Insurance risk scoring management system and method based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent management, and particularly discloses an insurance risk scoring management system and method based on artificial intelligence, and the system employs an artificial intelligence technology based on deep learning to carry out the semantic analysis of a data set of insurance data marked as high risk. And extracting the core semantic feature representation of the high-risk insurance data set, thereby intelligently judging whether the to-be-evaluated insurance data is high-risk based on the semantic feature difference between the core semantic feature of the high-risk insurance data and the to-be-evaluated insurance data. Therefore, the accuracy and efficiency of insurance risk assessment are improved, the interference of human factors can be reduced, and the risk caused by subjective judgment is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent management technology, and more specifically, to an insurance risk scoring management system and method based on artificial intelligence. Background Art

[0002] In the insurance industry, insurance risk assessment is the basis for insurance companies to conduct risk management, which is directly related to the profitability and business stability of insurance companies. Insurance companies determine the risk level of customers' insurance applications by evaluating their basic information and insurance needs, and then classify and screen customers, formulate more reasonable premium pricing strategies for customers, and ensure that premiums can cover expected risk losses, thereby maintaining the company's financial stability. In addition, insurance fraud is an important issue facing the insurance industry. By conducting risk assessments on insurance applications, insurance companies can identify and prevent potential fraud, reduce fraud risks, and protect the interests of the company and its customers.

[0003] Traditional insurance risk assessment methods mainly rely on manual analysis, using the experience and expertise of insurance brokers or risk assessors to assess the risk level of insurance applications. However, this method has many limitations, such as low efficiency, strong subjectivity, and difficulty in processing large-scale data. With the rapid development of the insurance industry, the variety of insurance products has become increasingly rich, and the complexity of insurance business has also increased. Traditional insurance risk assessment methods have been unable to meet the needs of modern insurance business. Therefore, an insurance risk scoring management system and method based on artificial intelligence is expected. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an insurance risk scoring management system and method based on artificial intelligence, which uses artificial intelligence technology based on deep learning to perform semantic analysis on a data set of insurance data marked as high-risk, and extracts the core semantic feature representation of the high-risk insurance data set, thereby intelligently judging whether the insurance data to be evaluated is high-risk based on the difference in semantic features between the core semantic features of the high-risk insurance data and the insurance data to be evaluated. In this way, the accuracy and efficiency of insurance risk assessment are improved, and the interference of human factors can be reduced, reducing the risks caused by subjective judgment.

[0005] Accordingly, according to one aspect of the present application, an insurance risk scoring management system based on artificial intelligence is provided, which includes:

[0006] An insurance data acquisition module, used to acquire the insurance content and customer information in the insurance to be evaluated, and extract a data set of insurance data marked as high risk from the database, wherein the insurance data includes claim records, customer information and insurance content;

[0007] A high-risk insurance data semantic coding module, used for semantically coding each insurance data marked as high-risk in the data set of insurance data marked as high-risk to obtain a set of high-risk insurance semantic coding vectors;

[0008] A high-risk insurance data semantic enhancement fusion module, used to perform semantic enhancement fusion on the set of high-risk insurance semantic encoding vectors to obtain a high-risk insurance semantic significant aggregation representation vector;

[0009] A semantic coding module for insurance data to be evaluated, used for semantically coding the insurance content and customer information in the insurance to be evaluated to obtain a semantic coding vector for the insurance to be evaluated;

[0010] The high-risk judgment module is used to determine whether the insurance to be evaluated is high-risk based on the semantic difference between the high-risk insurance semantic significant aggregation representation vector and the semantic encoding vector of the insurance to be evaluated.

[0011] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the high-risk insurance data semantic encoding module is used to: pass each insurance data marked as high-risk in the data set of insurance data marked as high-risk through an insurance data semantic encoder based on the Bert model to obtain a set of high-risk insurance semantic encoding vectors.

[0012] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the high-risk insurance data semantic enhancement fusion module is used to: input the set of high-risk insurance semantic encoding vectors into the significance enhancement fusion module based on the sequence distribution cluster center to obtain the high-risk insurance semantic significance aggregation representation vector.

