CN115205049A

CN115205049A - Insurance product pushing method and device

Info

Publication number: CN115205049A
Application number: CN202210832239.XA
Authority: CN
Inventors: 王祎琨; 马子琛; 凌华泽
Original assignee: Bank of China Ltd
Current assignee: Bank of China Ltd
Priority date: 2022-07-15
Filing date: 2022-07-15
Publication date: 2022-10-18

Abstract

The invention provides a method and a device for pushing insurance products, and particularly relates to the field of finance, wherein the method comprises the following steps: based on the historical user feature vectors, the corresponding historical user types and the quantity of the historical user feature vectors of each historical user type, determining the homogeneous local centroid vector of each historical user feature vector; according to a preset target user characteristic vector, obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor quantity of the target user characteristic vector in the historical user characteristic vectors; and determining the target user type of the target user according to the plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the plurality of homogeneous centroid distances between the target user characteristic vector and the homogeneous local centroid vectors corresponding to the nearest neighbor vectors, and pushing insurance products to the target user based on the target user type. The invention can improve the speed and accuracy of pushing the insurance product, thereby improving the efficiency of pushing the insurance product.

Description

Insurance product pushing method and device

Technical Field

The invention relates to the technical field of product pushing, in particular to the field of finance, and particularly relates to an insurance product pushing method and device.

Background

The existing insurance product pushing method mainly comprises manual pushing and model-based pushing. The manual pushing mainly comprises that a worker researches the characteristics of a user and determines the type of the user, so that insurance products are pushed to the user according to the type of the user, but the accuracy of the manual pushing depends on the experience of the worker, the pushing accuracy needs to be improved, a large amount of time is consumed for researching the relevant data of the user in the manual pushing, and the pushing speed is low. The model-based pushing mainly includes that a large amount of historical user data and corresponding user types are used for training a preset model, the trained model is put into insurance product pushing, during actual pushing, data of a current user are often input into the model, the model predicts the type of the current user, then the insurance product pushing is correspondingly carried out on the user based on the type predicted by the model, however, the model-based pushing needs a large amount of time and computing resources to train the model, and therefore the whole pushing time is long, and the speed is slow. In summary, the existing insurance product pushing method has the problems of low pushing speed and poor accuracy, so that the insurance product pushing efficiency is low, and the improvement of income of an insurance supplier and the user experience are not facilitated.

Disclosure of Invention

The invention aims to provide an insurance product pushing method to solve the problem that the pushing efficiency of an insurance product is low due to low pushing speed and poor accuracy of the insurance product in the prior art. Another object of the present invention is to provide an insurance product pushing device. It is a further object of this invention to provide such a computer apparatus. It is a further object of this invention to provide such a readable medium. It is a further object of this invention to provide a computer program product.

In order to achieve the above object, an aspect of the present invention discloses an insurance product pushing method, the method including:

based on the historical user feature vectors, the corresponding historical user types and the quantity of the historical user feature vectors of each historical user type, determining the homogeneous local centroid vector of each historical user feature vector;

according to a preset target user feature vector, obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor quantity of the target user feature vector in the historical user feature vectors;

determining the target user type of the target user according to the plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the plurality of homogeneous centroid distances between the target user characteristic vector and the homogeneous local centroid vectors corresponding to the nearest neighbor vectors, and pushing insurance products to the target user based on the target user type.

Optionally, further comprising:

before determining homogeneous local centroid vectors for each of the historical user feature vectors based on the historical user feature vectors, the corresponding historical user types, and the number of historical user feature vectors for each of the historical user types,

and carrying out feature vectorization processing on the plurality of historical user information to obtain corresponding historical user feature vectors.

Optionally, further comprising:

before obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor quantity of the target user feature vector in the historical user feature vectors according to the preset target user feature vector,

and performing feature vectorization processing on the target user information to obtain the corresponding target user feature vector.

Optionally, the determining a homogeneous local centroid vector of each historical user feature vector based on the historical user feature vector, the corresponding historical user type, and the number of the historical user feature vectors of each historical user type includes:

determining a plurality of similar neighbor vectors of each historical user feature vector based on the historical user feature vectors, the corresponding historical user types and the quantity of the historical user feature vectors of each historical user type;

and respectively determining the similar local centroid vectors of the corresponding historical user feature vectors based on the similar neighbor vectors.

Optionally, the determining, based on the historical user feature vector, the corresponding historical user type, and the quantity of the historical user feature vectors of each historical user type, a plurality of similar neighbor vectors of each historical user feature vector includes:

obtaining the minimum historical user type with the minimum historical user characteristic vector quantity based on the historical user types and the historical user characteristic vector quantity of each historical user type;

judging whether the historical user type is the minimum historical user type, if so, obtaining the quantity of similar neighbor vectors based on the quantity of the historical user characteristic vectors of the minimum historical user type; if not, obtaining the quantity of similar neighbor vectors based on the quantity of the historical user characteristic vectors of the minimum historical user type and the quantity of the historical user characteristic vectors corresponding to the historical user type;

and based on the first Euclidean distance between the historical user feature vector and other historical user feature vectors of the same historical user type, taking the other historical user feature vectors of the same historical user type with the same class neighbor vector quantity with the nearest first Euclidean distance as the similar neighbor vectors of the historical user feature vector.

