damage standard library stores an indemnity base value of 8000, a damage coefficient of 0.8 and a number of compensation months of 16, when the target coefficient includes a target base value 8050, a target damage coefficient of 0.8 and a target number of compensation months of 18, the indemnity base value 8000 and a first configuration value corresponding to the target base value 8050 are 100, the damage coefficient of 0.8 and a second configuration value corresponding to the target damage coefficient of 0.8 are 0.1, and the number of compensation months of 16 and the target number of compensation months of 18 are 1, the damage base value of the damage coefficient is 8000, the damage coefficient of 0.8, and the number of compensation months of 18.

Through the above embodiment, the matching parameters are generated by analyzing all information of the damage assessment object through the preset damage assessment standard library, so that logical operation processing is performed according to the obtained matching parameters to complete the output of the damage assessment coefficients, thereby not only realizing the automatic calculation of the damage assessment of the human injury, but also improving the accuracy of the damage assessment coefficients.

105, calculating the claim settlement data of the damage assessment object according to the preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

It is emphasized that, to further ensure the privacy and security of the claim data, the claim data can also be stored in a node of a blockchain.

In at least one embodiment of the present application, the damage assessment coefficient includes a damage assessment base number, a damage assessment month number, and a damage assessment residual coefficient, and the calculating, by the electronic device, the claim settlement data of the damage assessment object according to a plurality of preset damage assessment items and the damage assessment coefficients corresponding to each preset damage assessment item includes:

and the electronic equipment multiplies the loss assessment base number, the loss assessment month number and the loss assessment residue coefficient to obtain a loss assessment amount corresponding to each preset loss assessment item, and further adds the loss assessment amounts corresponding to a plurality of preset loss assessment items to obtain the claim settlement data.

In the above embodiment, the claims data can be calculated quickly by multiplying and adding the damage assessment base number, the damage assessment month number, and the damage assessment residual coefficient corresponding to the plurality of preset damage assessment items.

Fig. 2 is a functional block diagram of the damage settlement apparatus according to the preferred embodiment of the present invention. The damage settlement apparatus 11 includes an acquiring unit 110, a generating unit 111, an inputting unit 112, an adjusting unit 113, and a calculating unit 114. The module/unit referred to herein is a series of computer readable instruction segments that can be accessed by the processor 13 and perform a fixed function and that are stored in the memory 12. In the present embodiment, the functions of the modules/units will be described in detail in the following embodiments.

The obtaining unit 110 obtains an initial tag of a damage assessment object, initial tag information corresponding to the initial tag, and a claim form image, and identifies the claim form image by using a pre-trained character recognition model to obtain a keyword tag and a text keyword of the claim form image.

In at least one embodiment of the present application, the initial tag refers to partial text data of the damage assessment object, the initial tag information refers to keyword information corresponding to the initial tag, and the claim form image refers to an image including information required for claim settlement of the damage assessment object. For example: the initial tags include, but are not limited to: the general area where the damage assessment object is located, the social security card number of the damage assessment object, and when the initial tag is the general area where the damage assessment object is located, the initial tag information corresponding to the initial tag may be, for example, shanghai, and the claim settlement form image includes past disability identification information, labor contract performance, household registration nature, gender, date of birth, and the like of the damage assessment object.

In at least one embodiment of the present application, the obtaining unit 110 provides an information input interface, so that a user can input the initial tag information and import the claim form image through the information input interface. The information input interface comprises an input tag, an input button corresponding to the input tag, an information input field corresponding to the input tag, an image uploading button and an image import field corresponding to the image uploading button, the acquisition unit 110 receives a first trigger instruction generated by pressing the input button by a user, receives initial tag information input by the user in the information input field of the input tag corresponding to the initial tag according to the first trigger instruction, and the acquisition unit 110 receives a second trigger instruction generated by pressing the image uploading button by the user, and receives a claims form image uploaded by the user in the image import field according to the second trigger instruction.

