CN114998042A - Product recommendation processing method, device and equipment and computer-readable storage medium - Google Patents

Abstract

The present application belongs to the technical field of data processing, and provides a product recommendation processing method, device, computer equipment and computer-readable storage medium. Multi-layer graph model, preset product recommendation The multi-layer graph model contains at least three layers, each layer contains several nodes, and there is a business relationship between adjacent layers, including a user layer, and the user layer The layer contains only one node, and then according to all nodes, based on the recommendation algorithm of PersonalRank, the recommendation probability of each node is obtained, and then according to the recommendation probability, and based on the preset probability usage method, use the corresponding recommendation probability of different types of nodes to produce products. Recommendation can improve search efficiency, and implement multi-layer entity recommendation based on nodes of multiple layers, which improves the recommendation efficiency, diversity and accuracy of product recommendation.

Description

Product recommendation processing method, apparatus, device and computer-readable storage medium

technical field

The present application relates to the technical field of data processing, and in particular, to a product recommendation processing method, apparatus, computer device, and computer-readable storage medium.

Background technique

With the development of information technology and the Internet, the insurance industry has ushered in new opportunities and challenges. While the marketing channels of insurance products are diversified, the digital transformation of insurance product marketing also needs to keep up with the changes of the times. In traditional technology, when the insurance industry uses digital channels to market insurance products, it generally uses machine learning technology to support insurance product recommendation, but this insurance product recommendation method can only implement the recommendation of one entity of insurance products. For example, for For the recommendation of entities such as insurance products, sales channels or insurance categories, if you want to recommend multiple entities of insurance products, you need to use models corresponding to multiple entities to perform calculations, which reduces the efficiency of insurance entity recommendation.

SUMMARY OF THE INVENTION

The present application provides a product recommendation processing method, apparatus, computer equipment and computer-readable storage medium, which can solve the technical problem of low processing efficiency of product recommendation in the traditional technology.

In a first aspect, the present application provides a product recommendation processing method, including: acquiring a user ID of a target user, and obtaining a preset product recommendation multi-layer graph model corresponding to the user ID according to the user ID, The preset product recommendation multi-layer graph model includes at least three layers, each layer includes several nodes, and there is a business relationship between adjacent layers, including a user layer, and the user layer only A node is included as a user node, and the user node is used to describe the user identifier; according to all the nodes, a recommendation algorithm based on PersonalRank is used to obtain the recommendation probability of each of the nodes; Set the probability usage mode, and use the recommendation probability corresponding to different types of nodes to recommend products.

In a second aspect, the present application further provides a product recommendation processing device, including: a first acquiring unit, configured to acquire a user identifier of a target user, and acquire a preset corresponding to the user identifier according to the user identifier Product recommendation multi-layer graph model, the preset product recommendation multi-layer graph model includes at least three layers, each layer includes several nodes, and there is a business relationship between adjacent layers, which includes a user layer, and the user layer only includes one node as a user node, and the user node is used to describe the user identifier; the second obtaining unit is used to obtain each The recommendation probability of the node; the probability using unit is used for recommending products using the recommendation probability corresponding to different types of nodes according to the recommendation probability and based on the preset probability usage mode.

In a third aspect, the present application also provides a computer device, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the product recommendation processing method when the processor executes the computer program .

In a fourth aspect, the present application further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the processing of the product recommendation steps of the method.

The present application provides a product recommendation processing method, apparatus, computer device, and computer-readable storage medium. The processing method obtains the user ID of the target user, and according to the user ID, obtains a preset product recommendation multi-layer graph model corresponding to the user ID, and the preset product recommendation multi-layer graph model includes at least three layers, each layer. It contains several nodes, and there is a business relationship between adjacent layers, including a user layer, and the user layer contains only one node, and then according to all nodes, the recommendation algorithm based on PersonalRank is used to obtain the recommendation probability of each node , and then according to the recommendation probability and based on the preset probability usage mode, use the respective recommendation probability corresponding to different types of nodes to perform product recommendation, because the preset product recommendation multi-layer graph model, compared with the PersonalRank model, the structure is reduced. The storage space occupied by the model is small, which can improve the search efficiency, and the multi-layer entity recommendation can be realized based on the nodes of multiple layers, thereby improving the diversity, accuracy and recommendation efficiency of product recommendation.

