Insurance policy inspection method and system based on image recognition technology

Through image recognition technology, the policy data and information of similar user groups are extracted, and insurance vulnerability reports are generated, which solves the problems of time-consuming, labor-intensive and information processing pressure in traditional policy inspections, and realizes efficient, accurate and personalized analysis of policy inspections, ensuring insurance coverage and benefits .

CN119090657APending Publication Date: 2024-12-06YANDAO (NINGBO) DATA TECH CO LTD

Patent Information

Application Number
CN202411569947.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-06
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Traditional policy inspection methods rely on manual operations, which are time-consuming and labor-intensive, and are prone to omissions or errors. The policy formats and display methods of different insurance companies are different, making it difficult to quickly extract key information, resulting in high pressure on information processing, lack of data mining methods, and difficult to effectively analyze. Insurance loophole.

Method used

The policy inspection method based on image recognition technology is adopted. By obtaining the policy images and user portraits uploaded by users, the insurance data is extracted using image recognition technology, and similar user groups are screened based on the similarity algorithm to generate insurance vulnerabilities reports to identify unreasonable purchased and It is recommended to purchase insurance types.

Benefits of technology

It has achieved rapid and accurate extraction of key insurance data, reduced manual operation errors, ensured that insurance types were covered in comprehensively, avoided omissions, identified excessive or insufficient insurance through personalized analysis, allocated funds reasonably, and maximized insurance benefits.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090657A_ABST
    Figure CN119090657A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of data processing, and discloses an insurance policy inspection method and system based on an image recognition technology, and the method comprises the steps: obtaining M to-be-inspected insurance policy images uploaded by a target user, and obtaining a user portrait of the target user; identifying each to-be-inspected insurance policy image by using an image identification technology so as to extract first insurance data from the to-be-inspected insurance policy image; screening a similar user group matched with the user portrait of the target user based on a similarity algorithm, and obtaining corresponding second insurance data according to the insured information of the similar user group; generating an insurance vulnerability report of the user based on the first insurance data and the second insurance data; according to the invention, different users can be ensured to effectively cover various valuable insurance types for themselves, and the situation of excessive insurance or insufficient insurance of the users is avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically, to a policy review method and system based on image recognition technology. Background Art

[0002] With the development of social economy and the continuous expansion of the insurance market, more and more people have purchased different types of insurance to cope with future uncertainties; there are many types of insurance products, including life insurance, health insurance, property insurance, accident insurance, etc., and the number and coverage of insurance types launched by major insurance companies are also increasing; for individual users, with the increase in the number of insured persons, the management and review of purchased insurance policies have become more and more complicated; users may hold multiple policies from different insurance companies, which cover different types of insurance, insured amounts, and coverage, and the content is complicated; in order to ensure the comprehensiveness and rationality of insurance protection and avoid duplication or unnecessary policies, users need to review the policies they hold regularly.

[0003] Traditional policy review mainly relies on manual operation by users, who need to check the policy content, check the insurance information, and compare the coverage and the insured amount by themselves. However, this method has many problems: on the one hand, manual review is time-consuming and labor-intensive, and is prone to omissions or errors. On the other hand, as the number of user policies increases, it becomes difficult to manually review the policies one by one. In addition, different insurance companies use different policy formats and display methods, making it difficult for users to quickly extract key information from them. This puts users under great information processing pressure when reviewing policies, and manual review lacks the necessary data mining methods, making it difficult to effectively help users analyze insurance loopholes. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a policy review method and system based on image recognition technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A policy review method based on image recognition technology, the method comprising: Obtain M copies of insurance policy images uploaded by the target user to be reviewed, and obtain the user portrait of the target user, where M is an integer greater than zero; Using image recognition technology to identify each insurance policy image to be inspected, so as to extract first insurance data from the insurance policy image to be inspected, wherein the first insurance data includes a first insurance type set of multiple purchased insurance types and an insured amount of each purchased insurance type; Based on the similarity algorithm, a similar user group matching the user profile of the target user is screened out, and corresponding second insurance data is obtained according to the insurance information of the similar user group, wherein the insurance information includes multiple insured insurance types and the insurance amount of each insured insurance type; the second insurance data includes a second insurance type set of multiple recommended insurance types and the recommended insurance amount of each recommended insurance type; An insurance gap report for the user is generated based on the first insurance data and the second insurance data, wherein the insurance gap report includes unreasonable insurance types purchased and / or insurance types that are recommended to be purchased but not purchased.