In at least one embodiment of the present application, the obtaining unit 110 uses a pre-trained character recognition model to identify the claim form image, and obtains a keyword tag and a text keyword of the claim form image.

In at least one embodiment of the present application, the text recognition model includes a cyclic sequence generation network and a decoding network, and the obtaining unit 110 uses a pre-trained text recognition model to recognize the claim form image, so as to obtain the keyword tag and the text keyword of the claim form image, including:

the obtaining unit 110 locates a position of text information in the claim form image based on pixel values of pixels in the claim form image to obtain a text position, further, the obtaining unit 110 performs feature extraction on the text information based on the text position to obtain a feature sequence, further, the obtaining unit 110 inputs the feature sequence into the cyclic sequence generation network to obtain a cyclic sequence, further, the obtaining unit 110 decodes the cyclic sequence based on the decoding network to obtain text information of each loss-assessment image, further, the obtaining unit 110 determines a first preset keyword in the text information as the keyword tag and determines a second preset keyword in the text information as the text keyword.

Specifically, the obtaining unit 110 performs feature extraction on the text information based on a feature extraction network to obtain the feature sequence, where the feature extraction network may be a rescet 50 network, the cyclic sequence generation network may be a bidirectional long-short term memory neural network, and the decoding network may be a timing class classification algorithm (CTC) based on a neural network.

Specifically, the obtaining unit 110 locates the position of the text information in the claim form image based on the pixel values of the pixel points in the claim form image, and obtains the text position, including:

the obtaining unit 110 obtains pixel positions of all pixel points in the claim form image, the obtaining unit 110 traverses pixel values of all pixel points in the claim form image, and further, the obtaining unit 110 determines a plurality of pixel points corresponding to the pixel values larger than a preset value as text information, and determines the pixel positions of the corresponding pixel points as the text positions.

Specifically, the cyclic sequence generation network includes a convolutional layer and a pooling layer, and the obtaining unit 110 inputs the feature sequence into the cyclic sequence generation network to obtain a cyclic sequence, which includes:

the obtaining unit 110 performs segmentation processing on the feature sequence to obtain a plurality of segmentation features, further, the obtaining unit 110 performs convolution operation on the plurality of segmentation features based on the convolution layer to obtain a convolution result, further, the obtaining unit 110 performs pooling processing on the convolution result based on the pooling layer to obtain a feature vector corresponding to each segmentation feature, and further, the obtaining unit 110 splices the plurality of feature vectors to obtain the cyclic sequence.

The segmentation process is prior art, and is not described herein in detail.

Specifically, the obtaining unit 110 decodes the cyclic sequence based on the decoding network to obtain the text information of each damaged image, including:

the obtaining unit 110 obtains a feature sequence corresponding to each feature vector in the cyclic sequence, maps each feature vector in the cyclic sequence based on a pre-constructed dictionary to obtain a feature character corresponding to each feature vector, and the obtaining unit 110 splices a plurality of feature characters according to the feature sequence to obtain text information of each loss-assessment image.

In this embodiment, the accuracy of the text information can be ensured by combining a plurality of the characteristic characters according to the characteristic sequence.

The generating unit 111 generates keyword tag information corresponding to the keyword tag based on the tag position and the text position of the keyword tag and the text keyword in the claim form image.

In at least one embodiment of the present application, the generating unit 111 generates keyword tag information corresponding to the keyword tag based on the corresponding tag position and text position of the keyword tag and the text keyword in the claim form image, including:

the generating unit 111 detects whether a label rectangular region corresponding to the keyword label is intersected with a text rectangular region corresponding to the text keyword according to the label position and the text position, if the label rectangular region is intersected with the text rectangular region, the generating unit 111 calculates an intersection region of the label rectangular region and each intersected text rectangular region, further, the generating unit 111 calculates a first area ratio of the intersection region on the label rectangular region, calculates a second area ratio of the intersection region on the text rectangular region, further, the generating unit 111 selects a text rectangular region with the first area ratio and the second area ratio both greater than a preset threshold as a target rectangular region, and generates keyword label information corresponding to the keyword label based on the number of the target rectangular regions and the text keyword corresponding to the target rectangular region.

