Insurance information processing method, device and equipment, and storage medium

Abstract

The application provides an insurance information processing method, device and equipment and a storage medium, and belongs to the technical field of data processing and machine learning. The method comprises the following steps: receiving an insurance information acquisition request sent by a user terminal, wherein the insurance information acquisition request comprises a user identifier to be processed; determining a target feature vector corresponding to the user identifier to be processed; inputting the target feature vector into an insurance recommendation model to obtain a target insurance type, wherein the insurance recommendation model is obtained by pre-training user features, insurance identifiers of insurance purchased by the user and cover vectors corresponding to the insurance identifiers; acquiring a target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier comprises an insurance identifier of held insurance; determining a to-be-sent insurance identifier according to the target insurance type and the target insurance identifier; and sending target insurance information corresponding to the to-be-sent insurance identifier to the user terminal. The method solves the problem of long time for a user to find insurance.

Description

Technical Field

[0001] This application relates to the fields of data processing and machine learning technology, and in particular to an insurance information processing method, apparatus, device and storage medium. Background Technology

[0002] With the continuous development of the economy and society, the types of insurance business have gradually become more diverse. Users can choose different types of insurance according to different needs, and insurance providers can also provide different types of insurance products for different users.

[0003] Currently, existing methods for recommending insurance typically recommend fixed types of insurance products to different types of users.

[0004] However, the inventors have discovered that the existing technology has at least the following technical problems: recommending fixed types of insurance products to different types of users will lead to a long time for users to search for insurance products. Summary of the Invention

[0005] This application provides an insurance information processing method, apparatus, device, and storage medium to solve the problem of long search times for insurance information.

[0006] In a first aspect, this application provides an insurance information processing method, comprising: receiving an insurance information acquisition request sent by a user terminal, wherein the insurance information acquisition request includes a user identifier to be processed; determining a target feature vector corresponding to the user identifier to be processed; inputting the target feature vector into an insurance recommendation model to obtain a target insurance type, wherein the insurance recommendation model is pre-trained using user features, insurance identifiers of insurance already purchased by the user, and masking vectors corresponding to the insurance identifiers; acquiring a target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier includes an insurance identifier for holding insurance; determining an insurance identifier to be sent based on the target insurance type and the target insurance identifier; and sending the target insurance information corresponding to the insurance identifier to be sent to the user terminal.

[0007] In one possible design, determining the insurance identifier to be sent based on the target insurance type and the target insurance identifier includes: finding the insurance type to be removed corresponding to the target insurance identifier; removing the insurance type to be removed from the target insurance type to obtain at least one suggested insurance type; and determining the insurance identifier to be sent based on the suggested insurance type.

[0008] In one possible design, the insurance identifier to be sent is determined according to the suggested insurance type, including: determining the insurance identifiers of the top M insurance policies with the highest sales volume in any suggested insurance type as high-sales insurance identifiers, where M is a positive integer; jointly determining the high-sales insurance identifiers corresponding to each suggested insurance type as insurance identifiers to be sent; or, determining the insurance identifiers of the top P insurance policies with the highest sales volume corresponding to each suggested insurance type as insurance identifiers to be sent, where P is a positive integer.

[0009] In one possible design, the target insurance identifier also includes a purchased and expired insurance identifier; correspondingly, after removing the insurance types to be removed from the target insurance types to obtain at least one suggested insurance type, the design further includes: obtaining each suggested insurance identifier corresponding to the suggested insurance type; obtaining the feature parameters corresponding to each suggested insurance identifier; determining the recommended insurance identifier among the suggested insurance identifiers based on the feature parameters; and jointly determining the purchased and expired insurance identifier and the recommended insurance identifier as the insurance identifier to be sent.

[0010] In one possible design, before inputting the target feature vector into the insurance recommendation model to obtain the target insurance type, the process further includes: obtaining the user features corresponding to the target user identifier and the insurance identifier of the purchased insurance, where the target user identifier is the user identifier of any user; converting the user features into feature vectors; determining the target vector based on the insurance identifier of the purchased insurance; determining the masking vector based on the target vector; determining the output vector based on the target vector and the masking vector; determining the feature vectors corresponding to N target user identifiers and the output vector as training data, where N is a positive integer; and training the model using the training data to obtain the insurance recommendation model.

[0011] In one possible design, the target vector is determined based on the insurance identifier of the purchased insurance, including: determining the type of insurance purchased based on the insurance identifier of the purchased insurance; and determining the target vector based on the type of insurance purchased.

[0012] In one possible design, determining the masking vector based on the target vector includes: obtaining the target direction in the target vector, where the target direction is the direction corresponding to the type of insurance not purchased; determining at least one random direction from the target directions as the masking direction; and generating a masking vector based on each masking direction.

[0013] In one possible design, before determining the target vector based on the insurance identifiers of the purchased insurance, the process includes: reading the insurance information corresponding to each insurance identifier; obtaining the semantic feature vectors of each insurance information; calculating the similarity between each semantic feature vector; clustering each insurance identifier based on the similarity between each semantic feature vector to obtain multiple insurance types; and determining the insurance type vector corresponding to each insurance type.