[0006] Furthermore, before using image recognition technology to identify each insurance policy image to be inspected, the following steps are performed: A1: extracting the i-th insurance policy image to be inspected from the M insurance policy images to be inspected, and dividing the i-th insurance policy image to be inspected into equal parts to obtain N local areas in the form of images, where i is an integer greater than zero; A2: Calculate the local contrast of the vth local area, where v is an integer greater than zero; A3: Compare the local contrast of the vth local area with the preset contrast threshold. If the local contrast is less than the preset contrast threshold, mark the corresponding local area as a blurred local area, set v=v+1, and return to step A2; if the local contrast is greater than or equal to the preset contrast threshold, mark the corresponding local area as a clear local area, set v=v+1, and return to step A2; A4: Repeat the above steps A2 to A3 until v=N, and K fuzzy local areas are obtained; A5: extract the jth fuzzy local area from the K fuzzy local areas, where j is an integer greater than zero; A6: Use the pre-built contrast enhancement model to adjust the grayscale distribution of the j-th blurred local area, perform image enhancement on the j-th blurred local area according to the adjusted grayscale distribution, obtain the enhanced j-th blurred local area, set j=j+1, and return to step A5; A7: Repeat the above steps A5 to A6 until j=K, and obtain multiple enhanced fuzzy local areas. Merge the enhanced fuzzy local areas with the clear local areas to form a processed policy image to be inspected, set i=i+1, and return to step A1; A8: Repeat the above steps A1 to A7 until i=M, and obtain all processed insurance policy images to be inspected.

[0007] Furthermore, the expression of the pre-built contrast enhancement model is as follows: ; Where: is the gray value of the pixel at the coordinate (x, y) in the jth blurred local area after enhancement, is the gray value of the pixel at the coordinate (x, y) in the jth fuzzy local area, is the correction factor, Represents the average gray value of the jth blurred local area.

[0008] Furthermore, the method of using image recognition technology to identify each insurance policy image to be inspected includes: B1: Retrieve M insurance policy images to be reviewed, wherein the M insurance policy images to be reviewed include a plurality of insurance policy images to be reviewed before and after processing; B2: extracting the rth insurance policy image to be inspected from the M insurance policy images to be inspected, and performing optical character recognition on the rth insurance policy image to be inspected using OCR image recognition technology to obtain the content text of the rth insurance policy image to be inspected; B3: Perform word segmentation on the content text, remove meaningless words, and obtain a field set containing Q policy fields, where Q is an integer greater than zero; B4: Extract the dth policy field in the field set; B5: Input the dth policy field into the pre-built keyword library for traversal. If the traversal result shows a certain insurance type, mark the insurance type as the purchased insurance type, record the field position of the dth policy field in the field set, and search for the purchased insured amount starting from the field position of the dth policy field according to the pre-configured regular expression rule, and return to step B4; if the traversal result shows an empty set, set d=d+1, and return to step B4; B6: Repeat the above steps B4 to B5 until d=Q, obtain the purchased insurance type and the insured amount contained in the rth policy image to be reviewed, set r=r+1, and return to step B2; B7: Repeat the above steps B2 to B6 until r=M, and obtain a first insurance type set of multiple purchased insurance types and the insured amount of each purchased insurance type.

[0009] Furthermore, the screening out of similar user groups matching the user profile of the target user includes: Obtain the user profile of the target user, and extract multiple user groups and user profiles of each user group pre-stored in the system database; Calculate the similarity between the target user and the user portrait of each user group to obtain multiple similarities; The multiple similarities are sorted from large to small according to the numerical values, and the user group corresponding to the first similarity sort is marked as a similar user group.

[0010] Furthermore, the obtaining corresponding second insurance data according to the insurance information of the similar user group includes: Obtain the types of insurance policies that each user in a similar user group has purchased and the average amount of insurance for each type of insurance policy; Perform intersection statistics on the types of insurance policies that each user in a similar user group has purchased, so as to obtain the types of insurance policies that are common to similar user groups; Obtain the occurrence frequency of the common insured insurance types, mark the common insured insurance types whose occurrence frequency is greater than or equal to a preset occurrence frequency threshold as recommended insurance types, and obtain multiple recommended insurance types; Extract the insured amount of each insured type from all the common insured types, and calculate the mean of the insured amount of each insured type to obtain the average insured amount of each insured type; The average insurance amount is marked as the recommended insurance amount, and the recommended insurance amounts for all recommended insurance types are obtained.