Specifically, the positioning manner of the tag position includes:

the generating unit 111 acquires a minimum abscissa value corresponding to a pixel point in the rectangular tag region, acquires a minimum ordinate value corresponding to a pixel point in the rectangular tag region, acquires a maximum abscissa value corresponding to a pixel point in the rectangular tag region, and acquires a maximum ordinate value corresponding to a pixel point in the rectangular tag region, and the generating unit 111 sets a first abscissa interval formed by the minimum abscissa value and the maximum abscissa value and a first ordinate interval formed by the minimum ordinate value and the maximum ordinate value as the tag positions.

Specifically, the step of detecting, by the generating unit 111, whether a tag rectangular region corresponding to the keyword tag intersects with a text rectangular region corresponding to the text keyword according to the tag position and the text position includes:

the generating unit 111 identifies whether the first abscissa interval and the second abscissa interval overlap, and identifies whether the first ordinate interval and the second ordinate interval overlap, and determines that the label rectangular region intersects the text rectangular region if the first abscissa interval and the second abscissa interval overlap, and the first ordinate interval and the second ordinate interval overlap.

Specifically, the generating unit 111 generates keyword tag information corresponding to the keyword tag based on the number of the target rectangular areas and the text keyword corresponding to the target rectangular area, including:

if the number of the target rectangular areas is single, the generating unit 111 determines the text keyword corresponding to the target rectangular area as the keyword tag information corresponding to the keyword tag, or if the number of the target rectangular areas is multiple, the generating unit 111 performs weighting and operation on each first area ratio and the corresponding second area ratio to obtain a final score value of each target rectangular area, and selects the text keyword corresponding to the target rectangular area with the largest final score value as the keyword tag information corresponding to the keyword tag.

The input unit 112 inputs the initial label, the initial label information, the keyword label, and the keyword label information into a preset general coefficient model and a preset feature coefficient model respectively to calculate a general coefficient of a preset damage assessment item and a feature coefficient of the preset damage assessment item.

In at least one embodiment of the present application, the general coefficient model is configured to calculate a general coefficient, the feature coefficient model is configured to calculate a feature coefficient, and an intersection exists between the preset general data and the preset feature data. For example, the general coefficients include: the general base 8600, the general month number 18 and the general disability coefficient 0.8, and the characteristic coefficients comprise a characteristic base 8700 and a characteristic disability coefficient 0.8.

the input unit 112 is further configured to obtain a preset antagonistic neural network, obtain position information to which the damage assessment object belongs and training data corresponding to the position information, and select preset feature data from the training data based on a preset feature keyword, and further, the input unit 112 trains the preset antagonistic neural network based on the preset feature data to obtain the preset feature coefficient model.

In at least one embodiment of the present application, the inputting, by the input unit 112, the initial label information, the keyword label, and the keyword label information into a preset general coefficient model to calculate a general coefficient of a preset damage item includes:

the input unit 112 selects a disability label from the initial label and the keyword label based on a preset disability keyword, and uses a numerical value corresponding to the disability label in the claims form image as an initial disability grade of the damage assessment target, further, the input unit 112 determines a general proportion coefficient corresponding to a preset general disability grade identical to the initial disability grade in the general coefficient model as the general disability coefficient, and determines a general correspondence corresponding to the preset general disability grade in the general coefficient model as a target correspondence, further, the input unit 112 identifies an initial base label corresponding to a general base label in the target correspondence and an initial month label corresponding to a general month label in the target correspondence from the initial label and the keyword label, further, the input unit 112 calculates the general month label based on the initial base numerical value of the initial label and a general month number calculation method in the target correspondence, and calculates the month number based on the general month number calculation method in the initial base number label and the target month number correspondence method.

determining a reference base number value according to the region where the damage assessment object is located, determining a first payroll threshold value and a second payroll threshold value according to the reference base number value, wherein the first payroll threshold value is larger than the second payroll threshold value, comparing the first payroll threshold value with the initial base number value, if the initial base number value is larger than or equal to the first payroll threshold value, determining the first payroll threshold value as the general base number, if the initial base number value is smaller than or equal to the second payroll threshold value, determining the second payroll threshold value as the general base number, and if the initial base number value corresponding to the initial base number label is between the first payroll threshold value and the second payroll threshold value, determining the initial base number value as the general base number.