[0014] Secondly, this application provides an insurance information processing apparatus, comprising: a request receiving module for receiving an insurance information acquisition request sent by a user terminal, wherein the insurance information acquisition request includes a user identifier to be processed; a vector determination module for determining a target feature vector corresponding to the user identifier to be processed; a type acquisition module for inputting the target feature vector into an insurance recommendation model to obtain a target insurance type, wherein the insurance recommendation model is pre-trained using user features, insurance identifiers of insurance already purchased by the user, and the masking vectors corresponding to the insurance identifiers; an identifier acquisition module for acquiring a target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier includes an insurance identifier for holding insurance; an identifier determination module for determining an insurance identifier to be sent based on the target insurance type and the target insurance identifier; and an information sending module for sending the target insurance information corresponding to the insurance identifier to be sent to the user terminal.

[0015] Thirdly, this application provides an electronic device, including: a processor and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, causing the processor to perform the insurance information processing method as described in the first aspect.

[0016] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the insurance information processing method as described in the first aspect.

[0017] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the insurance information processing method as described in the first aspect.

[0018] The insurance information processing method, apparatus, equipment, and storage medium provided in this application, after receiving an insurance information retrieval request sent by a user terminal, determine the target user features corresponding to the user identifier to be processed, convert the target user features into a target feature vector, input the target feature vector into a pre-trained insurance recommendation model to obtain the target insurance type, obtain the target insurance identifier of the insurance already purchased corresponding to the user identifier to be processed, filter the target insurance types using the target insurance identifier to obtain the insurance identifier to be sent, and send the target insurance information corresponding to the insurance identifier to be sent to the user terminal. This achieves the goal of finding the corresponding target insurance type by combining the insurance recommendation model with the user's features. The insurance recommendation model used employs a masking vector to obtain the results of insurance types that have not been purchased, and filters the target insurance type using the insurance types already held by the user identifier to be processed, thus avoiding the problem of pushing the same type of insurance that has been purchased and has not expired to the user. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] Figure 1 This is a schematic diagram illustrating an application scenario of the insurance information processing method provided in the embodiments of this application;

[0021] Figure 2 A flowchart illustrating the insurance information processing method provided in this application embodiment;

[0022] Figure 3 A flowchart illustrating the process from insurance classification to insurance type recommendation, provided for an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of the insurance information processing device provided in the embodiments of this application;

[0024] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0025] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0026] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0027] With the continuous development of the economy and society, the public's demand for insurance is increasing in order to protect their rights. Correspondingly, the types of insurance are also increasing, making it more difficult for users to find suitable insurance for themselves.

[0028] Currently, to reduce the time users spend searching for insurance products, one approach is to obtain user type information and recommend corresponding insurance products based on that type. However, each user's individual preferences may differ, meaning the recommended insurance products might not be suitable for them, thus increasing the time users spend searching for insurance products.

[0029] To address the aforementioned technical problems, the inventors propose the following technical concept: Upon receiving an insurance information retrieval request from a user terminal, user characteristics are determined based on the user identifier. These user characteristics are then input into a pre-trained insurance recommendation model to obtain the target insurance type. Insurance types already held by the user are removed from the target insurance type, resulting in the final target insurance type. An insurance identifier to be sent is selected from the target insurance type, and the insurance information corresponding to this identifier is sent to the user terminal. The insurance recommendation model can be pre-trained using different user feature data, insurance identifiers of the user's purchased insurance, and corresponding masking vectors. The masking vector can represent at least one direction in the vector corresponding to an unpurchased insurance type.

[0030] This application is applied to scenarios involving the processing of insurance information. The collection, storage, use, processing, transmission, provision, and disclosure of financial data or user data involved in the technical solution of this application all comply with relevant laws and regulations and do not violate public order and good morals.

[0031] Figure 1 This is a schematic diagram illustrating an application scenario of the insurance information processing method provided in an embodiment of this application. For example... Figure 1 In this scenario, the components include: server 101 and user terminal 102.

[0032] In the specific implementation process, server 101 can be implemented using a cluster of one or more servers with more powerful processing capabilities and higher security. Where possible, it can also be replaced by computers, laptops, etc. with strong computing power.

[0033] User terminal 102 may include computers, servers, tablets, mobile phones, PDAs (personal digital assistants), and laptops, etc., which can input and output data.

[0034] Server 101 is used to receive insurance information retrieval requests sent by user terminal 102, filter out suitable insurance information from the insurance information retrieval requests, and send it to user terminal 102. User terminal 102 is used to accept user control, send insurance information retrieval requests to server 101, receive target insurance information sent by server 101, and output the target insurance information.

[0035] It is understood that the structure illustrated in the embodiments of this application does not constitute a specific limitation on the insurance information processing method. In other feasible embodiments of this application, the above architecture may include more or fewer components than illustrated, or combine some components, or split some components, or arrange different components, which can be determined according to the actual application scenario and is not limited here. Figure 1 The components shown can be implemented by hardware, software, or a combination of both.

[0036] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0037] Figure 2 This is a flowchart illustrating the insurance information processing method provided in an embodiment of this application. The executing entity of this embodiment may be... Figure 1 The server 101 in this embodiment can also be a computer and / or a mobile phone, etc., and this embodiment does not impose any special restrictions on it. Figure 2 As shown, the method includes:

[0038] S201: Receive an insurance information retrieval request sent by a user terminal, wherein the insurance information retrieval request includes the user identifier to be processed.

[0039] In this step, the insurance information retrieval request can be a message, a string, etc. The user identifier to be processed can be the user's identifier.

[0040] S202: Determine the target feature vector corresponding to the user identifier to be processed.