Optionally, the determining similar local centroid vectors of the corresponding historical user feature vectors based on the similar neighbor vectors respectively includes:

and taking the average value of the similar neighbor vectors as the similar local centroid vector of the corresponding historical user feature vector.

Optionally, the obtaining, according to a preset target user feature vector, a plurality of nearest neighbor vectors corresponding to a preset nearest neighbor number of the target user feature vector in the historical user feature vector includes:

according to the target user feature vector and the plurality of historical user feature vectors, respectively obtaining a second Euclidean distance between the target user feature vector and the historical user feature vectors;

and taking a plurality of historical user feature vectors with the nearest neighbor number of the second Euclidean distance as the nearest neighbor vector of the target user feature vector.

Optionally, the determining the target user type of the target user according to the multiple nearest neighbor distances between the target user feature vector and the nearest neighbor vector and the multiple homogeneous centroid distances between the target user feature vector and the nearest neighbor vector and the homogeneous local centroid vectors corresponding to the target user feature vector and the nearest neighbor vector includes:

determining a plurality of combined neighbor vectors with a preset combined neighbor quantity according to a plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vectors and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user characteristic vector and the nearest neighbor vectors;

and determining the target user type of the target user based on the historical user type corresponding to the combined neighbor vector.

Optionally, the determining, according to a plurality of nearest neighbor distances between the target user feature vector and the nearest neighbor vectors and a plurality of homogeneous centroid distances between the target user feature vector and the nearest neighbor vectors corresponding to the same local centroid vectors, a plurality of combined nearest neighbor vectors of a preset combined nearest neighbor number includes:

obtaining a plurality of combined distances between the target user characteristic vector and the nearest neighbor vector according to a plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user characteristic vector and the nearest neighbor vector;

and determining a plurality of nearest neighbor vectors corresponding to a plurality of the combination distances with the minimum combination neighbor number as the combination neighbor vectors.

Optionally, the determining a target user type of the target user based on the historical user type corresponding to the combined neighbor vector includes:

determining the number of combined neighbor vectors corresponding to the historical user type based on the historical user type corresponding to the combined neighbor vectors;

determining the historical user type with the maximum number of the combined neighbor vectors as the target user type.

In order to achieve the above object, another aspect of the present invention discloses an insurance product pushing device, including:

the local centroid determining module is used for determining the similar local centroid vector of each historical user feature vector based on the historical user feature vector, the corresponding historical user type and the quantity of the historical user feature vectors of each historical user type;

the nearest neighbor determining module is used for obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor quantity of the target user feature vector in the historical user feature vectors according to preset target user feature vectors;

and the pushing module is used for determining the target user type of the target user according to a plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user characteristic vector and the nearest neighbor vector, and pushing an insurance product to the target user based on the target user type.

The invention also discloses a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method when executing the program.

The invention also discloses a computer-readable medium, on which a computer program is stored which, when executed by a processor, implements a method as described above.

The invention also discloses a computer program product comprising a computer program which, when executed by a processor, implements the method as described above.

According to the insurance product pushing method and device, the similar local centroid vector of each historical user characteristic vector is determined based on the historical user characteristic vector, the corresponding historical user type and the quantity of the historical user characteristic vectors of each historical user type, the similar local centroid vector capable of representing the spatial distribution center of each historical user characteristic vector can be determined accurately and quickly based on the vector spatial distribution conditions of different types of historical user characteristics, so that the relevance between the target user type determined in the subsequent step and the actual historical user characteristic vector in the vector spatial distribution aspect can be indirectly improved, and the accuracy of the target user type determined in the subsequent step is further indirectly improved; by obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor number of the target user feature vector in the historical user feature vectors according to the preset target user feature vector, the nearest neighbor vectors closest to the target user feature vector can be quickly determined by a simple processing process, so that the speed of determining the type of the target user based on the nearest neighbor vectors in the subsequent steps is indirectly increased; the target user type of the target user is determined according to the nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the similar local centroid distances between the target user characteristic vector and the similar local centroid vectors corresponding to the nearest neighbor vectors, and an insurance product is pushed to the target user based on the target user type, so that the vector space symmetry constraint relation between the target user characteristic vector and the nearest neighbor vectors is further considered on the basis of fully considering the distance constraint relation between the target user characteristic vector and the nearest neighbor vectors, the target user type can better accord with the comprehensive characteristics of the nearest neighbor vectors, the accuracy of the determined target user type is remarkably improved, and the accuracy of the insurance product pushing based on the target user type is remarkably improved. In conclusion, the insurance product pushing method and device provided by the invention can improve the speed and accuracy of insurance product pushing, thereby improving the insurance product pushing efficiency, and further being beneficial to improving the income of insurance suppliers and the experience of users.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a flow chart of an insurance product pushing method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram illustrating an alternative step of determining homogeneous local centroid vectors according to an embodiment of the present invention;

FIG. 3 is a schematic diagram showing an alternative step of determining homogeneous local centroid vectors according to an embodiment of the present invention;

FIG. 4 is a schematic diagram illustrating an alternative method for determining nearest neighbor vectors according to an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating an alternative step of determining a type of target user according to an embodiment of the present invention;

FIG. 6 is a block diagram of an insurance product pushing device according to an embodiment of the present invention;

FIG. 7 illustrates a schematic block diagram of a computer device suitable for use in implementing embodiments of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.