[0011] Further, generating the insurance vulnerability report of the user based on the first insurance data and the second insurance data includes: Extracting a first insurance type set from the first insurance data, and extracting a second insurance type set from the second insurance data; Obtain the recommended insurance types that appear in the second insurance type set but not in the first insurance type set, and mark them as recommended but not purchased insurance types; Obtain purchased insurance types that do not appear in the second insurance type set but appear in the first insurance type set, and mark them as candidate insurance types; Calculate the matching degree between the candidate insurance types and the target users, and filter out unreasonable purchased insurance types from the candidate insurance types according to the matching degree; All unreasonable types of insurance purchased and / or types of insurance recommended but not purchased are counted, and an insurance vulnerability report for the user is generated based on them.

[0012] Furthermore, the step of screening out unreasonable purchased insurance types from among the candidate insurance types includes: Extract each candidate insurance type and the corresponding recommended insurance type that is consistent with each candidate insurance type, and extract the corresponding recommended insurance amount for the corresponding recommended insurance type; Get the insured amount of each candidate insurance type, and calculate the difference between the insured amount and the corresponding recommended insured amount; The corresponding candidate insurance types whose amount difference is less than the preset amount difference threshold are marked as reasonable purchased insurance types; The corresponding candidate insurance type whose amount difference is greater than or equal to the preset amount difference threshold is marked as an unreasonable purchased insurance type.