[0013] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the high-risk insurance data semantic enhancement fusion module includes: a sequence distribution cluster center calculation unit, which is used to calculate the mean vector of the set of high-risk insurance semantic coding vectors as the sequence distribution cluster center; a semantic relevance calculation unit, which is used to calculate the semantic relevance between each high-risk insurance semantic coding vector in the set of high-risk insurance semantic coding vectors and the sequence distribution cluster center to obtain a sequence of semantic relevance; a feature significance enhancement fusion unit, which is used to perform weighted fusion on the set of high-risk insurance semantic coding vectors based on the sequence of semantic relevance to obtain the high-risk insurance semantic saliency aggregation representation vector.

[0014] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the semantic relevance calculation unit is used to: respectively calculate the matrix product between each high-risk insurance semantic coding vector in the set of high-risk insurance semantic coding vectors and each predetermined weight matrix to obtain a set of high-risk insurance semantic representation coding feature vectors; add a bias vector to each high-risk insurance semantic representation coding feature vector in the set of high-risk insurance semantic representation coding feature vectors and then pass the resultant sigmoid activation function to obtain a set of activated high-risk insurance semantic representation coding feature vectors; respectively calculate the vector product between each activated high-risk insurance semantic representation coding feature vector in the set of activated high-risk insurance semantic representation coding feature vectors and the transposed vector of the sequence distribution cluster center to obtain the sequence of semantic relevance.

[0015] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the feature significance enhancement fusion unit is used to: input the sequence of semantic relevance into the softmax function for soft maximum normalization processing to obtain a sequence of sequence-related significance weight factors; using the sequence of sequence-related significance weight factors as the weight sequence, calculate the vector-by-vector weighted sum of the set of high-risk insurance semantic encoding vectors to obtain the high-risk insurance semantic significance aggregation representation vector.

[0016] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the semantic encoding module of the insurance data to be evaluated is used to: use the insurance data semantic encoder based on the Bert model to semantically encode the insurance content and customer information in the insurance to be evaluated to obtain the semantic encoding vector of the insurance to be evaluated.

[0017] In the above-mentioned artificial intelligence-based insurance risk scoring management system, the high-risk judgment module includes: a feature difference measurement unit, used to calculate the Hamming distance between the high-risk insurance semantic significant aggregation representation vector and the insurance semantic encoding vector to be evaluated; an evaluation result determination unit, used to determine whether the insurance to be evaluated is high-risk based on the comparison between the Hamming distance and a preset threshold.

[0018] According to another aspect of the present application, there is provided an insurance risk scoring management method based on artificial intelligence, which includes:

[0019] Obtaining the insurance content and customer information in the insurance to be evaluated, and extracting a data set of insurance data marked as high risk from a database, wherein the insurance data includes claim records, customer information and insurance content;

[0020] Performing semantic coding on each insurance data marked as high risk in the data set of insurance data marked as high risk to obtain a set of high-risk insurance semantic coding vectors;

[0021] Performing semantic enhancement fusion on the set of high-risk insurance semantic encoding vectors to obtain a high-risk insurance semantic significant aggregation representation vector;

[0022] Semantically encoding the insurance content and customer information in the insurance to be evaluated to obtain a semantic encoding vector of the insurance to be evaluated;

[0023] Based on the semantic difference between the high-risk insurance semantically significant aggregation representation vector and the semantic encoding vector of the insurance to be evaluated, it is determined whether the insurance to be evaluated is high-risk.

[0024] Compared with the prior art, the AI-based insurance risk scoring management system and method provided by the present application uses AI technology based on deep learning to perform semantic analysis on a data set of insurance data marked as high-risk, extracting the core semantic feature representation of the high-risk insurance data set, thereby intelligently judging whether the insurance data to be evaluated is high-risk based on the difference in semantic features between the core semantic features of the high-risk insurance data and the insurance data to be evaluated. In this way, the accuracy and efficiency of insurance risk assessment are improved, and the interference of human factors can be reduced, reducing the risks caused by subjective judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 The block diagram of the insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application.

[0027] Figure 2 Schematic diagram of the architecture of an insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application.

[0028] Figure 3 The present invention is a block diagram of a high-risk insurance data semantic enhancement fusion module in an insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application.