The terms "first," "second," "8230," "8230," and the like, as used herein, are not intended to be limited to a specific meaning or sequence, nor are they intended to limit the invention, but only to distinguish one element from another or to distinguish one element from another element.

As used herein, the terms "comprising," "including," "having," "containing," and the like are open-ended terms that mean including, but not limited to.

As used herein, "and/or" includes any and all combinations of the described items.

It should be noted that, in the technical solution of the present invention, the acquisition, storage, use, processing, etc. of the data all meet the relevant regulations of the national laws and regulations.

The embodiment of the invention discloses an insurance product pushing method, which specifically comprises the following steps of:

s101: and determining the homogeneous local centroid vector of each historical user feature vector based on the historical user feature vectors, the corresponding historical user types and the quantity of the historical user feature vectors of each historical user type.

S102: and obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor quantity of the target user feature vector in the historical user feature vectors according to the preset target user feature vector.

S103: determining the target user type of the target user according to the plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the plurality of homogeneous centroid distances between the target user characteristic vector and the homogeneous local centroid vectors corresponding to the nearest neighbor vectors, and pushing insurance products to the target user based on the target user type.

For example, the various distances involved in embodiments of the present invention may be, but are not limited to, euclidean distances in nature.

For example, the specific content and number of the historical user types may be determined by those skilled in the art according to practical situations, and the embodiment of the present invention is not limited thereto. For example, the historical user types may include, but are not limited to, "intent to apply", "intent to apply general", and "intent to apply weak", among others. Correspondingly, the specific content of the target user type is the same as the content of one of the historical user types.

For example, the specific implementation manner of pushing the insurance product to the target user based on the type of the target user may be determined by those skilled in the art according to practical situations, and the embodiment of the present invention is not limited thereto. For example, for a target user with a target user type of "strong insurance intention to apply", insurance products are pushed to the target user by a telephone contact mode, for a target user with a target user type of "general insurance intention", insurance products are pushed to the target user by a short message notification and application message push mode, and for a target user with a target user type of "weak insurance intention", insurance products are not pushed to the target user. The specific way of pushing the insurance product to the target user based on the type of the target user can be, but is not limited to, determining the intension of the target user based on the type of the target user, and then correspondingly pushing the insurance product to the target user according to the intension of the target user.

According to the insurance product pushing method and device, the similar local centroid vector of each historical user characteristic vector is determined based on the historical user characteristic vector, the corresponding historical user type and the quantity of the historical user characteristic vectors of each historical user type, the similar local centroid vector capable of representing the spatial distribution center of each historical user characteristic vector can be determined accurately and quickly based on the vector spatial distribution conditions of different types of historical user characteristics, so that the relevance between the target user type determined in the subsequent step and the actual historical user characteristic vector in the vector spatial distribution aspect can be indirectly improved, and the accuracy of the target user type determined in the subsequent step is further indirectly improved; by obtaining a plurality of nearest neighbor vectors corresponding to the preset nearest neighbor number of the target user feature vector in the historical user feature vectors according to the preset target user feature vector, the nearest neighbor vectors closest to the target user feature vector can be quickly determined by a simple processing process, so that the speed of determining the type of the target user based on the nearest neighbor vectors in the subsequent steps is indirectly increased; the target user type of the target user is determined according to the nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the similar local centroid distances between the target user characteristic vector and the similar local centroid vectors corresponding to the nearest neighbor vectors, and an insurance product is pushed to the target user based on the target user type, so that the vector space symmetry constraint relation between the target user characteristic vector and the nearest neighbor vectors is further considered on the basis of fully considering the distance constraint relation between the target user characteristic vector and the nearest neighbor vectors, the target user type can better accord with the comprehensive characteristics of the nearest neighbor vectors, the accuracy of the determined target user type is remarkably improved, and the accuracy of the insurance product pushing based on the target user type is remarkably improved. In summary, the insurance product pushing method and device provided by the invention can improve the speed and accuracy of insurance product pushing, thereby improving the insurance product pushing efficiency, and further being beneficial to improving the income of insurance suppliers and the user experience.

In an optional embodiment, further comprising:

before determining homogeneous local centroid vectors of each historical user feature vector based on the historical user feature vectors, the corresponding historical user types and the quantity of the historical user feature vectors of each historical user type,

For example, the feature vectorization processing is performed on the multiple pieces of historical user information to obtain the corresponding historical user feature vectors, which may be, but not limited to, performing corresponding format adjustment and feature extraction on each piece of attribute information in the historical user information to obtain vector elements corresponding to the attribute information, and then splicing and integrating each vector element to obtain the corresponding historical user feature vector. For attribute information of which the attribute value is a number in the attribute information, when determining the vector element, the attribute value serving as the number may be directly determined as the corresponding vector element, or the attribute value is normalized to obtain the corresponding vector element; for attribute information (for example, characters or characters) whose attribute values are not numbers in the attribute information, when determining vector elements, the attribute information may be digitized to obtain corresponding vectorized elements, and the digitized products may be normalized to obtain corresponding vectorized elements, and the implementation of the digitization may be, but is not limited to, processing using a digitization coding (for example, one-hot coding) algorithm or referring to an ASCII code table. It should be noted that, a specific implementation manner of performing the feature vectorization processing on the multiple pieces of historical user information to obtain the corresponding historical user feature vectors may be determined by those skilled in the art according to actual situations, and the above description is only an example, and is not limited thereto.