For example, the reference base value may be an average salary of employees in a planning region where the damage assessment object is located, the first salary threshold may be 300% of the average salary of the employees, and the second salary threshold may be 60% of the average salary of the employees.

The calculation mode of the general month number comprises the following steps:

In at least one embodiment of the present application, the feature coefficient model includes a tag radix value, a feature month number tag, and a tag month value, a feature coefficient tag, and a feature scale coefficient, where the tag month value, the feature month number tag, and the feature month number tag correspond to the feature month number tag, and the feature coefficient includes a feature radix, a feature month number, and a feature disability coefficient.

In at least one embodiment of the present application, the inputting unit 112 inputs the initial label, the initial label information, the keyword label, and the keyword label information into a preset feature coefficient model to calculate a feature coefficient of the preset impairment term includes:

the input unit 112 obtains a radix label type corresponding to the feature radix label, and detects whether a label exists in the initial label and the keyword label and corresponds to the feature radix label based on the radix label type.

In this embodiment, if the initial tag or the keyword tag has a tag corresponding to the feature radix tag, the input unit 112 determines the tag radix value corresponding to the feature radix tag as the feature radix, and the input unit 112 obtains the month tag type corresponding to the feature month tag, and detects whether the initial tag and the keyword tag have a tag corresponding to the feature month tag based on the month tag type.

In this embodiment, if a tag exists in the initial tag or the keyword tag and corresponds to the feature month number tag, the input unit 112 determines a tag month number value corresponding to the feature month number tag as the feature month number, and the input unit 112 obtains a coefficient tag type corresponding to the feature coefficient tag and detects whether the tag exists in the initial tag and the keyword tag and corresponds to the feature coefficient tag based on the coefficient tag type.

In this embodiment, if there is a label in the initial label or the keyword label corresponding to the feature coefficient label, the input unit 112 determines the label coefficient value corresponding to the feature coefficient label as the feature disability coefficient.

The adjusting unit 113 adjusts the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and corrects the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item.

In at least one embodiment of the present application, the adjusting unit 113 detects whether the feature coefficient corresponds to each general coefficient, if the feature coefficient does not correspond to each general coefficient, the adjusting unit 113 determines the general coefficient as the target coefficient, or if at least one feature coefficient corresponds to the general coefficient, the adjusting unit 113 replaces the general coefficient with the corresponding feature coefficient and determines the replaced general coefficient as the target coefficient.

In at least one embodiment of the present application, the library of predetermined damage criteria includes, but is not limited to: a country of disability standard library, a compensation standard library and a disability determination library. The disabled country standard library is used for storing disabled judgment standards corresponding to all geographic areas respectively, the indemnity standard library is used for storing indemnity standards issued according to regional maintenance national statistics bureaus, and the disabled country standard library is used for storing corresponding relations among the geographic areas, the age lower limit, the age upper limit and the indemnity evaluation algorithm.

In at least one embodiment of the present application, the adjusting unit 113 corrects the target coefficient based on a preset damage standard library, and obtaining the damage coefficient of the preset damage item includes:

the adjusting unit 113 inputs the initial tag, the initial tag information, the keyword tag, and the keyword tag information into the preset damage standard library, and determines a damage rating tag associated with each preset damage rating item according to a plurality of preset damage rating items matched with a damage rating object, further, the adjusting unit 113 determines a matching parameter of the damage rating object on the damage rating tag associated with each preset damage rating item according to information data stored in the preset damage rating standard library, further, the adjusting unit 113 compares the target coefficient with the matching parameter, if a difference between the target coefficient and the matching parameter is less than or equal to a configuration value, the adjusting unit 113 determines the target coefficient as the damage rating coefficient, or if the difference between the target coefficient and the matching parameter is greater than the configuration value, the adjusting unit 113 determines the matching parameter as the damage rating coefficient.