[0029] Figure 4 It is a block diagram of a high-risk assessment module in an insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application.

[0030] Figure 5 The present invention is a flowchart of an insurance risk scoring management method based on artificial intelligence according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above and other purposes, features and advantages of the present application will become more apparent. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0032] As mentioned in the above background technology, the traditional insurance risk assessment method is mainly based on manual analysis, relying on the rich experience and deep professional knowledge of insurance brokers or risk assessors to judge the risk level of insurance applications. However, this method has many disadvantages, such as low efficiency, a large proportion of subjective judgment, and difficulty in coping with challenges such as large-scale data processing. With the rapid development of the insurance industry and the continuous enrichment of insurance product types, the complexity of insurance business has become increasingly prominent, and the traditional insurance risk assessment method has gradually become difficult to adapt to the development needs of modern insurance business. In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to perform semantic analysis on a data set of insurance data marked as high-risk, and extract the core semantic feature representation of the high-risk insurance data set, so as to intelligently judge whether the insurance data to be evaluated is high-risk based on the core semantic features of the high-risk insurance data and the semantic feature differences between the insurance data to be evaluated. In this way, the accuracy and efficiency of insurance risk assessment are improved, and the interference of human factors can be reduced, and the risks caused by subjective judgment can be reduced.

[0033] Figure 1 The block diagram of the insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of an insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the artificial intelligence-based insurance risk scoring management system 100 includes: an insurance data acquisition module 110, which is used to acquire the insurance content and customer information in the insurance to be evaluated, and extract a data set of insurance data marked as high-risk from the database, wherein the insurance data includes claim records, customer information and insurance content; a high-risk insurance data semantic coding module 120, which is used to semantically encode each insurance data marked as high-risk in the data set of insurance data marked as high-risk to obtain a set of high-risk insurance semantic coding vectors; a high-risk insurance data semantic enhancement fusion module 130, which is used to semantically enhance and fuse the set of high-risk insurance semantic coding vectors to obtain a high-risk insurance semantic significant aggregation representation vector; a semantic coding module 140 for insurance data to be evaluated, which is used to semantically encode the insurance content and customer information in the insurance to be evaluated to obtain a semantic coding vector for insurance to be evaluated; a high-risk judgment module 150, which is used to determine whether the insurance to be evaluated is high-risk based on the semantic difference between the high-risk insurance semantic significant aggregation representation vector and the semantic coding vector of the insurance to be evaluated.

[0034] In the above-mentioned artificial intelligence-based insurance risk scoring management system 100, the insurance data acquisition module 110 is used to obtain the insurance content and customer information in the insurance to be evaluated, and extract the data set of insurance data marked as high-risk from the database, wherein the insurance data includes claim records, customer information and insurance content. It should be understood that insurance content and customer information are the basis for assessing insurance risks. The insurance content determines the scope, conditions and restrictions of insurance coverage, while customer information provides background information about the insured or beneficiary, such as age, occupation, health status, etc., which is an important factor in assessing risks. The data set of insurance data marked as high-risk contains detailed information on high-risk cases that occurred in the past. By analyzing these historical high-risk insurance data, the common characteristics and patterns of high-risk cases can be learned to better predict and assess the risk level of new insurance applications.

[0035] In the above-mentioned insurance risk scoring management system 100 based on artificial intelligence, the high-risk insurance data semantic encoding module 120 is used to semantically encode each insurance data marked as high risk in the data set of insurance data marked as high risk to obtain a set of high-risk insurance semantic encoding vectors. In a specific example of the present application, each insurance data marked as high risk in the data set of insurance data marked as high risk is passed through an insurance data semantic encoder based on the Bert model to obtain a set of high-risk insurance semantic encoding vectors. It should be understood that traditional text processing methods are often based on keyword matching or simple statistical information, and cannot fully understand the deep meaning and contextual relationship in the text. Therefore, in order to achieve an in-depth understanding of insurance data, in the technical solution of the present application, an insurance data semantic encoder based on the Bert model is used to semantically encode each insurance data marked as high risk. Those of ordinary skill in the art should know that the Bert model is a bidirectional encoder representation based on Transformer, which can deeply understand the contextual semantics of the text, thereby retaining more text information. Here, semantic encoding of insurance data through the Bert model can effectively mine the deep meaning of insurance data, including the meaning of words, the structure of sentences, and the theme of paragraphs, and generate semantic coding vectors corresponding to each high-risk insurance data, thereby providing a valuable information basis for subsequent risk assessment and improving the accuracy of risk assessment.