Through the steps, the historical user information can be quickly and accurately converted into a vector form which is convenient to participate in operation and processing, so that the operation of subsequent steps in relevant operation and processing is simpler, and the speed of pushing the whole insurance product is effectively improved.

In an optional embodiment, further comprising:

For example, the performing the feature vectorization processing on the target user information to obtain the corresponding target user feature vector may be, but is not limited to, performing corresponding format adjustment and feature extraction on each attribute information in the target user information to obtain vector elements corresponding to the attribute information, and then splicing and integrating each vector element to obtain the corresponding target user feature vector. For attribute information of which the attribute value is a number in the attribute information, when determining the vector element, the attribute value serving as the number may be directly determined as the corresponding vector element, or the attribute value is normalized to obtain the corresponding vector element; for attribute information (for example, characters or characters) whose attribute values are not numbers in the attribute information, in determining vector elements, the attribute information may be digitized to obtain corresponding vectorized elements, and the digitized product may be normalized to obtain corresponding vectorized elements, and the implementation of the digitization may be, but is not limited to, processing using a digital coding (for example, one-hot coding) algorithm or matching an ASCII code table. It should be noted that, a specific implementation manner of performing the feature vectorization processing on the target user information to obtain the corresponding target user feature vector may be determined by those skilled in the art according to actual situations, and the above description is only an example, and does not limit this.

Through the steps, the target user information can be quickly and accurately converted into a vector form which is convenient to participate in operation and processing, so that the operation of subsequent steps in relevant operation and processing is simpler, and the speed of pushing the whole insurance product is effectively improved.

In an alternative embodiment, as shown in fig. 2, the determining the homogeneous local centroid vector of each historical user feature vector based on the historical user feature vector, the corresponding historical user type, and the number of the historical user feature vectors of each historical user type includes the following steps:

s201: and determining a plurality of similar neighbor vectors of each historical user feature vector based on the historical user feature vectors, the corresponding historical user types and the quantity of the historical user feature vectors of each historical user type.

S202: and respectively determining the similar local centroid vectors of the corresponding historical user feature vectors based on the similar neighbor vectors.

Illustratively, the historical user type corresponds to historical user information for generating a historical user feature vector, so the historical user type can directly correspond to the historical user feature vector. The historical user type may be, but is not limited to, determined by the relevant staff member after previous analysis based on the historical user information. It should be noted that the source of the historical user type can be determined by those skilled in the art according to practical situations, and the above description is only an example and is not limiting.

Illustratively, the historical user feature vector includes the following examples:

(1352，22，5，20000)

wherein 1352 corresponds to the name information "Wangzhi" of the historical user, 22 corresponds to the age "22 years" of the historical user, 5 corresponds to the average annual insurance number "5 times" of the historical user, 20000 corresponds to the average annual insurance amount "20000" of the historical user, and the historical user insurance type corresponding to the vector is "strong insurance intention". It should be noted that, the historical user feature vector is only used as an example, and the number of elements, specific values of the elements, attribute information represented by the elements, and the like of the historical user feature vector may be determined by those skilled in the art according to actual situations, which is not limited in this embodiment of the present invention.

Through the steps S201 and S202, the characteristic center which accords with the spatial distribution condition of the historical user characteristic vector relative to the same type of overall vector can be determined to be the same type local centroid vector which can represent the spatial distribution condition of the historical user characteristic vector more accurately based on the same type of neighbor vector of the historical user characteristic vector, so that the accuracy of the determined same type local centroid vector is higher, the relevance of the target user type determined in the subsequent step and the actual historical user characteristic vector in the aspect of vector spatial distribution is more favorably and indirectly improved, and the accuracy of the target user type determined in the subsequent step is more favorably and indirectly improved.

In an optional embodiment, the determining, based on the historical user feature vector, the corresponding historical user type, and the number of historical user feature vectors for each historical user type, a plurality of homogeneous neighbor vectors for each historical user feature vector includes:

obtaining a minimum historical user type with the minimum historical user characteristic vector quantity based on the historical user type and the historical user characteristic vector quantity of each historical user type;

and based on the first Euclidean distance between the historical user characteristic vector and other historical user characteristic vectors of the same historical user type, taking the other historical user characteristic vectors of the same historical user type with the quantity of the similar neighboring vectors with the nearest Euclidean distance as the similar neighboring vectors of the historical user characteristic vector.

For example, the minimum historical user type with the minimum number of the historical user feature vectors is obtained based on the historical user types and the number of the historical user feature vectors of each historical user type, and may be, but is not limited to, determining each historical user type set (each set includes the historical user feature vectors that belong to the type) based on the historical user types and the number of the historical user feature vectors of each historical user type, determining the historical user type set with the minimum number of the included vectors based on the number of the vectors in each historical user type set, and taking the historical user type corresponding to the historical user type set with the minimum number of the included vectors as the minimum historical user type. It should be noted that, for a specific implementation manner of obtaining the minimum historical user type with the minimum historical user feature vector number based on the historical user type and the historical user feature vector number of each historical user type, the implementation manner may be determined by a person skilled in the art according to an actual situation, and the foregoing description is only an example, and does not limit this.