[0041] This step may include determining the target user features corresponding to the user identifier to be processed, and converting the target user features into a target feature vector. Specifically, this could involve finding the correspondence between user identifiers and user features to obtain the target user features corresponding to the user identifier to be processed. Alternatively, it could involve searching a database based on the user identifier to be processed to obtain the target user features. Another approach is to send a user feature acquisition request to the user terminal and receive the target user features sent by the user terminal based on the authorization information input by the user. A preset algorithm is then used to convert each target user feature into a vector, and the vectors are concatenated to obtain the target feature vector. Alternatively, the textual descriptions in the target user features can be converted into numbers using a natural language model, the originally numerical content in the target user features can be converted into a preset number format, and the converted numbers can be concatenated to obtain the target feature vector.

[0042] User characteristics may include gender, age, income, etc. It should be noted that all this information and data is authorized by the user or fully authorized by all parties, and the collection, use, and processing of this data must comply with relevant laws, regulations, and standards. A corresponding access point is provided for users to choose whether to authorize or refuse authorization.

[0043] For example, the "male" in the target user feature is converted to 1, the age or age range is converted to binary, and then concatenated to obtain the target feature vector. Another example is converting "male" in the target user feature to "

[01] ", the age is 25, and the age or age range is converted to "[1,1,0,0,1]". Alternatively, blank vectors can be prepared in advance for each user feature, and the converted data can be written into these blank vectors. For example, the blank vector for age is "[x,x,x,x,x,x,x]", and the target user feature is age 32. The binary number 32 is written into the blank vector, resulting in "[0,1,0,0,0,0,0]". Since the number of bits in binary 32 is less than the number of bits in the blank vector, 0 is used to fill the empty positions. This application does not impose specific restrictions on the specific content of the target user feature or the number of bits in the converted vector; the vector can also use other bases besides binary. The concatenation method can be to connect the first and last elements in a preset order. For example, if we have two vectors, “[1,0,1,0,0,1]” and “[0,1,0]”, then the target feature vector obtained by concatenation is “[1,0,1,0,0,1,0,1,0]”.

[0044] S203: Input the target feature vector into the insurance recommendation model to obtain the target insurance type. The insurance recommendation model is pre-trained using user features, insurance identifiers of insurance already purchased by the user, and the corresponding masking vectors of the insurance identifiers.

[0045] In this step, the target feature vector can be used as input, and the output of the insurance recommendation model can be the target insurance type. Alternatively, the output of the insurance recommendation model can be a vector corresponding to the target insurance type, and the target insurance type can be found by looking up the correspondence between the vector and the insurance type.

[0046] During the training of the model, the user features and insurance identifiers of the user's purchased insurance were obtained with the user's authorization.

[0047] S204: Obtain the target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier includes the insurance identifier that holds insurance.

[0048] In this step, the database or a mapping between the user identifier and the insurance identifier held by the user can be searched to obtain the corresponding target insurance identifier. The held insurance can be insurance that the user has purchased and that has not yet expired. The target insurance identifier can also include the insurance identifier corresponding to insurance that has been purchased but expired.

[0049] S205: Determine the insurance identifier to be sent based on the target insurance type and target insurance identifier.

[0050] This step may include finding the insurance type corresponding to the target insurance identifier, removing the insurance type corresponding to the target insurance identifier from the target insurance types to obtain filtered insurance types, and selecting insurance with higher sales volume or positive review rate from the filtered insurance types as the insurance to be sent, with the identifier of the insurance to be sent being the insurance to be sent identifier. It may also include prioritizing the insurance identifiers of insurance products that the user has previously purchased as the insurance to be sent identifier, and secondarily using the insurance identifiers of insurance products with high sales volume or positive review rate.

[0051] For example, if the current target insurance types include A, B, C, and D, and the target insurance identifiers include a and c, where a and c correspond to insurance types A and C respectively, then after filtering, the insurance types include B and D. In this case, the top 10% in sales or positive review rate, the top 10, or the top 20 in ranking are selected from B and D as insurance to be sent, and the identifier corresponding to the insurance to be sent is the insurance to be sent identifier.

[0052] S206: Send the target insurance information corresponding to the insurance identifier to be sent to the user terminal.

[0053] In this step, the target insurance information corresponding to the insurance identifier to be sent can be read and then sent to the user terminal.

[0054] The target insurance information may include the name of the insurance, the type of insurance, and a description of the insurance.

[0055] As described in the above embodiments, this application embodiment, after receiving an insurance information retrieval request sent by a user terminal, determines the target user features corresponding to the user identifier to be processed, converts the target user features into a target feature vector, inputs the target feature vector into a pre-trained insurance recommendation model to obtain the target insurance type, obtains the target insurance identifier of the purchased insurance corresponding to the user identifier to be processed, filters the target insurance type using the target insurance identifier to obtain the insurance identifier to be sent, and sends the target insurance information corresponding to the insurance identifier to be sent to the user terminal. This achieves the goal of finding the corresponding target insurance type by combining the insurance recommendation model with the user's features. The insurance recommendation model used employs a masking vector, which can obtain the results of insurance types that have not been purchased. Furthermore, the target insurance type is filtered using the insurance types already held by the user identifier to be processed, thus avoiding the problem of pushing the same type of insurance that has been purchased and has not expired to the user.

[0056] In one possible implementation, in step S205 above, the insurance identifier to be sent is determined based on the target insurance type and the target insurance identifier, including:

[0057] S2051: Locate the insurance type to be removed corresponding to the target insurance identifier.

[0058] In this step, you can look up the correspondence between insurance identifiers and insurance types to obtain the insurance type to which the target insurance identifier belongs, i.e., the insurance type to be removed.