[0036] In the above-mentioned artificial intelligence-based insurance risk scoring management system 100, the high-risk insurance data semantic enhancement fusion module 130 is used to perform semantic enhancement fusion on the set of high-risk insurance semantic coding vectors to obtain a high-risk insurance semantically significant aggregation representation vector. It should be understood that, considering that there may be semantic redundancy or differences between the various insurance data in the data set of insurance data marked as high-risk, that is, it contains some irrelevant or repeated semantic information, which in turn interferes with risk assessment. Therefore, in order to more accurately express the core semantic features of high-risk insurance data, in the technical solution of the present application, the set of high-risk insurance semantic coding vectors is further subjected to semantic enhancement aggregation processing.

[0037] In a specific example of the present application, the encoding method for semantically reinforcing and fusing the set of high-risk insurance semantic coding vectors is to input the set of high-risk insurance semantic coding vectors into a saliency reinforcing fusion module based on the sequence distribution cluster center to obtain the high-risk insurance semantic saliency aggregation representation vector. Specifically, the saliency reinforcing fusion module based on the sequence distribution cluster center uses the mean vector of the set of high-risk insurance semantic coding vectors as the sequence distribution cluster center to reflect the overall semantic characteristics of the high-risk insurance data set, and determines the importance of each high-risk insurance semantic coding vector in the overall semantic space based on the semantic difference between each high-risk insurance semantic coding vector and the sequence distribution cluster center, so as to perform weighted aggregation on the set of high-risk insurance semantic coding vectors to emphasize the semantic features that are closer to the overall semantics of the high-risk insurance data set, while suppressing redundant or irrelevant semantic information, thereby effectively removing interference information, and mining the core semantic features of high-risk insurance data, providing more accurate and reliable information support for subsequent insurance risk assessment.

[0038] Figure 3 FIG. 1 is a block diagram of a high-risk insurance data semantic enhancement fusion module in an insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application. Figure 3 As shown, the high-risk insurance data semantic enhancement fusion module 130 includes: a sequence distribution cluster center calculation unit 131, used to calculate the mean vector of the set of high-risk insurance semantic coding vectors as the sequence distribution cluster center; a semantic relevance calculation unit 132, used to calculate the semantic relevance between each high-risk insurance semantic coding vector in the set of high-risk insurance semantic coding vectors and the sequence distribution cluster center to obtain a sequence of semantic relevance; a feature significance enhancement fusion unit 133, used to perform weighted fusion on the set of high-risk insurance semantic coding vectors based on the sequence of semantic relevance to obtain the high-risk insurance semantic saliency aggregation representation vector.

[0039] More specifically, the semantic relevance calculation unit 132 is used to: respectively calculate the matrix product between each high-risk insurance semantic coding vector in the set of high-risk insurance semantic coding vectors and each predetermined weight matrix to obtain a set of high-risk insurance semantic representation coding feature vectors; add a bias vector to each high-risk insurance semantic representation coding feature vector in the set of high-risk insurance semantic representation coding feature vectors and then pass the sigmoid activation function to obtain a set of activated high-risk insurance semantic representation coding feature vectors; respectively calculate the vector product between each activated high-risk insurance semantic representation coding feature vector in the set of activated high-risk insurance semantic representation coding feature vectors and the transposed vector of the sequence distribution cluster center to obtain the sequence of semantic relevance.

[0040] More specifically, the feature significance enhancement fusion unit 133 is used to: input the sequence of semantic relevance into a softmax function for soft maximum normalization processing to obtain a sequence of sequence-related significance weight factors; using the sequence of sequence-related significance weight factors as a weight sequence, calculate the vector-by-vector weighted sum of the set of high-risk insurance semantic encoding vectors to obtain the high-risk insurance semantic significance aggregation representation vector.