Description of drawings

In order to explain the technical solutions of the embodiments of the present application more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. For those of ordinary skill, other drawings can also be obtained from these drawings without any creative effort.

1 is a schematic flowchart of a processing method for product recommendation provided by an embodiment of the present application;

FIG. 2 is an example of a model structure of a multi-layer graph model of preset product recommendation involved when the processing method for product recommendation is used in an insurance product other than auto insurance provided by an embodiment of the present application;

3 is a schematic diagram of the first sub-flow of the processing method for product recommendation provided by the embodiment of the present application;

FIG. 4 is a second sub-flow schematic diagram of the processing method for product recommendation provided by the embodiment of the present application;

5 is a schematic diagram of a third sub-flow of the processing method for product recommendation provided by the embodiment of the present application;

6 is a schematic diagram of the fourth sub-flow of the processing method for product recommendation provided by the embodiment of the present application;

FIG. 7 is a schematic block diagram of a processing device for product recommendation provided by an embodiment of the present application;

FIG. 8 is a schematic block diagram of a computer device according to an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

It is to be understood that, when used in this specification and the appended claims, the terms "comprising" and "comprising" indicate the presence of the described features, integers, steps, operations, elements and/or components, but do not exclude one or The presence or addition of a number of other features, integers, steps, operations, elements, components, and/or sets thereof.

The embodiment of the present application provides a processing method for product recommendation. The processing method can be applied to computer equipment such as servers, and used in business scenarios such as non-auto insurance products, online movie tickets, or e-commerce shopping in the insurance industry. Product recommendation with multi-layer business nodes. Please refer to FIG. 1 , which is a schematic flowchart of a processing method for product recommendation provided by an embodiment of the present application. As shown in Figure 1, the method includes the following steps S11-S13:

S11. Acquire a user ID of a target user, and according to the user ID, obtain a preset product recommendation multi-layer graph model corresponding to the user ID, where the preset product recommendation multi-layer graph model includes at least three layers, Each layer contains several nodes, and there is a business relationship between adjacent layers, including a user layer, and the user layer only contains one node as a user node, and the user node is used to describe the User ID.

Specifically, in application scenarios such as non-auto insurance products (referred to as non-car products), online movie ticket sales, or e-commerce shopping in the insurance industry, for example, if it is detected that an old user logs in to an online website or APP, the The old user can be regarded as the target user, or when the insurance agent clicks the preset evaluation button to know the product recommendation method for a certain target insured user, the target insured user can be regarded as the target user, and it is considered that the product recommendation has been activated. The activation instruction, in response to the activation instruction for product recommendation, obtains the user ID of the target user, and the user ID can be the content that uniquely identifies the user, such as the user's login account, user name, ID number or mobile phone number.