[0059] S2052: Remove the insurance types to be removed from the target insurance types to obtain at least one suggested insurance type.

[0060] In this step, the insurance types to be removed from the target insurance types can be deleted to obtain suggested insurance types. This step is similar to the example in step S205 above, and will not be repeated here.

[0061] S2053: Determine the insurance identifier to be sent based on the recommended insurance type.

[0062] In this step, you can select X insurance identifiers for each suggested insurance type as insurance identifiers to be sent.

[0063] The selection method can be either using preset filtering conditions or random selection.

[0064] As can be seen from the description of the above embodiments, the embodiments of this application obtain suggested insurance types by finding the insurance type corresponding to the target insurance identifier and removing the insurance type to which the target insurance identifier belongs in the target insurance type. Then, the suggested insurance type is found to be sent. This realizes the elimination of the insurance type currently held by the user, making the recommended insurance more in line with the user's needs and reducing the time the user spends searching for insurance.

[0065] In one possible implementation, step S2053 above determines the insurance identifier to be sent based on the suggested insurance type, including: S53A or S53B.

[0066] S53A: For any suggested insurance type, the insurance identifiers of the top M best-selling insurance policies are identified as high-selling insurance identifiers, where M is a positive integer. The high-selling insurance identifiers corresponding to all suggested insurance types are collectively identified as insurance identifiers to be sent.

[0067] In this step, for example, if there are three suggested insurance types A, B, and C, and the M value is 5, then the top 5 insurance types in terms of sales volume for type A are identified as high-sales insurance types, the top 5 insurance types in terms of sales volume for type B are also identified as high-sales insurance types, and the top 5 insurance types in terms of sales volume for type C are also identified as high-sales insurance types. All high-sales insurance types are identified as insurance types to be sent.

[0068] S53B: Determine the insurance identifiers of the top P insurance policies with the highest sales for each suggested insurance type as the insurance identifiers to be sent, where P is a positive integer.

[0069] In this step, for example, if there are three suggested insurance types A, B, and C, and P is 10, then the insurance identifiers of the top 10 best-selling insurance types (A, B, and C) are determined as the insurance identifiers to be sent. This application embodiment does not impose specific limitations on the value of P.

[0070] As can be seen from the description of the above embodiments, the embodiments of this application determine the insurance identifiers of the top-selling insurance products as the insurance identifiers to be sent, thereby achieving the effect of prioritizing the recommendation of insurance products that are frequently selected by users and reducing the time users spend selecting insurance products.

[0071] In one possible implementation, the target insurance identifier also includes an identifier for purchased and expired insurance.

[0072] Among them, the "purchased and expired insurance" identifier refers to the insurance identifier of the user identifier that has been purchased and has expired.

[0073] Accordingly, after removing the insurance type to be removed from the target insurance type in step S2052 above to obtain at least one suggested insurance type, the process further includes:

[0074] S521: Obtain the suggested insurance identifiers corresponding to the suggested insurance types.

[0075] In this step, you can look up the correspondence between insurance types and insurance identifiers to obtain the recommended insurance identifier corresponding to the recommended insurance type.

[0076] The correspondence between insurance types and insurance identifiers can be pre-determined through clustering of insurance identifiers. The process of clustering insurance identifiers can be found in the following example.

[0077] S522: Obtain the feature parameters corresponding to each suggested insurance identifier.

[0078] In this step, the feature parameters can be sales volume, click-through rate, etc.

[0079] S523: Determine the recommended insurance identifier in the suggested insurance identifier based on the feature parameters.

[0080] In this step, you can either select the top 10% of insurance products by sales volume or P insurance products as the recommended insurance products. Alternatively, you can select the top 10% of insurance products by click-through rate or P insurance products as the recommended insurance products.

[0081] S524: The purchased and expired insurance label and the recommended insurance label are jointly identified as the insurance label to be sent.

[0082] In this step, for example, if the purchased and expired insurance is identified as A, and the recommended insurance identifiers include B, C, and D, then A, B, C, and D will be collectively identified as the insurance identifiers to be sent.

[0083] As can be seen from the description of the above embodiments, the embodiments of this application, by selecting insurance identifiers that have been purchased and expired, as well as insurance identifiers that have not been purchased, as insurance identifiers to be sent, achieve the effect of showing users insurance with a higher probability of purchase, thereby reducing the time users spend selecting insurance.

[0084] In one possible implementation, before step S203 above, which inputs the target feature vector into the insurance recommendation model to obtain the target insurance type, the following is also included:

[0085] S210: Obtain the user characteristics corresponding to the target user identifier and the insurance identifier of the purchased insurance, wherein the target user identifier is the user identifier of any user.

[0086] In this step, the process of obtaining user characteristics is similar to step S202 above, and will not be repeated here. Obtaining the insurance identifier corresponding to the target user identifier can be done by searching the database based on the target user identifier, or by finding the correspondence between user identifiers and insurance identifiers for purchased insurance.

[0087] The correspondence between user identifiers and insurance identifiers that have been purchased can be predefined or stored in a table or key-value pair format.

[0088] S211: Convert user features into feature vectors.

[0089] In this step, BERT (Bidirectional Encoder Representation from Transformers) can be used to convert user features into vectors, and the vectors can be concatenated end to end to obtain a feature vector. During the conversion of user features into feature vectors, the user's personal features are mapped to a low-dimensional vector space. Textual descriptive features are mapped using a natural language model, while numerical and category features are processed as one-hot vectors.

[0090] S212: Determine the target vector based on the insurance identifier that has been purchased.