For example, the number of homogeneous neighbor vectors is obtained based on the number of the historical user feature vectors of the minimum historical user type, and may be, but is not limited to, taking an upward rounded value of a square root of the number of the historical user feature vectors of the minimum historical user type as the number of homogeneous neighbor vectors. The number of similar neighbor vectors is obtained based on the number of the historical user feature vectors of the minimum historical user type and the number of the historical user feature vectors corresponding to the historical user type, and may be, but not limited to, a square root of the number of the historical user feature vectors of the minimum historical user type is multiplied by the number of the historical user feature vectors corresponding to the historical user type, and then the number of the historical user feature vectors of the minimum historical user type is divided by the number of the historical user feature vectors of the historical user type to obtain an intermediate similar neighbor quantity value, and then the intermediate similar neighbor quantity value is rounded up to obtain the number of the similar neighbor vectors. The specific implementation manner for determining the number of the similar neighbor vectors can be expressed as the following equation:

wherein k is _tr,j Number of similar neighbor vectors, n, representing historical user feature vector numbered j _min Number of historical user feature vectors, n, representing minimum historical user type _j Representing the number, w, of the historical user feature vectors corresponding to the historical user type corresponding to the historical user feature vector numbered j _j The historical user type, w, corresponding to the historical user feature vector with the number j is shown _min Representing a minimum historical user type.

It should be noted that, for a specific implementation manner for determining the number of homogeneous neighbor vectors, the above description is only an example, and is not limited thereto, which can be determined by those skilled in the art according to practical situations.

For example, determining the first euclidean distance between the vectors is a conventional technique in the art and will not be described herein.

Exemplarily, the other historical user feature vectors of the same historical user type, which are counted by the similar neighbor vector closest to the first euclidean distance, are used as the similar neighbor vectors of the historical user feature vectors, and the following examples are provided:

and a certain historical user type A corresponds to a historical user feature vector A, a historical user feature vector B, a historical user feature vector C, a historical user feature vector D and a historical user feature vector E. And the current historical user feature vector is A, and the similar neighbor vector of A needs to be solved. In this case, it is known that the number of homogeneous neighbor vectors is 2, the first euclidean distance from a to B is 8, the first euclidean distance from a to C is 2, the first euclidean distance from a to D is 3, and the first euclidean distance from a to E is 1. At this time, 2 other historical user feature vectors with the minimum first euclidean distance to a need to be selected from B, C, D and E as similar neighbor vectors of the historical user feature vector a, and therefore, it can be seen that if the two other historical user feature vectors with the minimum first euclidean distance to a are C and E, the similar neighbor vectors of the historical user feature vector a are determined to be C and E.

It should be noted that, a specific implementation manner of using a plurality of other historical user feature vectors of the same historical user type, which are the number of similar neighbor vectors closest to the first euclidean distance, as the similar neighbor vectors of the historical user feature vectors may be determined by those skilled in the art according to actual situations, and the foregoing description is only an example, and does not limit this.

Through the steps, a plurality of similar neighbor vectors of each historical user feature vector can be quickly and accurately determined through simple calculation and simple processing steps, so that the overall pushing speed and accuracy are indirectly improved.

In an optional embodiment, as shown in fig. 3, the determining homogeneous local centroid vectors of the corresponding historical user feature vectors based on the homogeneous neighbor vectors respectively includes the following steps:

s301: and taking the average value of the similar neighbor vectors as the similar local centroid vector of the corresponding historical user characteristic vector.

For example, a specific implementation manner of step S301 may be represented by the following formula:

wherein u is _i The similar local centroid vector of the historical user characteristic vector with the number i is shown, N shows the quantity of the existing historical user characteristic vectors at present, and k _tr,j The number of homogeneous neighbor vectors representing the historical user feature vector numbered i,

and the similar neighbor vector with the number of m represents the historical user feature vector with the number of i.

It should be noted that, for the specific implementation manner of step S301, it can be determined by those skilled in the art according to practical situations, and the above description is only an example, and is not limited thereto.

Through the step S301, the similar local centroid vector can comprehensively represent the vector space symmetry constraint characteristics of a plurality of similar neighbor vectors, so that the accuracy of the similar local centroid vector is improved, and the vector space symmetry constraint relation between the target user characteristic vector and the nearest neighbor vector is accurately considered in the subsequent step when the target user type is determined, so that the accuracy of the determined target user type is improved, and the accuracy of the pushing of the insurance product is improved.

In an optional implementation manner, as shown in fig. 4, the obtaining, according to a preset target user feature vector, a plurality of nearest neighbor vectors corresponding to a preset nearest neighbor number of the target user feature vector in the historical user feature vectors includes the following steps:

s401: and respectively obtaining second Euclidean distances of the target user characteristic vector and the historical user characteristic vectors according to the target user characteristic vector and the plurality of historical user characteristic vectors.

S402: and taking a plurality of historical user feature vectors with the nearest neighbor number of the second Euclidean distance as the nearest neighbor vector of the target user feature vector.

For example, the second euclidean distances of the target user feature vector and the historical user feature vectors obtained according to the target user feature vector and the plurality of historical user feature vectors may be expressed as the following equation:

wherein x represents a target user feature vector, y _i Representing the historical user feature vector with the number i, N representing the number of the existing historical user feature vectors, T representing the transposition operation of the vector (matrix), d (x, y) _i ) And a second Euclidean distance between the target user feature vector and the historical user feature vector with the number i is represented.