[0041] That is, the high-risk insurance data semantic enhancement fusion module 130 is used to process the set of high-risk insurance semantic encoding vectors using the following saliency enhancement fusion formula to obtain the high-risk insurance semantic saliency aggregation representation vector, wherein the saliency enhancement fusion formula is:

[0042]

[0043] Among them, v i is the ith high-risk insurance semantic coding vector in the set of high-risk insurance semantic coding vectors, N is the number of high-risk insurance semantic coding vectors, v r is the cluster center of sequence distribution, W i is the predetermined weight matrix, B i is the bias vector, sigmoid is the activation function, e i is the ith semantic relevance, softmax is the normalized exponential function, * represents dot product, and V is the semantically significant aggregation representation vector of the high-risk insurance.

[0044] In the above-mentioned insurance risk scoring management system 100 based on artificial intelligence, the semantic coding module 140 of the insurance data to be evaluated is used to semantically encode the insurance content and customer information in the insurance to be evaluated to obtain the semantic coding vector of the insurance to be evaluated. In a specific example of the present application, the insurance data semantic encoder based on the Bert model is used to semantically encode the insurance content and customer information in the insurance to be evaluated to obtain the semantic coding vector of the insurance to be evaluated. That is, similar to the processing method of the aforementioned high-risk insurance data, for the insurance data to be evaluated, the insurance data semantic encoder based on the Bert model is also used to semantically encode it, so as to fully understand and extract the contextual semantic information in the insurance to be evaluated, and convert it into a computer-processable semantic coding vector form, so as to achieve the unification of data form, thereby facilitating semantic analysis and comparison with known high-risk cases. In this way, the same semantic coding model is used to process the insurance to be evaluated and the known high-risk insurance data, which can reduce the errors introduced by different processing methods and ensure the consistency and accuracy of the evaluation process.

[0045] In the above-mentioned insurance risk scoring management system 100 based on artificial intelligence, the high-risk evaluation module 150 is used to determine whether the insurance to be evaluated is high-risk based on the semantic difference between the high-risk insurance semantic significant aggregation representation vector and the semantic encoding vector of the insurance to be evaluated. Figure 4 FIG. 1 is a block diagram of a high risk assessment module in an insurance risk scoring management system based on artificial intelligence according to an embodiment of the present application. Figure 4 As shown, the high-risk assessment module 150 includes: a feature difference measurement unit 151, used to calculate the Hamming distance between the semantically significant aggregation representation vector of the high-risk insurance and the semantic encoding vector of the insurance to be evaluated; an evaluation result determination unit 152, used to determine whether the insurance to be evaluated is high-risk based on the comparison between the Hamming distance and a preset threshold.

[0046] Specifically, the high-risk assessment module 150 includes: a feature difference measurement unit 151, which is used to calculate the Hamming distance between the high-risk insurance semantic significant aggregation representation vector and the insurance semantic encoding vector to be evaluated. That is, the semantic feature difference between the high-risk insurance data and the insurance data to be evaluated is measured by calculating the Hamming distance. A person of ordinary skill in the art should know that the Hamming distance is an effective indicator for measuring the difference between two equal-length character strings. In information theory, the Hamming distance is used to describe the number of different characters at corresponding positions between two equal-length character strings, so as to reflect the similarity or difference between the two character strings. Therefore, in the technical solution of the present application, by calculating the Hamming distance between the high-risk insurance semantic significant aggregation representation vector and the insurance semantic encoding vector to be evaluated, the degree of proximity between the insurance data to be evaluated and the known high-risk cases in the semantic space can be intuitively and quantitatively evaluated, thereby providing a reference basis for subsequent risk judgment.