[0091] In this step, the mapping relationship between the insurance type and the vector can be found based on the insurance type to which the purchased insurance identifier belongs, thus obtaining the target vector. The target vector can be the insurance type vector corresponding to the insurance identifier.

[0092] In one possible implementation, in this step, the target vector is determined based on the insurance identifier of the purchased insurance, including steps S2121 and S2122.

[0093] S2121: Determine the type of insurance purchased based on the insurance identifier of the purchased insurance.

[0094] In this step, you can look up the correspondence between the insurance identifier and the insurance type based on the insurance identifier that has been purchased, and then obtain the type of insurance that has been purchased corresponding to the insurance identifier.

[0095] The correspondence between insurance identifiers and insurance types can be pre-stored, and the format can be a table or key-value pairs.

[0096] S2122: Determine the target vector based on the type of insurance already purchased.

[0097] In this step, the correspondence between insurance types and vectors can be found to obtain the vectors corresponding to the purchased insurance types, and the vectors of each purchased insurance type can be superimposed to obtain the target vector.

[0098] For example, if the purchased insurance type is insurance type A, and the vector corresponding to insurance type A is "[1,0,0,0,0]", then the target vector is "[1,0,0,0,0]". As another example, if the purchased insurance type is insurance type B, and the vector corresponding to insurance type B is "[0,1,0,0,0]", then the target vector is "[0,1,0,0,0]". Furthermore, if the purchased insurance types are insurance types A and B, then the target vector is "[1,1,0,0,0]". Yet another example is if the purchased insurance types are A, B, and C, where the vectors corresponding to A and B are the same as in the examples above, and the vector corresponding to C is "[0,0,1,0,0]", then the target vector is "[1,1,1,0,0]". Finally, if the user has purchased products ABC, with AB belonging to category 1 and C belonging to category 2, then the target vector is "[1,1,0,0,…]", with a dimension of 1×K, where K is the total number of insurance categories.

[0099] S213: Determine the occlusion vector based on the target vector.

[0100] In this step, the masking vector can be selected from the direction of the target vector that represents the type of insurance that has not been purchased, and the masking vector is generated from the masking direction.

[0101] For example, in step S212 above, if the purchased insurance type is A or B, and the target vector is "[1,1,0,0,0]", then at least one of the directions of the last three "0"s can be selected as the masking direction. The generated masking vector can be "[1,1,0,1,1]", "[1,1,1,0,1]", or "[1,1,1,0,0]", etc. In the masking vector, "0" represents the masking direction. For another example, if the target vector is "[1,1,1,0,0]", at least one of the directions of the last two "0"s can be selected as the masking direction, resulting in masking vectors such as "[1,1,1,0,1]", "[1,1,1,1,0]", or "[1,1,1,0,0]". When "0" represents the purchased insurance type, "1" can also be selected as the masking vector. For yet another example, the target vector corresponding to a user identifier is y = [y1,y2,…,y]. k ], where y i =1 indicates that the user has purchased the i-th category, y i =0 indicates that the user has not purchased the i-th category. The randomly generated masking vector is m = [m1, m2, ..., m...]. k ], where m i ∈{0,1} indicates whether the i-th position is covered, 0 indicates covered, 1 indicates uncovered. When yi is 0, mi may be 0.

[0102] S214: Determine the output vector based on the target vector and the occlusion vector.

[0103] In this step, the occlusion direction value of the occlusion vector in the target vector can be removed to obtain the output vector. Introducing the occlusion direction does not affect the result of the loss function.

[0104] For example, if the target vector is "[0,0,1,0,1]" and the masking vector is "[0,1,1,1,1]", then removing the first "0" will result in the output vector "[,0,1,0,1]". As another example, if the target vector is "[0,1,1,0,1]" and the masking vector is "[1,1,1,0,1]", then removing the fourth "0" will result in the output vector "[0,1,1,,1]".

[0105] S215: Determine the feature vectors and output vectors corresponding to the N target user identifiers as training data, where N is a positive integer.

[0106] In this step, the training data may include a training set and a test set. This application embodiment does not impose specific limitations on the value of N.

[0107] S216: Use training data to train the model and obtain the insurance recommendation model.

[0108] In this step, feature vectors from the training data can be used as input parameters to train the model, and the output vectors can be used as the model's output values. Once the error is less than a preset error, an insurance recommendation model is obtained. The trained insurance recommendation model can be a multi-class, multi-label model.

[0109] As described in the above embodiments, this application embodiment obtains the user features corresponding to the target user identifier and the insurance identifier of the purchased insurance, converts the user features into feature vectors, obtains the corresponding target vector from the insurance identifier of the purchased insurance, determines the masking vector from the direction in the target vector that indicates that the insurance has not been purchased, and obtains the output vector for training by combining the target vector and the masking vector. The feature vector corresponding to the user features is used as the input parameter for model training, so that the trained model can obtain the vector corresponding to the insurance that the user may purchase from the user features. Furthermore, since the masking vector is used, the target insurance type obtained by the model can also include the insurance type that the user has not purchased, increasing the breadth of insurance type recommendations.

[0110] In one possible implementation, in step S213 above, the occlusion vector is determined based on the target vector, including steps S13A, S13B, and S13C.

[0111] S13A: Obtain the target direction in the target vector, where the target direction corresponds to the direction of the unpurchased insurance type.

[0112] In this step, for example, if the current target vector is "[0,1,0,0,1]", where "1" represents the type of insurance that has been purchased and "0" represents the type of insurance that has not been purchased, then the direction of "0" corresponds to the direction of the purchased insurance type. This application embodiment does not impose specific limitations on the content of the target vector.