The calculating unit 114 calculates the claim settlement data of the damage assessment object according to the plurality of preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

In at least one embodiment of the present application, the damage assessment coefficients include a damage assessment base number, a damage assessment month number, and a damage assessment residual coefficient, and the calculating unit 114 calculates the claim settlement data of the damage assessment object according to a plurality of the preset damage assessment items and the damage assessment coefficients corresponding to each preset damage assessment item includes:

the calculating unit 114 multiplies the damage assessment base number, the damage assessment month number, and the damage assessment residual coefficient to obtain a damage assessment amount corresponding to each preset damage item, and further, the calculating unit 114 adds up the damage assessment amounts corresponding to the preset damage items to obtain the claim settlement data.

Fig. 3 is a schematic structural diagram of an electronic device according to a preferred embodiment of the method for claim settlement.

In one embodiment of the present application, the electronic device 1 includes, but is not limited to, a memory 12, a processor 13, and computer readable instructions, such as a damage settlement program, stored in the memory 12 and executable on the processor 13.

It will be appreciated by a person skilled in the art that the schematic diagram is only an example of the electronic device 1 and does not constitute a limitation of the electronic device 1, and that it may comprise more or less components than shown, or some components may be combined, or different components, e.g. the electronic device 1 may further comprise an input output device, a network access device, a bus, etc.

The processor 13 may be a Central Processing Unit (CPU), other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The processor 13 is an operation core and a control center of the electronic device 1, and is connected to each part of the whole electronic device 1 by various interfaces and lines, and executes an operating system of the electronic device 1 and various installed application programs, program codes, and the like.

Illustratively, the computer readable instructions may be divided into one or more modules/units, such as the acquisition unit, generation unit, input unit, adjustment unit, calculation unit, etc. shown in fig. 2, which are stored in the memory 12 and executed by the processor 13 to accomplish the present application. The one or more modules/units may be a series of computer readable instruction segments capable of performing specific functions, which are used for describing the execution process of the computer readable instructions in the electronic device 1.

The memory 12 may be used for storing the computer readable instructions and/or modules, and the processor 13 implements various functions of the electronic device 1 by executing or executing the computer readable instructions and/or modules stored in the memory 12 and invoking data stored in the memory 12. The memory 12 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data created according to use of the electronic device, and the like. The memory 12 may include non-volatile and volatile memories, such as: a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card (FlashCard), at least one disk storage device, a flash memory device, or other storage device.

The memory 12 may be an external memory and/or an internal memory of the electronic device 1. Further, the memory 12 may be a memory in a physical form, such as a memory stick, a TF card (Trans-flash card), and the like.

The integrated modules/units of the electronic device 1 may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, all or part of the flow in the method of the embodiments described above can be realized by the present application, and can also be realized by hardware related to computer readable instructions, which can be stored in a computer readable storage medium, and when the computer readable instructions are executed by a processor, the steps of the above described method embodiments can be realized.

Wherein the computer readable instructions comprise computer readable instruction code which may be in source code form, object code form, an executable file or some intermediate form, and the like. The computer-readable medium may include: any entity or device capable of carrying said computer readable instruction code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random Access Memory (RAM).

The block chain referred by the application is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.