[0013] An insurance policy review system based on image recognition technology, comprising: A data acquisition module is used to acquire M copies of insurance policy images to be reviewed uploaded by a target user, and to acquire a user portrait of the target user, where M is an integer greater than zero; An information extraction module, configured to use image recognition technology to identify each insurance policy image to be inspected, so as to extract first insurance data from the insurance policy image to be inspected, wherein the first insurance data includes a first insurance type set of multiple purchased insurance types and an insured amount of each purchased insurance type; A data matching module is used to screen out similar user groups that match the user profile of the target user based on a similarity algorithm, and obtain corresponding second insurance data based on the insurance information of the similar user groups, wherein the insurance information includes multiple types of insured insurance products and the insurance amount of each type of insured insurance product; the second insurance data includes a second insurance type set of multiple recommended insurance types and the recommended insurance amount of each recommended insurance type; The insurance policy review module is used to generate an insurance vulnerability report for the user based on the first insurance data and the second insurance data, wherein the insurance vulnerability report includes unreasonable types of insurance purchased and / or types of insurance recommended to be purchased but not purchased.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present application discloses a method and system for reviewing insurance policies based on image recognition technology, including: obtaining M insurance policy images to be reviewed uploaded by a target user, and obtaining a user portrait of the target user; using image recognition technology to identify each insurance policy image to be reviewed, so as to extract first insurance data from the insurance policy image to be reviewed; screening out similar user groups that match the user portrait of the target user based on a similarity algorithm, and obtaining corresponding second insurance data based on the insurance information of the similar user groups; generating an insurance vulnerability report for the user based on the first insurance data and the second insurance data; based on the above technical features, the present invention can ensure that various valuable insurance types are effectively covered for different users and avoid omissions of important insurance types; at the same time, through personalized insurance analysis, it helps users identify over-insurance or under-insurance, ensures that funds are reasonably allocated and risks are fully covered, so as to achieve the maximum benefit of insurance protection; in addition, by introducing image recognition technology and automatically processing insurance policy images uploaded by users, the present invention can quickly and accurately extract key data in the insurance policy, greatly improving the efficiency of insurance policy review, and at the same time, reducing the risk of errors caused by manual operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flow chart of the insurance policy review method based on image recognition technology provided by the present invention; Figure 2 A schematic diagram of the module structure of the insurance policy review system based on image recognition technology provided by the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Example 1 See also Figure 1 As shown, this embodiment discloses a policy review method based on image recognition technology, the method comprising: S101: Obtain M copies of insurance policy images to be reviewed uploaded by a target user, and obtain a user portrait of the target user, where M is an integer greater than zero; It should be understood that the uploaded policy images to be reviewed support multiple image input methods, including but not limited to mobile phone photography, scanner scanning, etc., and support multiple formats, such as JPEG, PDF and PNG, etc. These policy images may come from multiple insurance companies and may involve different types of insurance, such as life insurance, health insurance, property insurance, etc.; Specifically, the user portrait includes demographic characteristics and health status characteristics; wherein the demographic characteristics include but are not limited to age, gender, occupation, income, marital status (such as married, single, divorced, etc.) and family members (such as whether there are children, the number of family members), etc., and the health status characteristics include but are not limited to medical history (whether there are records of chronic diseases, major diseases, such as heart disease, diabetes, etc.) and health habits (lifestyle habits, such as whether smoking, drinking, exercise frequency, etc.); S102: using image recognition technology to identify each insurance policy image to be inspected, so as to extract first insurance data from the insurance policy image to be inspected, wherein the first insurance data includes a first insurance type set of multiple purchased insurance types and an insured amount of each purchased insurance type; In implementation, before using image recognition technology to identify each insurance policy image to be inspected, the following steps are performed: A1: extracting the i-th insurance policy image to be inspected from the M insurance policy images to be inspected, and dividing the i-th insurance policy image to be inspected into equal parts to obtain N local areas in the form of images, where i is an integer greater than zero; A2: Calculate the local contrast of the vth local area, where v is an integer greater than zero; It should be understood that the local contrast is calculated based on the brightness range (maximum value - minimum value), and its specific calculation formula is: ; Where: is the maximum brightness value (grayscale value) in the local area, is the minimum brightness value in the local area, is the local contrast; A3: Compare the local contrast of the vth local area with the preset contrast threshold. If the local contrast is less than the preset contrast threshold, mark the corresponding local area as a blurred local area, set v=v+1, and return to step A2; if the local contrast is greater than or equal to the preset contrast threshold, mark the corresponding local area as a clear local area, set v=v+1, and return to step A2; A4: Repeat the above steps A2 to A3 until v=N, and K fuzzy local areas are obtained; A5: extract the jth fuzzy local area from the K fuzzy local areas, where j is an integer greater than zero; A6: Use the pre-built contrast enhancement model to adjust the grayscale distribution of the j-th blurred local area, perform image enhancement on the j-th blurred local area according to the adjusted grayscale distribution, obtain the enhanced j-th blurred local area, set j=j+1, and return to step A5; The expression of the pre-built contrast enhancement model is as follows: ; Where: is the gray value of the pixel at the coordinate (x, y) in the jth blurred local area after enhancement, is the gray value of the pixel at the coordinate (x, y) in the jth fuzzy local area, is the correction factor, Represents the average gray value of the jth fuzzy local area; A7: Repeat the above steps A5 to A6 until j=K, and obtain multiple enhanced fuzzy local areas. Merge the enhanced fuzzy local areas with the clear local areas to form a processed policy image to be inspected, set i=i+1, and return to step A1; A8: Repeat the above steps A1 to A7 until i=M, and obtain all processed insurance policy images to be inspected; In implementation, the image recognition technology is used to identify each insurance policy image to be inspected, including: B1: Retrieve M insurance policy images to be reviewed, wherein the M insurance policy images to be reviewed include a plurality of insurance policy images to be reviewed before and after processing; B2: Extract the r-th policy image to be inspected from the M policy images to be inspected, and perform optical character recognition on the r-th policy image to be inspected using OCR image recognition technology to obtain the content text of the r-th policy image to be inspected; B3: Perform word segmentation on the content text, remove meaningless words, and obtain a field set containing Q policy fields, where Q is an integer greater than zero; It should be noted that: the word segmentation is implemented based on existing word segmentation tools, such as the jieba word segmentation tool; the meaningless words include but are not limited to repeated words, stop words removed, onomatopoeic words and mimetic words, such as: "ah", "de" and "le", etc.; B4: Extract the d-th policy field from the field set; B5: Input the d-th policy field into the pre-constructed keyword library for traversal. If the traversal result shows a certain insurance type, mark the insurance type as the purchased insurance type, record the field position of the d-th policy field in the field set, and according to the pre-configured regular expression rules, start from the field position of the d-th policy field, search for the insured amount that has been purchased, and return to step B4; if the traversal result shows an empty set, let d = d + 1, and return to step B4; It should be noted that: the pre-constructed keyword library is set by technicians in advance, and it includes field styles of several insurance types, such as life insurance, health insurance, accident insurance, critical illness insurance, etc.; among them, the regular expression (RegularExpression, abbreviated as regex or regexp) is a powerful tool for matching text patterns, and the regular expression rules are implemented by technicians manually customizing regular expression patterns. For example, the corresponding regular expression for matching the insured amount is "¥?\d{1,3}(,\d{3})*(\.\d{2})? yuan?"; It can be understood that: when the purchased insurance types included in the r-th policy image to be inspected are searched through the keyword library, and the regular expression rules for the insured amount are matched according to the order of the policy fields of the purchased insurance types in the content text position, which is beneficial to shortening the extraction time of the first insurance data and improving the data extraction efficiency; B6: Repeat the above steps B4 - B5 until d = Q, obtain the purchased insurance types and the insured amounts included in the r-th policy image to be inspected, let r = r + 1, and return to step B2; B7: Repeat the above steps B2 - B6 until r = M, obtain a first insurance type set of multiple purchased insurance types and the insured amounts of each purchased insurance type; It is understandable that: by introducing image recognition technology and automatically processing the insurance policy images uploaded by users, key data in the insurance policy (such as insurance type and insured amount) can be extracted quickly and accurately; compared with the traditional manual review method, automated processing has greatly improved the efficiency of insurance policy review and reduced the time and energy invested by users in tedious operations; in addition, through the pre-processing of insurance policy images, including analysis and enhancement of local contrast, and OCR technology to identify the policy content, it ensures that even blurred or low-quality images can be accurately parsed; combined with the keyword library and regular expression rules, it can more accurately identify the insurance type and amount information in the policy, reducing the risk of errors caused by manual operations.