[0113] S13B: Determine at least one random direction from the target directions as the occlusion direction.

[0114] In this step, for example, the direction of any "0" in step S13A above can be selected as the masking direction. The masking direction can be the direction represented by the position of the "0" in the target vector.

[0115] S13C: Generates occlusion vectors based on each occlusion direction.

[0116] In this step, the same dimension as the target vector or a preset dimension can be used to generate the vector to be filled. The position corresponding to the occlusion direction in the vector to be filled is written with a value indicating occlusion, and the other positions are written with a value indicating no occlusion, thus obtaining the occlusion vector.

[0117] For example, if the preset dimension is 6, the vector to be filled is "[,,,,,]". If the position corresponding to the occlusion direction is the second or fourth position, it means the value to be occluded is "0" and the value to be unoccluded is "1", then the resulting occlusion vector is "[1,0,1,0,1,1]". As another example, if the preset dimension is 5, the vector to be filled is "[,,,,]". If the position corresponding to the occlusion direction is the first or third position, it means the value to be occluded is "0" and the value to be unoccluded is "1", then the resulting occlusion vector is "[0,1,0,1,1,1]".

[0118] As can be seen from the description of the above embodiments, the embodiments of this application obtain the target direction in the target vector, determine at least one random direction in the target direction as the occlusion direction, generate an occlusion vector according to each occlusion direction, so as to facilitate the subsequent removal of the value in the target vector according to the occlusion vector to obtain the output vector. By using the occlusion vector, the output vector corresponding to the insurance type that the user has not purchased can be output.

[0119] In one possible implementation, before determining the target vector based on the insurance identifier of the purchased insurance in step S212, the following method is also included:

[0120] S220: Read the insurance information corresponding to each insurance identifier.

[0121] In this step, you can read the insurance information document corresponding to the insurance identifier to obtain the insurance information in the insurance information document.

[0122] The insurance information may include the insurance product name, insurance type, coverage, coverage period, premium, sum insured, deductible, waiting period, and claims process.

[0123] S221: Obtain the semantic feature vectors of each insurance information.

[0124] In this step, the method for obtaining semantic feature vectors can be to use the BERT model described above to convert insurance information into vectors.

[0125] S222: Calculate the similarity between each semantic feature vector.

[0126] In this step, the cosine value between each semantic feature vector can be calculated. The larger the cosine value, the more similar the documents are.

[0127] S223: Based on the similarity between the semantic feature vectors, the insurance information is clustered to obtain multiple insurance types.

[0128] The insurance identifiers can be clustered using K-Means (K-means clustering algorithm) or other clustering algorithms to obtain multiple insurance types.

[0129] One method for clustering using K-Means is to randomly select several semantic feature vectors as initial cluster centers, use the similarity between semantic feature vectors as the distance between them, and assign each semantic feature vector to the nearest cluster center until no (or the minimum number) semantic feature vectors are reassigned to different clusters, no (or the minimum number) cluster centers change, or the sum of squared errors is locally minimized.

[0130] S224: Determine the insurance type vector corresponding to each insurance type.

[0131] In this step, the number of clusters obtained from clustering can be used as the dimension of the insurance type vector, with each cluster representing an insurance type. A vector is then assigned to each insurance type.

[0132] The process of assigning a vector to each insurance type can be done in two ways: either randomly assigning a vector to each insurance type, or assigning a vector to each insurance type based on the number of insurance identifiers included in that insurance type.

[0133] For example, if we have 5 insurance types A, B, C, D, and E, we can assign insurance type vectors as “[1,0,0,0,0]”, “[0,1,0,0,0]”, “[0,0,1,0,0]”, “[0,0,0,1,0]”, “[0,0,0,1,0]”, and “[0,0,0,0,1]” respectively.

[0134] As can be seen from the description of the above embodiments, the embodiments of this application read the insurance information corresponding to each insurance identifier, extract the semantic feature vector of each insurance information, calculate the similarity between each semantic feature vector, and use the semantic feature vector and the similarity between each semantic feature vector to cluster each insurance identifier to obtain multiple insurance types. An insurance type vector is assigned to each insurance type to realize the representation of insurance type with vectors, which facilitates the subsequent use of vectors as input to the model.

[0135] Figure 3 This is a schematic diagram illustrating the process from insurance classification to insurance type recommendation, provided as an embodiment of this application. Figure 3As shown, the insurance classification process includes: reading the insurance document corresponding to the insurance identifier; obtaining the semantic feature vector corresponding to the insurance identifier using the insurance document; clustering using the semantic feature vector to obtain multiple insurance types; and assigning an insurance type vector to each insurance type. The model training process after insurance classification includes: determining training data using user characteristics of users who have purchased insurance and the insurance identifiers of those purchased insurance; and training the model using the training data to obtain an insurance recommendation model. The insurance type recommendation process includes: determining the target user characteristics based on the received user identifiers to be processed, inputting the target user characteristics into the insurance recommendation model to obtain the target insurance type; removing insurance types held by the target user from the target insurance type and filtering the remaining insurance to obtain the insurance identifier to be sent; and sending the target insurance information corresponding to the insurance identifier to be sent to the user terminal.

[0136] Figure 4 This is a schematic diagram of the structure of the insurance information processing device provided in an embodiment of this application. Figure 4 As shown, the insurance information processing device 400 includes: a request receiving module 401, a vector determination module 402, a type obtaining module 403, an identifier obtaining module 404, an identifier determining module 405, and an information sending module 406.