With reference to fig. 1, the memory 12 of the electronic device 1 stores computer-readable instructions to implement a damage settlement method, and the processor 13 can execute the computer-readable instructions to implement:

acquiring an initial label of a damage assessment object, initial label information corresponding to the initial label and a claim form image, and identifying the claim form image by using a pre-trained character identification model to obtain a keyword label and a text keyword of the claim form image; generating keyword tag information corresponding to the keyword tag based on the corresponding tag position and text position of the keyword tag and the text keyword in the claim form image; inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset damage assessment item and a characteristic coefficient of the preset damage assessment item; adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and correcting the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item; and calculating the claim settlement data of the damage assessment object according to the preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

Specifically, the processor 13 may refer to the description of the relevant steps in the embodiment corresponding to fig. 1 for a specific implementation method of the computer readable instructions, which is not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is only one logical functional division, and other divisions may be realized in practice.

The computer readable storage medium has computer readable instructions stored thereon, wherein the computer readable instructions when executed by the processor 13 are configured to implement the steps of:

acquiring an initial label of a damage assessment object, initial label information corresponding to the initial label and a claim form image, and identifying the claim form image by using a pre-trained character identification model to obtain a keyword label and a text keyword of the claim form image; generating keyword tag information corresponding to the keyword tag based on the keyword tag and the text keyword corresponding to the tag position and the text position in the claim form image; inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset damage assessment item and a characteristic coefficient of the preset damage assessment item; adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and correcting the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item; and calculating the claim settlement data of the damage assessment object according to the preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.

In addition, functional modules in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional module.

The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.

Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. The plurality of units or devices may also be implemented by one unit or device through software or hardware. The terms first, second, etc. are used to denote names, but not any particular order.

Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present application and not for limiting, and although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present application without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A damage assessment method, comprising:

acquiring an initial label of a damage assessment object, initial label information corresponding to the initial label and a claim form image, and identifying the claim form image by using a pre-trained character recognition model to obtain a keyword label and a text keyword of the claim form image;

generating keyword tag information corresponding to the keyword tag based on the keyword tag and the text keyword corresponding to the tag position and the text position in the claim form image;

inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset damage assessment item and a characteristic coefficient of the preset damage assessment item;

adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and correcting the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item;

and calculating the claim settlement data of the damage assessment object according to the preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

2. The damage assessment method of claim 1, wherein the pre-trained text recognition model comprises a recurrent sequence generation network and a decoding network, and the recognition of the claim form image using the pre-trained text recognition model to obtain the keyword tag and the text keyword of the claim form image comprises:

positioning the position of text information in the claim form image based on pixel values of pixel points in the claim form image to obtain a character position;

extracting the characteristics of the text information based on the character positions to obtain a characteristic sequence;

inputting the characteristic sequence into the cyclic sequence generation network to obtain a cyclic sequence;

decoding the cyclic sequence based on the decoding network to obtain the text information of each loss-assessment image;

and determining a first preset keyword in the character information as the keyword tag, and determining a second preset keyword in the character information as the text keyword.

3. The damage settlement method of claim 2, wherein the generating keyword tag information corresponding to the keyword tag based on the corresponding tag position and text position of the keyword tag and the text keyword in the claims form image comprises:

detecting whether a label rectangular area corresponding to the keyword label is intersected with a text rectangular area corresponding to the text keyword according to the label position and the text position;

if the label rectangular area is intersected with the text rectangular area, calculating an intersection area of the label rectangular area and each intersected text rectangular area;

calculating a first area ratio of the intersection region on the label rectangular region, and calculating a second area ratio of the intersection region on the text rectangular region;

selecting a text rectangular area with the first area ratio and the second area ratio both larger than a preset threshold value as a target rectangular area, and generating keyword label information corresponding to the keyword label based on the number of the target rectangular areas and the text keyword corresponding to the target rectangular area.

4. The damage settlement method of claim 3, wherein the generating keyword tag information corresponding to the keyword tags based on the number of the target rectangular areas and the text keywords corresponding to the target rectangular areas comprises:

if the number of the target rectangular areas is single, determining the text keywords corresponding to the target rectangular areas as keyword label information corresponding to the keyword labels; or

And if the number of the target rectangular areas is multiple, performing weighting and operation on each first area ratio and the corresponding second area ratio to obtain a final score value of each target rectangular area, and selecting the text keyword corresponding to the target rectangular area with the largest final score value as keyword tag information corresponding to the keyword tag.