[0018] S103: Screening out similar user groups that match the user profile of the target user based on a similarity algorithm, and acquiring corresponding second insurance data based on the insurance information of the similar user groups, wherein the insurance information includes a plurality of insured insurance types and the insured amount of each insured insurance type; the second insurance data includes a second insurance type set of a plurality of recommended insurance types and the recommended insured amount of each recommended insurance type; In implementation, the screening out of similar user groups that match the user profile of the target user includes: Obtain the user profile of the target user, and extract multiple user groups and user profiles of each user group pre-stored in the system database; It should be noted that the multiple user groups pre-stored in the system database are obtained by technical personnel through manual statistical division based on the user portrait of each user; For example, suppose there are multiple users, among which the user portraits of a certain part of the users show that the age of these users is 20-24 years old, the gender is male, the occupation is teacher, the income is 10,000 yuan per month, the marital status is and the family members are children, wherein the number of family members is 3, the health status characteristics include but are not limited to the medical history of chronic diseases, specifically diabetes, and the health habits are smoking and drinking, therefore, this type of users is regarded as a user group; Calculate the similarity between the target user and the user portrait of each user group to obtain multiple similarities; It should be noted that: the similarity is calculated by cosine similarity. When calculating the similarity between the target user and the user portrait of a single user group, the similarity calculation is performed on each feature in the user portrait by cosine similarity, and then the sum of the similarities of each feature is accumulated and the mean is calculated to obtain the similarity between the target user and the user portrait of the single user group; For example, following the above example, assuming that the user profile of a target user is completely consistent with the user profile of a specific user group in the above example, it means that the similarity between the target user and the specific user group in terms of age, gender, occupation and other characteristics is 100%. Therefore, after accumulating the sum of the similarities of each characteristic and performing the mean calculation, the similarity between the user profile of the target user and the user group is still 100%; Sort multiple similarities by value from large to small, and mark the user group corresponding to the first similarity as a similar user group; In implementation, the step of obtaining corresponding second insurance data according to insurance information of similar user groups includes: Obtain the types of insurance policies that each user in a similar user group has purchased and the average amount of insurance for each type of insurance policy; Perform intersection statistics on the types of insurance policies that each user in a similar user group has purchased, so as to obtain the types of insurance policies that are common to similar user groups; Obtain the occurrence frequency of the common insured insurance types, mark the common insured insurance types whose occurrence frequency is greater than or equal to a preset occurrence frequency threshold as recommended insurance types, and obtain multiple recommended insurance types; For example, suppose there are three users in the similar user group, namely Z1, Z2 and Z3, among which Z1 purchased life insurance, health insurance and accident insurance, Z2 purchased life insurance, health insurance and property insurance, and Z3 purchased life insurance, accident insurance and property insurance. The occurrence frequency of life insurance is 3, and the occurrence frequencies of health insurance, accident insurance and property insurance are 2. If the preset occurrence frequency threshold is 3, only life insurance is used as the recommended insurance type; Extract the insured amount of each insured type from all the common insured types, and calculate the mean of the insured amount of each insured type to obtain the average insured amount of each insured type; The average insurance amount is marked as the recommended insurance amount, and the recommended insurance amounts for all recommended insurance types are obtained.