[0137] The request receiving module 401 is used to receive an insurance information retrieval request sent by a user terminal, wherein the insurance information retrieval request includes a user identifier to be processed.

[0138] The vector determination module 402 is used to determine the target feature vector corresponding to the user identifier to be processed.

[0139] The type acquisition module 403 is used to input the target feature vector into the insurance recommendation model to obtain the target insurance type. The insurance recommendation model is trained in advance using user features, insurance identifiers of insurance that the user has purchased, and the masking vectors corresponding to the insurance identifiers.

[0140] The identifier acquisition module 404 is used to acquire the target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier includes the insurance identifier of the holder of insurance.

[0141] The identifier determination module 405 is used to determine the insurance identifier to be sent based on the target insurance type and the target insurance identifier.

[0142] The information sending module 406 is used to send the target insurance information corresponding to the insurance identifier to be sent to the user terminal.

[0143] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0144] In one possible implementation, the identifier determination module 405 is specifically used to find the insurance type to be removed corresponding to the target insurance identifier. The insurance types to be removed from the target insurance types are then removed to obtain at least one suggested insurance type. Based on the suggested insurance type, the insurance identifier to be sent is determined.

[0145] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0146] In one possible implementation, the identifier determination module 405 is specifically used to determine the insurance identifiers of the top M best-selling insurance policies for any suggested insurance type as high-selling insurance identifiers, where M is a positive integer. The high-selling insurance identifiers corresponding to each suggested insurance type are then collectively determined as insurance identifiers to be sent. Alternatively, the insurance identifiers of the top P best-selling insurance policies corresponding to each suggested insurance type are determined as insurance identifiers to be sent, where P is a positive integer.

[0147] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0148] In one possible implementation, the target insurance identifier also includes a purchased and expired insurance identifier. The identifier determination module 405 is further used to obtain each suggested insurance identifier corresponding to the suggested insurance type; obtain the feature parameters corresponding to each suggested insurance identifier; determine the recommended insurance identifier among the suggested insurance identifiers based on the feature parameters; and jointly determine the purchased and expired insurance identifier and the recommended insurance identifier as the insurance identifier to be sent.

[0149] In one possible implementation, the insurance information processing device 400 further includes a model training module 407.

[0150] The model training module 407 is used to obtain the user features corresponding to the target user identifier and the insurance identifier of the purchased insurance, where the target user identifier is the user identifier of any user. The user features are converted into feature vectors. Based on the insurance identifier of the purchased insurance, the target vector is determined. Based on the target vector, the masking vector is determined. Based on the target vector and the masking vector, the output vector is determined. The feature vectors corresponding to N target user identifiers and the output vector are used as training data, where N is a positive integer. The model is trained using the training data to obtain the insurance recommendation model.

[0151] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0152] In one possible implementation, the model training module 407 is specifically used to determine the type of insurance purchased based on the insurance identifier. Based on the type of insurance purchased, the target vector is then determined.

[0153] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0154] In one possible implementation, the model training module 407 is specifically used to obtain the target direction in the target vector, where the target direction corresponds to the type of insurance not purchased. At least one random direction from the target directions is determined as the occlusion direction. An occlusion vector is generated based on each occlusion direction.

[0155] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0156] In one possible implementation, the insurance information processing device 400 further includes a type determination module 408.

[0157] The type determination module 408 is used to read the insurance information corresponding to each insurance identifier. It obtains the semantic feature vector of each insurance information item. It calculates the similarity between each semantic feature vector. Based on the similarity between the semantic feature vectors, it clusters each insurance identifier to obtain multiple insurance types. Finally, it determines the insurance type vector corresponding to each insurance type.

[0158] The apparatus provided in this embodiment can be used to execute the technical solutions of the above method embodiments. Its implementation principle and technical effects are similar, and will not be described again here.

[0159] To implement the above embodiments, this application also provides an electronic device.

[0160] refer to Figure 5 The diagram illustrates a structural schematic of an electronic device 500 suitable for implementing embodiments of this application. The electronic device 500 can be a terminal device or a server. The terminal device can include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, personal digital assistants (PDAs), portable Android devices (PADs), portable media players (PMPs), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0161] like Figure 5 As shown, the electronic device 500 may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 501 and a memory 502 communicatively connected to the processor. The processor can perform various appropriate actions and processes based on programs stored in the memory 502, computer-executed instructions, or programs loaded from the storage device 508 into the random access memory (RAM) 503, to implement the insurance information processing method in any of the above embodiments. The memory may be a read-only memory (ROM). The RAM 503 also stores various programs and data required for the operation of the electronic device 500. The processing device 501, the memory 502, and the RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0162] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0163] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a memory 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of the embodiments of this application.

[0164] It should be noted that the computer-readable storage medium described above in this application can be a computer-readable signal medium or a computer storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0165] The aforementioned computer-readable storage medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0166] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0167] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0168] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0169] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the units are not necessarily limiting of the module itself; for example, a request receiving module can also be described as an "insurance information acquisition request receiving module".

[0170] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0171] This application also provides a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, it implements the technical solution of the insurance information processing method in any of the above embodiments. Its implementation principle and beneficial effects are similar to those of the insurance information processing method. Please refer to the implementation principle and beneficial effects of the insurance information processing method, which will not be repeated here.

[0172] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0173] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the technical solution of the insurance information processing method in any of the above embodiments. Its implementation principle and beneficial effects are similar to those of the insurance information processing method, and can be found in the implementation principle and beneficial effects of the insurance information processing method, which will not be repeated here.