5. The damage settlement method of claim 1, wherein prior to inputting the initial label, the initial label information, the keyword label and the keyword label information to a preset feature coefficient model, the method further comprises:

acquiring a preset antagonistic neural network, and acquiring position information to which the damage assessment object belongs and training data corresponding to the position information;

selecting preset feature data from the training data based on preset feature keywords;

and training the preset antagonistic neural network based on the preset characteristic data to obtain the preset characteristic coefficient model.

6. The damage settlement method of claim 1, wherein the common coefficient model comprises a plurality of preset common damage levels and a common correspondence corresponding to each preset common damage level, the common coefficients comprise a common damage coefficient, a common base and a common month number, and the inputting the initial label, the initial label information, the keyword label and the keyword label information into the preset common coefficient model to calculate the common coefficient of the preset damage item comprises:

selecting a disability label from the initial label and the keyword label based on a preset disability keyword, and taking a numerical value corresponding to the disability label in the claim form image as an initial disability grade of the damage assessment object;

determining a general proportionality coefficient corresponding to a preset general disability grade in the general coefficient model, which is the same as the initial disability grade, as the general disability coefficient, and determining a general corresponding relation in the general coefficient model, which corresponds to the preset general disability grade, as a target corresponding relation;

identifying an initial cardinal number tag corresponding to a universal cardinal number tag in the target corresponding relationship and an initial month number tag corresponding to a universal month number tag in the target corresponding relationship from the initial tag and the keyword tag;

and calculating the universal base number based on the initial base number value of the initial base number label and a universal base number calculation mode in the target corresponding relation, and calculating the universal month number based on the initial month number value of the initial month number label and a universal month number calculation mode in the target corresponding relation.

7. The damage settlement method of claim 1, wherein the adjusting the generic coefficients based on the feature coefficients to obtain the target coefficients of the preset damage terms comprises:

detecting whether the characteristic coefficients correspond to each general coefficient;

if the characteristic coefficient does not correspond to each general coefficient, determining the general coefficient as the target coefficient; or

And if at least one characteristic coefficient corresponds to the general coefficient, replacing the general coefficient with the corresponding characteristic coefficient, and determining the replaced general coefficient as the target coefficient.

8. A damage settlement apparatus, comprising:

the system comprises an acquisition unit, a database unit and a database unit, wherein the acquisition unit is used for acquiring an initial label of a damage assessment object, initial label information corresponding to the initial label and a claim form image, and recognizing the claim form image by using a pre-trained character recognition model to obtain a keyword label and a text keyword of the claim form image;

the generating unit is used for generating keyword tag information corresponding to the keyword tag based on the corresponding tag position and text position of the keyword tag and the text keyword in the claim form image;

the input unit is used for inputting the initial label, the initial label information, the keyword label and the keyword label information into a preset general coefficient model and a preset characteristic coefficient model respectively to calculate a general coefficient of a preset loss assessment item and a characteristic coefficient of the preset loss assessment item;

the adjusting unit is used for adjusting the general coefficient based on the characteristic coefficient to obtain a target coefficient of the preset damage assessment item, and correcting the target coefficient based on a preset damage assessment standard library to obtain a damage assessment coefficient of the preset damage assessment item;

and the calculation unit is used for calculating the claim settlement data of the damage assessment object according to the preset damage assessment items and the damage assessment coefficient corresponding to each preset damage assessment item.

9. An electronic device, characterized in that the electronic device comprises:

a memory storing computer readable instructions; and

a processor executing computer readable instructions stored in the memory to implement the damage settlement method of any one of claims 1 to 7.

10. A computer-readable storage medium characterized by: the computer-readable storage medium has stored therein computer-readable instructions that are executed by a processor in an electronic device to implement the damage settlement method of any one of claims 1-7.