[0019] S104: Generate an insurance vulnerability report for the user based on the first insurance data and the second insurance data, wherein the insurance vulnerability report includes unreasonable types of insurance purchased and / or types of insurance recommended for purchase but not purchased; In implementation, generating the insurance vulnerability report of the user based on the first insurance data and the second insurance data includes: Extracting a first insurance type set from the first insurance data, and extracting a second insurance type set from the second insurance data; Obtain the recommended insurance types that appear in the second insurance type set but not in the first insurance type set, and mark them as recommended but not purchased insurance types; Obtain purchased insurance types that do not appear in the second insurance type set but appear in the first insurance type set, and mark them as candidate insurance types; Calculate the matching degree between the candidate insurance types and the target users, and filter out unreasonable purchased insurance types from the candidate insurance types based on the matching degree; Specifically, the screening out of unreasonable purchased insurance types from candidate insurance types includes: Extract each candidate insurance type and the corresponding recommended insurance type that is consistent with each candidate insurance type, and extract the corresponding recommended insurance amount for the corresponding recommended insurance type; Get the insured amount of each candidate insurance type, and calculate the difference between the insured amount and the corresponding recommended insured amount; The corresponding candidate insurance types whose amount difference is less than the preset amount difference threshold are marked as reasonable purchased insurance types; The corresponding candidate insurance types whose amount difference is greater than or equal to the preset amount difference threshold are marked as unreasonable purchased insurance types; For example, following the above example, it is assumed that the target user's recommended insurance type is life insurance, and the recommended insured amount is 20,000, and the target user's purchased insurance types also include life insurance, but the actual insured amount is 30,000. Therefore, the difference between the insured amount and the corresponding recommended insured amount is 10,000. If the preset amount difference threshold is 5,000, it means that the corresponding life insurance is an unreasonable purchased insurance type. This means that the target user's investment in life insurance exceeds his or her own economic level, or the target user's investment in life insurance has not achieved the ideal expected return. Therefore, it is marked as an unreasonable purchased insurance type, prompting the target user to make up for the insurance loophole to maximize the rationality of the target user's insurance purchase; Collect statistics on all unreasonable insurance types purchased and / or insurance types recommended but not purchased, and generate insurance vulnerability reports for users based on them; Through image recognition technology, key information in user insurance policies is automatically extracted, and combined with user portraits and insurance data of similar user groups, potential insurance loopholes of users are deeply explored and analyzed. For different users, the present invention can ensure that various valuable insurance types are effectively covered and avoid omissions of important insurance types. At the same time, through personalized insurance analysis, users are helped to identify over-insurance or under-insurance, ensure that funds are reasonably allocated and risks are fully covered, and ultimately optimize the user's insurance portfolio to maximize the benefits of insurance protection.

[0020] Example 2 See also Figure 2 As shown, based on the same inventive concept, this embodiment discloses a policy review system based on image recognition technology. For details not provided in this embodiment, please refer to the description of the relevant parts in Embodiment 1. The system includes: The data acquisition module 210 is used to acquire M copies of insurance policy images to be reviewed uploaded by the target user, and to acquire the user portrait of the target user, where M is an integer greater than zero; The information extraction module 220 is used to identify each insurance policy image to be inspected by using image recognition technology to extract first insurance data from the insurance policy image to be inspected, wherein the first insurance data includes a first insurance type set of multiple purchased insurance types and an insured amount of each purchased insurance type; The data matching module 230 is used to screen out similar user groups that match the user profile of the target user based on a similarity algorithm, and obtain corresponding second insurance data according to the insurance information of the similar user groups, wherein the insurance information includes multiple insured insurance types and the insurance amount of each insured insurance type; the second insurance data includes a second insurance type set of multiple recommended insurance types and the recommended insurance amount of each recommended insurance type; The insurance policy review module 240 is used to generate an insurance vulnerability report for the user based on the first insurance data and the second insurance data, wherein the insurance vulnerability report includes unreasonable insurance types purchased and / or insurance types recommended to be purchased but not purchased.

[0021] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to actual conditions.