[0047] In a preferred example of the present application, calculating the Hamming distance between the high-risk insurance semantic significant aggregation representation vector and the insurance semantic encoding vector to be evaluated includes the following steps: cascading the high-risk insurance semantic significant aggregation representation vector and the insurance semantic encoding vector to be evaluated into a high-risk-insurance semantic joint representation vector; calculating the high-risk-insurance semantic joint mean representation matrix and the high-risk-insurance semantic joint standard deviation representation matrix of the high-risk-insurance semantic joint representation vector, wherein the value of each position of the high-risk-insurance semantic joint mean representation matrix is ​​the value of the high-risk-insurance semantic joint representation vector. The value of each position of the high-risk-to-be-assessed insurance semantics joint standard deviation representation matrix is ​​the standard deviation of a pair of eigenvalues ​​of the two positions of the high-risk-to-be-assessed insurance semantics joint representation vector corresponding to the position coordinates; the transposed vector of the high-risk-to-be-assessed insurance semantics joint representation vector is matrix-multiplied with the high-risk-to-be-assessed insurance semantics joint mean representation matrix to obtain a first high-risk-to-be-assessed insurance semantics joint intermediate vector, and the high-risk-to-be-assessed insurance semantics joint standard deviation representation matrix is ​​matrix-multiplied with the high-risk-to-be-assessed insurance semantics joint representation vector The second high risk-to-be-assessed insurance semantics joint intermediate vector is obtained by multiplying the high risk-to-be-assessed insurance semantics joint representation vector, wherein the high risk-to-be-assessed insurance semantics joint representation vector is a column vector; the point sum of the first high risk-to-be-assessed insurance semantics joint intermediate vector and the transposed vector of the second high risk-to-be-assessed insurance semantics joint intermediate vector is calculated to obtain a third high risk-to-be-assessed insurance semantics joint intermediate vector; the matrix product of the high risk-to-be-assessed insurance semantics joint mean representation matrix and the high risk-to-be-assessed insurance semantics joint representation matrix is ​​calculated, and the transposed vector of the high risk-to-be-assessed insurance semantics joint representation vector is matrix multiplied by the matrix product to obtain a fourth high risk-to-be-assessed insurance semantics joint representation vector. Estimate the insurance semantic joint intermediate vector; calculate the point sum of the high-risk-to-be-evaluated insurance semantic joint intermediate vector and the fourth high-risk-to-be-evaluated insurance semantic joint intermediate vector to obtain a corrected high-risk-to-be-evaluated insurance semantic joint representation vector; split the corrected high-risk-to-be-evaluated insurance semantic joint representation vector into a corrected high-risk insurance semantic significant aggregation representation vector and a corrected insurance semantic coding vector to be evaluated based on the cascade of the high-risk insurance semantic significant aggregation representation vector and the insurance semantic coding vector to be evaluated; calculate the Hamming distance between the corrected high-risk insurance semantic significant aggregation representation vector and the corrected insurance semantic coding vector to be evaluated.

[0048] Here, considering that the semantic coding representation of the high-risk insurance semantically significant aggregation representation vector for each high-risk insurance data in the high-risk insurance data set is fused based on the significance enhancement of sequence distribution, the high-risk insurance semantically significant aggregation representation vector as a whole is relative to the semantic coding representation of the single-sample insurance content and customer information by the insurance semantic coding vector to be evaluated. This will cause local overflow characteristics due to the significance difference between samples, thereby affecting the mapping consistency of the high-risk insurance semantically significant aggregation representation vector and the insurance semantic coding vector to be evaluated to the common semantic Hamming distance representation space across sample domains, and affecting the calculation accuracy of the Hamming distance between the high-risk insurance semantically significant aggregation representation vector and the insurance semantic coding vector to be evaluated.

[0049] Based on this, the high-risk-to-be-assessed insurance semantics joint mean representation matrix and the high-risk-to-be-assessed insurance semantics joint standard deviation representation matrix of the group aggregation statistical evaluation of the local distribution of the feature value granularity of the high-risk-to-be-assessed insurance semantics joint representation vector for the joint semantic representation of the high-risk insurance semantics significant aggregation representation vector and the insurance semantics encoding vector to be assessed are used as the retrieval and response distribution enhancement of the high-risk-to-be-assessed insurance semantics joint representation vector, and a reference-free distribution retrieval response framework based on the group aggregation feature distribution open domain of the high-risk-to-be-assessed insurance semantics joint representation vector is constructed to avoid the distribution response redundancy caused by the local overflow characteristics of the high-risk-to-be-assessed insurance semantics joint representation vector through response superposition, so as to realize the faithfulness constraint of the response self-aggregation statistics correlation of the high-risk-to-be-assessed insurance semantics joint representation vector to the target common semantic Hamming distance representation space, improve the mapping consistency of the high-risk insurance semantics significant aggregation representation vector and the insurance semantics encoding vector to the common semantic Hamming distance representation space under cross-sample domains, and improve the calculation accuracy of the Hamming distance between the high-risk insurance semantics significant aggregation representation vector and the insurance semantics encoding vector to be assessed.