Illustratively, the nearest neighbor number may be, but is not limited to, an integer of a product obtained by multiplying the number N of existing historical user feature vectors by a preset nearest neighbor scaling factor p, where the nearest neighbor scaling factor p may be, but is not limited to, 15%, 20%, or 10%, and is preferably 15%. It should be noted that, the specific determination manner of the nearest neighbor number can be determined by those skilled in the art according to practical situations, and the above description is only an example, and is not limited thereto.

Illustratively, the history user feature vectors with the nearest neighbor number of the second euclidean distance are used as the nearest neighbor vector of the target user feature vector, and the following example is given:

the method comprises a historical user feature vector A, a historical user feature vector B, a historical user feature vector C, a historical user feature vector D, a historical user feature vector E, a historical user feature vector F, a historical user feature vector G and a historical user feature vector H, wherein the Euclidean distance between a target user feature vector and A is 1, the Euclidean distance between a target user feature vector and B is 2, the second Euclidean distance between a target user feature vector and C is 3, the second Euclidean distance between a target user feature vector and D is 4, the second Euclidean distance between a target user feature vector and E is 5, the second Euclidean distance between a target user feature vector and F is 6, the second Euclidean distance between a target user feature vector and G is 7, the second Euclidean distance between a target user feature vector and H is 8, and the known nearest neighbor number is 4. Therefore, it is necessary to select 4 historical user feature vectors with the minimum second euclidean distance to the target user feature vector from a, B, C, D, E, F, G, and H as the nearest neighbor vectors of the target user feature vector, and thus, if these 4 historical user feature vectors with the minimum second euclidean distance to the target user feature vector are a, B, C, and D, the nearest neighbor vectors of the target user feature vector are determined to be a, B, C, and D.

It should be noted that, a specific implementation manner of using the plurality of history user feature vectors with the nearest neighbor number to the second euclidean distance as the nearest neighbor vector of the target user feature vector may be determined by those skilled in the art according to actual situations, and the above description is only an example, and is not limited thereto.

Through the above steps, the feature proximity degree of the target user feature vector and the feature proximity degree of the different history user feature vectors can be determined more accurately based on the standard euclidean distance, and the nearest neighbor vector closest to the feature of the target user feature vector can be determined more accurately. Therefore, the accuracy and the speed of determining the nearest neighbor vector can be improved through the steps, and the accuracy and the speed of pushing the whole insurance product are indirectly improved.

In an alternative embodiment, as shown in fig. 5, the determining a target user type of a target user according to a plurality of nearest neighbor distances between the target user feature vector and the nearest neighbor vector and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user feature vector and the nearest neighbor vector includes the following steps:

s501: and determining a plurality of combined neighbor vectors with preset combined neighbor quantity according to the plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user characteristic vector and the nearest neighbor vector.

S502: and determining the target user type of the target user based on the historical user type corresponding to the combined neighbor vector.

Illustratively, since the nearest neighbor vectors belong to the historical user feature vectors, and the second euclidean distance of the target user feature vector from each historical user feature vector is known in the previous step, a plurality of nearest neighbor distances of the target user feature vector from the nearest neighbor vectors (the second euclidean distance of the target user feature vector from the nearest neighbor vectors) are also known at this time.

Illustratively, since the nearest neighbor vector belongs to the historical user feature vector, and the homogeneous local centroid vector of the historical user feature vector is known in the previous step, the homogeneous local centroid vector corresponding to the nearest neighbor vector is also known at this time.

Illustratively, the plurality of homogeneous centroid distances of the homogeneous local centroid vector corresponding to the target user feature vector and the nearest neighbor vector are euclidean distances of the target user feature vector and the plurality of homogeneous local centroid vectors, and the determination of the homogeneous centroid distances may be represented by the following equation:

wherein,

the similar local centroid vector corresponding to the target user characteristic vector x and the nearest neighbor vector with the number of i

The Euclidean distance (homogeneous centroid distance) of the user is obtained, N represents the number of the existing historical user feature vectors, and p represents a preset nearest neighbor proportion coefficient.

Through the steps S501 and S502, the distance constraint relation and the vector space symmetry constraint relation between the target user feature vector and the nearest neighbor vector can be comprehensively considered to a deeper degree, so that the type of the target user can be determined more accurately on the basis of the distribution characteristics of the target user feature vector and the nearest neighbor vector, and the pushing accuracy of the insurance product is improved.

In an optional embodiment, the determining, according to a plurality of nearest neighbor distances between the target user feature vector and the nearest neighbor vector and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user feature vector and the nearest neighbor vector, a plurality of combined neighbor vectors of a preset combined neighbor number includes:

For example, the deriving a plurality of combined distances of the target user feature vector and the nearest neighbor vector according to a plurality of nearest neighbor distances of the target user feature vector and the nearest neighbor vector and a plurality of homogeneous centroid distances of homogeneous local centroid vectors corresponding to the target user feature vector and the nearest neighbor vector may be represented by, but not limited to, the following equation:

wherein,

representing the target user's feature vector x and its nearest neighbor vector with number i

The combined distance of (a) to (b),

representing the target user's feature vector x and its nearest number iSimilar local centroid vector corresponding to adjacent vector

Euclidean distance (homogeneous centroid distance),

and the nearest neighbor distance between the target user characteristic vector and the nearest neighbor with the number of i is represented, N represents the number of the existing historical user characteristic vectors, and p represents a preset nearest neighbor proportionality coefficient.