[0022] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired network or a wireless network. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD) or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0023] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0024] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0025] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0026] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0027] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0028] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0029] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A policy review method based on image recognition technology. It is characterized in that The method comprises: Obtain M copies of insurance policy images uploaded by the target user to be reviewed, and obtain the user portrait of the target user, where M is an integer greater than zero; Using image recognition technology to identify each insurance policy image to be inspected, so as to extract first insurance data from the insurance policy image to be inspected, wherein the first insurance data includes a first insurance type set of multiple purchased insurance types and an insured amount of each purchased insurance type; Based on the similarity algorithm, a similar user group matching the user profile of the target user is screened out, and corresponding second insurance data is obtained according to the insurance information of the similar user group, wherein the insurance information includes multiple insured insurance types and the insurance amount of each insured insurance type; the second insurance data includes a second insurance type set of multiple recommended insurance types and the recommended insurance amount of each recommended insurance type; An insurance gap report for the user is generated based on the first insurance data and the second insurance data, wherein the insurance gap report includes unreasonable insurance types purchased and / or insurance types that are recommended to be purchased but not purchased.

2. The insurance policy review method based on image recognition technology according to claim 1, It is characterized in that Before using image recognition technology to identify each policy image to be reviewed, including: A1: extracting the i-th insurance policy image to be inspected from the M insurance policy images to be inspected, and dividing the i-th insurance policy image to be inspected into equal parts to obtain N local areas in the form of images, where i is an integer greater than zero; A2: Calculate the local contrast of the vth local area, where v is an integer greater than zero; A3: Compare the local contrast of the vth local area with the preset contrast threshold. If the local contrast is less than the preset contrast threshold, mark the corresponding local area as a blurred local area, set v=v+1, and return to step A2; if the local contrast is greater than or equal to the preset contrast threshold, mark the corresponding local area as a clear local area, set v=v+1, and return to step A2; A4: Repeat the above steps A2 to A3 until v=N, and K fuzzy local areas are obtained; A5: extract the jth fuzzy local area from the K fuzzy local areas, where j is an integer greater than zero; A6: Use the pre-built contrast enhancement model to adjust the grayscale distribution of the j-th blurred local area, perform image enhancement on the j-th blurred local area according to the adjusted grayscale distribution, obtain the enhanced j-th blurred local area, set j=j+1, and return to step A5; A7: Repeat the above steps A5 to A6 until j=K, and obtain multiple enhanced fuzzy local areas. Merge the enhanced fuzzy local areas with the clear local areas to form a processed policy image to be inspected, set i=i+1, and return to step A1; A8: Repeat the above steps A1 to A7 until i=M, and obtain all processed insurance policy images to be inspected.

3. The insurance policy review method based on image recognition technology according to claim 2, It is characterized in that The expression of the pre-built contrast enhancement model is as follows: ; Where: is the gray value of the pixel at the coordinate (x, y) in the jth blurred local area after enhancement, is the gray value of the pixel at the coordinate (x, y) in the jth fuzzy local area, is the correction factor, Represents the average gray value of the jth blurred local area.

4. The insurance policy review method based on image recognition technology according to claim 3, It is characterized in that The method of using image recognition technology to identify each insurance policy image to be inspected includes: B1: Retrieve M insurance policy images to be reviewed, wherein the M insurance policy images to be reviewed include a plurality of insurance policy images to be reviewed before and after processing; B2: extracting the rth insurance policy image to be inspected from the M insurance policy images to be inspected, and performing optical character recognition on the rth insurance policy image to be inspected using OCR image recognition technology to obtain the content text of the rth insurance policy image to be inspected; B3: Perform word segmentation on the content text, remove meaningless words, and obtain a field set containing Q policy fields, where Q is an integer greater than zero; B4: Extract the dth policy field in the field set; B5: Input the dth policy field into the pre-built keyword library for traversal. If the traversal result shows a certain insurance type, mark the insurance type as the purchased insurance type, record the field position of the dth policy field in the field set, and search for the purchased insured amount starting from the field position of the dth policy field according to the pre-configured regular expression rule, and return to step B4; if the traversal result shows an empty set, set d=d+1, and return to step B4; B6: Repeat the above steps B4 to B5 until d=Q, obtain the purchased insurance type and the insured amount contained in the rth policy image to be reviewed, set r=r+1, and return to step B2; B7: Repeat the above steps B2 to B6 until r=M, and obtain a first insurance type set of multiple purchased insurance types and the insured amount of each purchased insurance type.