[0050] Specifically, the evaluation result determination unit 152 is used to determine whether the insurance to be evaluated is high risk based on the comparison between the Hamming distance and the preset threshold. That is, by comparing the Hamming distance with the preset threshold, the semantic difference between the insurance data to be evaluated and the known high-risk cases is converted into a clear risk level judgment, making the risk assessment process more objective and quantifiable. For example, when the Hamming distance exceeds the preset threshold, the insurance to be evaluated can be considered to be high risk; conversely, when the Hamming distance is lower than the preset threshold, it is low risk or medium risk. In this way, the risk assessment process of insurance data is simplified to the comparison between the Hamming distance and the preset threshold, which effectively simplifies the decision-making process and improves decision-making efficiency.

[0051] In summary, the insurance risk scoring management system based on artificial intelligence according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to perform semantic analysis on the data set of insurance data marked as high-risk, and extracts the core semantic feature representation of the high-risk insurance data set, thereby intelligently judging whether the insurance data to be evaluated is high-risk based on the difference in semantic features between the core semantic features of the high-risk insurance data and the insurance data to be evaluated. In this way, the accuracy and efficiency of insurance risk assessment are improved, and the interference of human factors can be reduced, reducing the risks caused by subjective judgment.

[0052] Figure 5 FIG. 1 is a flow chart of an insurance risk scoring management method based on artificial intelligence according to an embodiment of the present application. Figure 5 As shown, according to the embodiment of the present application, the insurance risk scoring management method based on artificial intelligence includes the steps of: S1, obtaining the insurance content and customer information in the insurance to be evaluated, and extracting a data set of insurance data marked as high-risk from a database, wherein the insurance data includes claim records, customer information and insurance content; S2, semantically encoding each insurance data marked as high-risk in the data set of insurance data marked as high-risk to obtain a set of high-risk insurance semantic coding vectors; S3, semantically strengthening and fusing the set of high-risk insurance semantic coding vectors to obtain a high-risk insurance semantic significant aggregation representation vector; S4, semantically encoding the insurance content and customer information in the insurance to be evaluated to obtain a semantic coding vector of the insurance to be evaluated; S5, determining whether the insurance to be evaluated is high-risk based on the semantic difference between the high-risk insurance semantic significant aggregation representation vector and the semantic coding vector of the insurance to be evaluated.

[0053] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned insurance risk scoring management method based on artificial intelligence have been referred to above. Figures 1 to 4 The invention has been described in detail in the description of the artificial intelligence-based insurance risk scoring management system, and therefore, its repeated description will be omitted.

[0054] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0055] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical sub-unit, that is, it may be located in one place, or it may be distributed on multiple network sub-units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0056] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0057] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. An insurance risk scoring management system based on artificial intelligence, characterized in that: include: An insurance data acquisition module, used to acquire the insurance content and customer information in the insurance to be evaluated, and extract a data set of insurance data marked as high risk from the database, wherein the insurance data includes claim records, customer information and insurance content; A high-risk insurance data semantic coding module, used for semantically coding each insurance data marked as high-risk in the data set of insurance data marked as high-risk to obtain a set of high-risk insurance semantic coding vectors; A high-risk insurance data semantic enhancement fusion module, used to perform semantic enhancement fusion on the set of high-risk insurance semantic encoding vectors to obtain a high-risk insurance semantic significant aggregation representation vector; A semantic coding module for insurance data to be evaluated, used for semantically coding the insurance content and customer information in the insurance to be evaluated to obtain a semantic coding vector for the insurance to be evaluated; The high-risk judgment module is used to determine whether the insurance to be evaluated is high-risk based on the semantic difference between the high-risk insurance semantic significant aggregation representation vector and the semantic encoding vector of the insurance to be evaluated.

2. The insurance risk scoring management system based on artificial intelligence according to claim 1 is characterized in that: The high-risk insurance data semantic encoding module is used to: Each insurance data marked as high risk in the data set of insurance data marked as high risk is passed through an insurance data semantic encoder based on the Bert model to obtain a set of high-risk insurance semantic encoding vectors.