[0174] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

[0175] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0176] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for processing insurance information, characterized in that, include: Receive an insurance information retrieval request sent by a user terminal, wherein the insurance information retrieval request includes the user identifier to be processed; Determine the target feature vector corresponding to the user identifier to be processed; The target feature vector is input into the insurance recommendation model to obtain the target insurance type, wherein the insurance recommendation model is pre-trained using user features, insurance identifiers of insurance purchased by the user, and the masking vectors corresponding to the insurance identifiers. Obtain the target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier includes an insurance identifier that holds insurance; Based on the target insurance type and the target insurance identifier, determine the insurance identifier to be sent; Send the target insurance information corresponding to the insurance identifier to be sent to the user terminal; Before inputting the target feature vector into the insurance recommendation model to obtain the target insurance type, the method further includes: obtaining the user features corresponding to the target user identifier and the insurance identifier of the purchased insurance, wherein the target user identifier is the user identifier of any user; converting the user features into feature vectors; determining the target vector based on the insurance identifier of the purchased insurance; determining the masking vector based on the target vector; determining the output vector based on the target vector and the masking vector; determining the feature vectors corresponding to N target user identifiers and the output vector as training data, where N is a positive integer; and training the model using the training data to obtain the insurance recommendation model. The step of determining the occlusion vector based on the target vector includes: obtaining the target direction in the target vector, wherein the target direction is the direction corresponding to the type of insurance not purchased; determining at least one random direction among the target directions as the occlusion direction; and generating the occlusion vector based on each occlusion direction.

2. The method according to claim 1, characterized in that, The step of determining the insurance identifier to be sent based on the target insurance type and the target insurance identifier includes: Find the type of insurance to be removed corresponding to the target insurance identifier; Remove the insurance type to be removed from the target insurance type to obtain at least one suggested insurance type; Based on the suggested insurance type, determine the insurance identifier to be sent.

3. The method according to claim 2, characterized in that, The step of determining the insurance identifier to be sent based on the suggested insurance type includes: For any given insurance type, the top M best-selling insurance policies are identified as high-selling insurance indicators, where M is a positive integer; the high-selling insurance indicators for all suggested insurance types are collectively identified as insurance indicators to be sent; or, The insurance identifiers of the top P insurance policies with the highest sales for each suggested insurance type are identified as insurance identifiers to be sent, where P is a positive integer.

4. The method according to claim 2, characterized in that, The target insurance identifier also includes an identifier for purchased but expired insurance; Accordingly, after removing the insurance type to be removed from the target insurance type to obtain at least one suggested insurance type, the process further includes: Obtain the suggested insurance identifiers corresponding to the suggested insurance types; Obtain the feature parameters corresponding to each suggested insurance identifier; Based on the aforementioned feature parameters, the recommended insurance identifier in the suggested insurance identifier is determined; The purchased and expired insurance identifier and the recommended insurance identifier are jointly identified as the insurance identifier to be sent.

5. The method according to claim 1, characterized in that, The step of determining the target vector based on the insurance identifier of the purchased insurance includes: The type of insurance purchased is determined based on the insurance identifier of the purchased insurance. Determine the target vector based on the type of insurance already purchased.

6. The method according to claim 1, characterized in that, Before determining the target vector based on the insurance identifier of the purchased insurance, the method further includes: Read the insurance information corresponding to each insurance identifier; Obtain the semantic feature vectors of each insurance information; Calculate the similarity between each semantic feature vector; Based on the similarity between the semantic feature vectors, the insurance identifiers are clustered to obtain multiple insurance types; Determine the insurance type vector corresponding to each insurance type.

7. An insurance information processing device, characterized in that, include: The request receiving module is used to receive insurance information retrieval requests sent by user terminals, wherein the insurance information retrieval request includes the user identifier to be processed; The vector determination module is used to determine the target feature vector corresponding to the user identifier to be processed; The type acquisition module is used to input the target feature vector into the insurance recommendation model to obtain the target insurance type, wherein the insurance recommendation model is pre-trained using user features, insurance identifiers of insurance purchased by the user, and the masking vectors corresponding to the insurance identifiers. The identifier acquisition module is used to acquire the target insurance identifier corresponding to the user identifier to be processed, wherein the target insurance identifier includes an insurance identifier that holds insurance. The identifier determination module is used to determine the insurance identifier to be sent based on the target insurance type and the target insurance identifier; The information sending module is used to send the target insurance information corresponding to the insurance identifier to be sent to the user terminal; The insurance information processing device also includes: a model training module; The model training module is used to obtain the user features corresponding to the target user identifier and the insurance identifier of the purchased insurance, where the target user identifier is the user identifier of any user; convert the user features into feature vectors; determine the target vector based on the insurance identifier of the purchased insurance; determine the masking vector based on the target vector; determine the output vector based on the target vector and the masking vector; determine the feature vectors and output vectors corresponding to N target user identifiers as training data, where N is a positive integer; and train the model using the training data to obtain the insurance recommendation model. The model training module is specifically used to obtain the target direction in the target vector, where the target direction is the direction corresponding to the type of insurance not purchased; to determine at least one random direction in the target direction as the occlusion direction; and to generate an occlusion vector based on each occlusion direction.

8. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the insurance information processing method as described in any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the insurance information processing method as described in any one of claims 1 to 5.

10. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the insurance information processing method according to any one of claims 1 to 5.

Images