5. The insurance policy review method based on image recognition technology according to claim 4, It is characterized in that The screening out of similar user groups that match the user profile of the target user includes: Obtain the user profile of the target user, and extract multiple user groups and user profiles of each user group pre-stored in the system database; Calculate the similarity between the target user and the user portrait of each user group to obtain multiple similarities; The multiple similarities are sorted from large to small according to the numerical values, and the user group corresponding to the first similarity sort is marked as a similar user group.

6. The insurance policy review method based on image recognition technology according to claim 5, It is characterized in that The obtaining corresponding second insurance data according to the insurance information of the similar user group includes: Obtain the types of insurance policies that each user in a similar user group has purchased and the average amount of insurance for each type of insurance policy; Perform intersection statistics on the types of insurance policies that each user in a similar user group has purchased, so as to obtain the types of insurance policies that are common to similar user groups; Obtain the occurrence frequency of the common insured insurance types, mark the common insured insurance types whose occurrence frequency is greater than or equal to a preset occurrence frequency threshold as recommended insurance types, and obtain multiple recommended insurance types; Extract the insured amount of each insured type from all the common insured types, and calculate the mean of the insured amount of each insured type to obtain the average insured amount of each insured type; The average insurance amount is marked as the recommended insurance amount, and the recommended insurance amounts for all recommended insurance types are obtained.

7. The insurance policy review method based on image recognition technology according to claim 6, It is characterized in that Generating the insurance vulnerability report of the user based on the first insurance data and the second insurance data includes: Extracting a first insurance type set from the first insurance data, and extracting a second insurance type set from the second insurance data; Obtain the recommended insurance types that appear in the second insurance type set but not in the first insurance type set, and mark them as recommended but not purchased insurance types; Obtain purchased insurance types that do not appear in the second insurance type set but appear in the first insurance type set, and mark them as candidate insurance types; Calculate the matching degree between the candidate insurance types and the target users, and filter out unreasonable purchased insurance types from the candidate insurance types according to the matching degree; All unreasonable types of insurance purchased and / or types of insurance recommended but not purchased are counted, and an insurance vulnerability report for the user is generated based on them.

8. The insurance policy review method based on image recognition technology according to claim 7, It is characterized in that The method of screening out unreasonable purchased insurance types from among the candidate insurance types includes: Extract each candidate insurance type and the corresponding recommended insurance type that is consistent with each candidate insurance type, and extract the corresponding recommended insurance amount for the corresponding recommended insurance type; Get the insured amount of each candidate insurance type, and calculate the difference between the insured amount and the corresponding recommended insured amount; The corresponding candidate insurance types whose amount difference is less than the preset amount difference threshold are marked as reasonable purchased insurance types; The corresponding candidate insurance type whose amount difference is greater than or equal to the preset amount difference threshold is marked as an unreasonable purchased insurance type.

9. An insurance policy review system based on image recognition technology. It is characterized in that include: A data acquisition module is used to acquire M copies of insurance policy images to be reviewed uploaded by a target user, and to acquire a user portrait of the target user, where M is an integer greater than zero; An information extraction module, configured to use image recognition technology to identify each insurance policy image to be inspected, so as to extract first insurance data from the insurance policy image to be inspected, wherein the first insurance data includes a first insurance type set of multiple purchased insurance types and an insured amount of each purchased insurance type; A data matching module is used to screen out similar user groups that match the user profile of the target user based on a similarity algorithm, and obtain corresponding second insurance data based on the insurance information of the similar user groups, wherein the insurance information includes multiple types of insured insurance products and the insurance amount of each type of insured insurance product; the second insurance data includes a second insurance type set of multiple recommended insurance types and the recommended insurance amount of each recommended insurance type; The insurance policy review module is used to generate an insurance vulnerability report for the user based on the first insurance data and the second insurance data, wherein the insurance vulnerability report includes unreasonable types of insurance purchased and / or types of insurance recommended to be purchased but not purchased.

Citation Information

Patent Citations

  • Big data insurance recommendation method and system

    CN113822743A

  • Insurance data processing method and device, equipment and storage medium

    CN114429387A

  • Accurate insurance product recommendation method based on hybrid recommendation algorithm

    CN115409630A

  • Health asset insurance policy recommendation method based on big data analysis

    CN118505402A

  • Method for providing insurance product information and apparatus therefor

    WO2022092524A1

Cited By

  • Firmware homologous vulnerability mining method and system based on multi-dimensional feature portrait

    CN120408648A

  • Method and system for analyzing insurance policy risks based on association rule mining and storage medium

    CN122736788A