3. The insurance risk scoring management system based on artificial intelligence according to claim 2 is characterized in that: The high-risk insurance data semantic enhancement fusion module is used to: The set of high-risk insurance semantic encoding vectors is input into a saliency enhancement fusion module based on sequence distribution cluster centers to obtain the high-risk insurance semantic saliency aggregation representation vector.

4. The insurance risk scoring management system based on artificial intelligence according to claim 3 is characterized in that: The high-risk insurance data semantic enhancement fusion module includes: A sequence distribution cluster center calculation unit, used to calculate the mean vector of the set of high-risk insurance semantic encoding vectors as the sequence distribution cluster center; A semantic relevance calculation unit, used for calculating the semantic relevance between each high-risk insurance semantic coding vector in the set of high-risk insurance semantic coding vectors and the sequence distribution cluster center to obtain a sequence of semantic relevance; A feature significance enhancement fusion unit is used to perform weighted fusion on the set of high-risk insurance semantic encoding vectors based on the sequence of semantic relevance to obtain the high-risk insurance semantic significance aggregation representation vector.

5. The artificial intelligence-based insurance risk scoring management system according to claim 4 is characterized in that: The semantic relevance calculation unit is used to: Respectively calculating the matrix products between each high-risk insurance semantic encoding vector in the set of high-risk insurance semantic encoding vectors and each predetermined weight matrix to obtain a set of high-risk insurance semantic representation encoding feature vectors; Adding a bias vector to each high-risk insurance semantic representation encoding feature vector in the set of high-risk insurance semantic representation encoding feature vectors and performing a Sigmoid activation function to obtain a set of activated high-risk insurance semantic representation encoding feature vectors; The vector products of each activated high-risk insurance semantic representation encoding feature vector in the set of activated high-risk insurance semantic representation encoding feature vectors and the transposed vector of the sequence distribution cluster center are calculated respectively to obtain the sequence of semantic relevance.

6. The artificial intelligence-based insurance risk scoring management system according to claim 5, characterized in that: The feature significance enhancement fusion unit is used to: Inputting the sequence of semantic relevance into a softmax function for soft maximum normalization processing to obtain a sequence of sequence-related significance weight factors; Taking the sequence of the sequence-related significance weight factors as the weight sequence, the vector-by-vector weighted sum of the set of the high-risk insurance semantic encoding vectors is calculated to obtain the high-risk insurance semantic significance aggregation representation vector.

7. The artificial intelligence-based insurance risk scoring management system according to claim 6, characterized in that: The semantic encoding module of the insurance data to be evaluated is used to: The insurance data semantic encoder based on the Bert model is used to semantically encode the insurance content and customer information in the insurance to be evaluated to obtain the semantic encoding vector of the insurance to be evaluated.

8. The artificial intelligence-based insurance risk scoring management system according to claim 7, characterized in that: The high-risk assessment module includes: A feature difference measurement unit, used to calculate the Hamming distance between the high-risk insurance semantic significant aggregation representation vector and the insurance semantic encoding vector to be evaluated; An evaluation result determination unit is used to determine whether the insurance to be evaluated is high risk based on a comparison between the Hamming distance and a preset threshold.

9. An insurance risk scoring management method based on artificial intelligence, characterized in that: include: Obtaining the insurance content and customer information in the insurance to be evaluated, and extracting a data set of insurance data marked as high risk from a database, wherein the insurance data includes claim records, customer information and insurance content; Performing semantic coding on each insurance data marked as high risk in the data set of insurance data marked as high risk to obtain a set of high-risk insurance semantic coding vectors; Performing semantic enhancement fusion on the set of high-risk insurance semantic encoding vectors to obtain a high-risk insurance semantic significant aggregation representation vector; Semantically encoding the insurance content and customer information in the insurance to be evaluated to obtain a semantic encoding vector of the insurance to be evaluated; Based on the semantic difference between the semantically significant aggregation representation vector of the high-risk insurance and the semantic encoding vector of the insurance to be evaluated, it is determined whether the insurance to be evaluated is high-risk.

Images