It should be noted that, for a specific implementation manner of obtaining a plurality of combined distances between the feature vector of the target user and the nearest neighbor vector according to a plurality of nearest neighbor distances between the feature vector of the target user and the nearest neighbor vector and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the feature vector of the target user and the nearest neighbor vector, the implementation manner may be determined by those skilled in the art according to actual situations, and the above description is only an example, and does not limit the present invention.

Illustratively, the number of combined neighbors, which is a preset number, may be, but is not limited to, an integer in the range of [1, 20], preferably 10. It should be noted that the specific value of the number of the combined neighbors can be determined by those skilled in the art according to practical situations, and the above description is only an example and is not limiting.

For example, the determining a plurality of nearest neighbor vectors corresponding to a plurality of the combined distances with the smallest number of combined neighbors as the combined neighbor vector may include:

a nearest neighbor vector a, a nearest neighbor vector B, a nearest neighbor vector C, a nearest neighbor vector D, and a nearest neighbor vector E have been determined at present, and the combined distance of the target user feature vector and a is 20, the combined distance of the target user feature vector and B is 16, the combined distance of the target user feature vector and C is 12, the combined distance of the target user feature vector and D is 8, the combined distance of the target user feature vector and E is 4, and at this time, the known combined nearest neighbor number is 3. It is thus understood that 3 nearest neighbor vectors having the smallest combined distance to the target user feature vector need to be selected from a, B, C, D, and E as the combined nearest neighbor vector of the target user feature vector, and that if the 3 nearest neighbor vectors having the smallest combined distance to the target user feature vector are C, D, and E, the combined nearest neighbor vector of the target user feature vector is determined to be C, D, and E.

It should be noted that, for a specific implementation manner of determining a plurality of nearest neighbor vectors corresponding to a plurality of combination distances with the smallest number of combination neighbors as the combination nearest neighbor vector, the determination may be determined by a person skilled in the art according to actual situations, and the above description is only an example, and does not limit the present invention.

Through the steps, the obtained combined neighbor vector can fully and accurately represent the distance constraint relation and the vector space symmetry constraint relation between the characteristic vector and the nearest neighbor vector of the target user, so that the obtained combined neighbor vector has higher accuracy and authenticity, the accuracy of the type of the target user determined based on the combined neighbor vector is improved, and the accuracy of the pushing of the insurance product is improved. In addition, the step of determining the combined neighbor vector does not need complicated execution steps and is simple in calculation, so that the speed of determining the combined neighbor vector in the step is high, the speed of determining the type of the target user can be increased, and the speed of pushing the insurance product is increased.

In an optional embodiment, the determining a target user type of a target user based on a historical user type corresponding to the combined neighbor vector includes:

and determining the historical user type with the maximum number of the combined neighbor vectors as the target user type.

For example, since the combined neighbor vector belongs to the historical user feature vector, and the historical user type corresponding to the historical user feature vector is known, the historical user type corresponding to the combined neighbor vector can be directly determined.

For example, since the corresponding relationship between the combined neighbor vector and the historical user type is known, the number of combined neighbor vectors corresponding to the historical user type can be determined directly based on the historical user type corresponding to the combined neighbor vector.

Illustratively, the historical user type with the largest number of combined neighbor vectors is determined as the target user type, and the following example is given:

knowing that the number of the combined neighbors corresponding to the historical user type A is 3, the number of the combined neighbors corresponding to the historical user type B is 2 and the number of the combined neighbors corresponding to the historical user type C is 1, determining that the historical user type with the maximum number of the combined neighbor vectors is A, and determining the A as the target user type.

It should be noted that, for a specific implementation manner of determining the historical user type with the largest number of combined neighbor vectors as the target user type, the determination may be performed by a person skilled in the art according to practical situations, and the above description is only an example, and does not limit this.

Through the steps, the historical user type with the maximum combined neighbor vector quantity can be determined as the target user type according to the majority voting principle, and the historical user type with the maximum combined neighbor vector quantity can always represent the target user type sufficiently and accurately.

Based on the same principle, the embodiment of the present invention discloses an insurance product pushing device 600, as shown in fig. 6, the insurance product pushing device 600 includes:

a local centroid determining module 601, configured to determine a similar local centroid vector of each historical user feature vector based on the historical user feature vector, a corresponding historical user type, and the number of historical user feature vectors of each historical user type;

a nearest neighbor determining module 602, configured to obtain, according to a preset target user feature vector, multiple nearest neighbor vectors corresponding to a preset nearest neighbor number of the target user feature vector in the historical user feature vectors;

the pushing module 603 is configured to determine a target user type of the target user according to the multiple nearest neighbor distances between the target user feature vector and the nearest neighbor vector and the multiple homogeneous centroid distances between the target user feature vector and the homogeneous local centroid vectors corresponding to the nearest neighbor vectors, and push an insurance product to the target user based on the target user type.

In an optional embodiment, the history information vectorization module is further configured to:

and performing feature vectorization processing on the plurality of historical user information to obtain corresponding historical user feature vectors.

In an optional embodiment, the system further comprises a current information vectorization module, configured to:

and carrying out feature vectorization processing on the target user information to obtain the corresponding target user feature vector.

In an optional embodiment, the local centroid determining module 601 is configured to:

and taking the average value of the similar neighbor vectors as the similar local centroid vector of the corresponding historical user characteristic vector.

In an optional embodiment, the nearest neighbor determination module 602 is configured to:

according to the target user characteristic vector and the plurality of historical user characteristic vectors, second Euclidean distances of the target user characteristic vector and the historical user characteristic vectors are respectively obtained;

and taking a plurality of historical user feature vectors with the nearest neighbor number of a second Euclidean distance as the nearest neighbor vector of the target user feature vector.

In an optional embodiment, the pushing module 603 is configured to:

determining a plurality of combined neighbor vectors with preset combined neighbor quantity according to a plurality of nearest neighbor distances between the target user feature vector and the nearest neighbor vectors and a plurality of homogeneous centroid distances between homogeneous local centroid vectors corresponding to the target user feature vector and the nearest neighbor vectors;

In an optional embodiment, the pushing module 603 is configured to:

and determining a plurality of nearest neighbor vectors corresponding to a plurality of combination distances with the minimum combination nearest number as the combination nearest vector.

In an optional embodiment, the pushing module 603 is configured to:

Since the principle of solving the problem of the insurance product pushing device 600 is similar to the above method, the implementation of the insurance product pushing device 600 can refer to the implementation of the above method, and will not be described herein again.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. A typical implementation device is a computer device, which may be, for example, a personal computer, laptop computer, cellular telephone, camera phone, smart phone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or a combination of any of these devices.

In a typical example, the computer device comprises in particular a memory, a processor and a computer program stored on the memory and executable on the processor, which when executed by the processor implements the method as described above.

Referring now to FIG. 7, shown is a block diagram of a computer device 700 suitable for use in implementing embodiments of the present application.

As shown in fig. 7, the computer device 700 includes a Central Processing Unit (CPU) 701, which can perform various appropriate works and processes according to a program stored in a Read Only Memory (ROM) 702 or a program loaded from a storage section 708 into a Random Access Memory (RAM) 703. In the RAM703, various programs and data necessary for the operation of the system 700 are also stored. The CPU701, ROM702, and RAM703 are connected to each other via a bus 704. An input/output (I/O) interface 705 is also connected to bus 704.

The following components are connected to the I/O interface 705: an input portion 706 including a keyboard, a mouse, and the like; an output section 707 including components such as a Cathode Ray Tube (CRT), a liquid crystal feedback (LCD), and a speaker; a storage section 708 including a hard disk and the like; and a communication section 709 including a network interface card such as a LAN card, a modem, or the like. The communication section 709 performs communication processing via a network such as the internet. A drive 710 is also connected to the I/O interface 705 as needed. A removable medium 711 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 710 as necessary, so that the computer program read out therefrom is mounted as necessary in the storage section 708.

In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the invention include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication section 709, and/or installed from the removable medium 711.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the various elements may be implemented in the same one or more pieces of software and/or hardware in the practice of the present application.

The present invention has been described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of other like elements in a process, method, article, or apparatus comprising the element.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.

The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. An insurance product pushing method, comprising:

2. The method of claim 1, further comprising:

3. The method of claim 1, further comprising:

4. The method of claim 1, wherein determining the homogeneous local centroid vector for each historical user feature vector based on the historical user feature vector, the corresponding historical user type, and the number of historical user feature vectors for each historical user type comprises:

and respectively determining similar local centroid vectors of the corresponding historical user characteristic vectors based on the similar neighbor vectors.

5. The method according to claim 4, wherein the determining a plurality of homogeneous neighbor vectors for each historical user feature vector based on the historical user feature vector, the corresponding historical user type, and the number of historical user feature vectors for each historical user type comprises:

6. The method according to claim 4, wherein the determining homogeneous local centroid vectors of the corresponding historical user feature vectors based on the homogeneous neighbor vectors respectively comprises:

7. The method according to claim 1, wherein the obtaining a plurality of nearest neighbor vectors corresponding to a preset nearest neighbor number of the target user feature vector from the historical user feature vectors according to a preset target user feature vector comprises:

8. The method according to claim 1, wherein the determining a target user type of a target user according to a plurality of nearest neighbor distances of the target user feature vector and the nearest neighbor vectors and a plurality of homogeneous centroid distances of homogeneous local centroid vectors corresponding to the target user feature vector and the nearest neighbor vectors comprises:

9. The method according to claim 8, wherein the determining a plurality of combined neighbor vectors of a preset number of combined neighbors according to a plurality of nearest neighbor distances of the target user feature vector to the nearest neighbor vectors and a plurality of homogeneous centroid distances of homogeneous local centroid vectors corresponding to the target user feature vector to the nearest neighbor vectors comprises:

10. The method of claim 8, wherein the determining a target user type of a target user based on a historical user type corresponding to the combined neighbor vector comprises:

11. An insurance product pusher apparatus, comprising:

and the pushing module is used for determining the target user type of the target user according to the plurality of nearest neighbor distances between the target user characteristic vector and the nearest neighbor vector and the plurality of homogeneous mass center distances between the target user characteristic vector and the homogeneous local mass center vector corresponding to the nearest neighbor vector, and pushing the insurance product to the target user based on the target user type.

12. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1-10 when executing the program.

13. A computer-readable medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-10.

14. A computer program product, characterized in that the computer program product comprises a computer program which, when being executed by a processor, carries out the method of any one of claims 1 to 10.