To recommend products with multi-layer business nodes, build a preset product recommendation multi-layer graph model applicable to the product recommendation in advance according to business needs. The preset product recommendation multi-layer graph model may be a graph (algorithm)-based model Model (Graph-Based Model in English), the preset product recommended multi-layer graph model contains at least three layers, each layer is used to describe the same type of entity, each layer contains several nodes, each node Describes an entity, and different nodes of each layer describe different entities belonging to the same category. The entities are used to describe discrete objects existing in the business. For example, in non-auto insurance products, entities can be user names, insurance product names, Objects such as insurance category name, sales channel name, etc., there is a business relationship between adjacent layers, and the relationship is the correlation between the entities corresponding to the nodes. For example, it can be that a user has purchased a product. , or a product can be sold based on a certain sales channel and other business dependencies, the preset product recommendation multi-layer graph model includes a user layer, and the user layer only includes one node as a user node, and the user node Used to describe the user ID. Especially for the application scenario of product recommendation based on historical transaction data, the preset multi-layer graph model for product recommendation may also include a historical transaction data layer, and the nodes of the historical transaction data layer are used to describe historical transaction data. The data can include the identifiers corresponding to the products or services that the target user has purchased. The products can be tangible commodities such as books and smartphones, and the services can be intangible commodities such as insurance services, financial services, and communication services. , which can involve the product name, product type, APP or website and other related entities such as online channels or offline channels of insurance products. Based on historical transaction data such as products or services that users have purchased, statistics on product recommendations can be made to users. It can more accurately predict user preferences, thereby improving the accuracy of product recommendations and the effectiveness of marketing resources. For example, please refer to FIG. 2 . FIG. 2 is an example of a model structure of a multi-layer graph model for preset product recommendation involved in applying a processing method for product recommendation in an insurance product other than auto insurance provided by an embodiment of the present application, as shown in FIG. 2 . , in this non-auto insurance product example, the built preset product recommendation multi-layer graph model may include 4 layers, and there is a business relationship between adjacent layers, and the relationship may be the front and back of the business node. The sequential relationship, or the correlation of the product elements involved in the business, are:

1) The first layer (that is, the starting layer of the graph structure) is generally the user layer. The nodes of the user layer are used to describe the recommended user (ie the target user), and the user layer only contains one node as the user node, The user node adopts the user ID to describe the target user of the recommended product;

2) The second layer is the purchased insurance product layer, which is used to describe the insurance products that the recommended user has purchased. Each node describes an insurance product that the recommended user has purchased, and the node corresponding to the purchased insurance product 1. Describe the insurance product 1 that the recommended user has purchased, and the node corresponding to the purchased insurance product 2 describes the insurance product 2 that the recommended user has purchased;

3) The third layer is the sales channel layer, which is used to describe the sales channels of each purchased insurance product in the second layer. Therefore, the second layer and the third layer are adjacent layers. There is a business relationship between them;

4) The fourth layer is the insurance category layer, which is used to describe the set of insurance categories sold by each sales channel in the third layer. Therefore, the third layer and the fourth layer are related to each other. There is a business relationship between adjacent layers.

Among them, the third layer and the fourth layer can be used for product recommendation, the third layer is used to filter out the sales channels for recommending insurance products to user 1, and the fourth layer is used to filter out the User 1 selects the insurance categories recommended by insurance products, and then can filter out the insurance categories based on the fourth layer, and then filter out the specific insurance products included in this insurance category as the recommended insurance products, as shown in Figure 2 In the example of , if the Chuangzhan channel is the most important among all the sales channels related to the user, it can be considered that the user may prefer the Chuangzhan channel as a sales channel, and he can use the Chuangzhan channel for insurance product marketing in the future. In order to improve the accuracy and effectiveness of sales channel recommendations, and then improve the accuracy and adaptability of insurance products to target users, so as to improve the marketing effect of insurance products. At the same time, as shown in Figure 2, after the fourth layer, a fifth layer can be constructed. The fifth layer can be an insurance product layer, which is used to describe the insurance products included in each insurance category. Set, according to the fifth layer, the specific insurance products included in the insurance category can be filtered out as the recommended insurance products, wherein the edge connections between the nodes in the Figure 2 can be undirected connections, or For the directed connection, the directed connection is used to describe the directed correlation between nodes.

Based on the above-mentioned preset product recommendation multi-layer graph model, for each target user, user target data can be combined, the user target data can include user historical data, and the user historical data can include historical transaction data and historical search data. Behavioral data can also include data generated by different processes such as original historical data and incremental data. Among them, historical search data is the behavioral data that users searched for in the past, such as when a user searches for a certain product in an e-commerce APP The search data for products but not purchased, the historical transaction data may include the identification and sales channels corresponding to the products or services purchased by the target user, and the historical transaction time, so that it can be combined with user target data, such as historical transaction data. The user target data is filled when needed, or the user target data is filled into each node included in the preset product recommendation multi-layer graph model in time when the data is generated, so as to construct the target user's preset product recommendation multi-layer graph model, A multi-layer graph model of preset product recommendations for specific users can be obtained. Please continue to refer to FIG. 2. As shown in FIG. 2, in this example, for user 1, the historical data of the product purchased by user 1 can be described as follows through the structure of the diagram in FIG. 2:

1) The insurance products that the user has purchased, including the purchased insurance product 1 and the purchased insurance product 2;

2) Purchased insurance product 1 can be sold through sales channel 1, and purchased insurance product 2 can be sold through sales channel 1 and sales channel 2;

3) The insurance categories sold by sales channel 1 include insurance category 1 and insurance category 3, and the insurance categories sold by sales channel 2 include insurance category 2 and insurance category 3.

Meanwhile, optionally, the set of insurance products included in insurance category 1 may be insurance product set 1, insurance product set 1 belongs to several insurance products of insurance category 1, and the set of insurance products included in insurance category 2 may be Insurance product set 2. Insurance product set 2 belongs to several insurance products of insurance category 2. The set of insurance products included in insurance category 3 can be insurance product set 3. Insurance product set 3 includes several insurance products belonging to insurance category 3. Insurance Products.

Based on the above-mentioned preset product recommendation multi-layer graph model constructed, after obtaining the user ID of the target user, according to the user ID, obtain the preset product recommendation multi-layer graph model corresponding to the user ID, based on the preset product recommendation multi-layer graph model. Nodes included in the product recommendation multi-layer graph model, the preset product recommendation multi-layer graph model describes the relevant entities and associations included in the user target data, and multiple entities involve multi-layer business nodes and/or business factors. According to the preset product recommendation multi-layer graph model, the recommendation probability of each entity in the preset product recommendation multi-layer graph model is counted, and based on the recommendation probability of each entity, a product recommendation is made to the target user.

S12. According to all the nodes, the recommendation algorithm based on PersonalRank is used to obtain the recommendation probability of each of the nodes.

Specifically, for the above-mentioned preset product recommendation multi-layer graph model, a recommendation algorithm based on PersonalRank, that is, recommending nodes of the multi-layer graph model according to the preset product, if there is a correlation between different nodes, the associated node Connecting, if the user has purchased a certain product (that is, there is a purchase correlation between the user and the product), the user and the respective nodes of the product are connected. For example, please continue to refer to Figure 2. As shown in Figure 2, for user 1, according to the purchased insurance product node generated by the purchase of insurance products, based on the business logic relationship between adjacent layers, the purchased insurance The product node can be associated with the sales channel node and the insurance category node, and connect multiple nodes with business associations, so that the path can be obtained: user 1 - purchased insurance product 1 - sales channel 1 - insurance category 1; User 1 - Purchased insurance product 1 - Sales channel 1 - Insurance category 3; User 2 - Purchased insurance product 2 - Sales channel 1 - Insurance category 3; User 1 - Purchased insurance product 2 - Sales channel 2 -Insurance category 2 and other paths, and then according to the connected paths, the user node in the first layer can go farthest to the node in the last layer by walking, and the importance of all nodes to the user after multiple iterations It will converge to a certain value, which is used to describe the importance of each node, and then sort according to the importance of each node to obtain the node set required for product recommendation. Among them, the correlation degree of the vertices (ie nodes) in the graph mainly depends on the following factors: 1) the number of paths between two vertices; 2) the length of the path between the two vertices; 3) the vertices that the path between the two vertices passes through. Vertices with high correlation generally have the following characteristics: 1) There are many paths connecting the two vertices; 2) The length of the path connecting the two vertices is relatively short; 3) The path connecting the two vertices does not pass the out-degree larger vertices.

Please continue to refer to Figure 2. As shown in Figure 2, for personalized recommendation for user 1, start walking from the node corresponding to user 1 in the figure. When walking to a node, first decide whether to continue walking according to a preset probability , or stop the walk and start the walk again from the node corresponding to user 1. If you decide to continue to walk, then randomly select a node from the nodes pointed to by the current node as the next node according to a uniform distribution, so that after many walks, the probability of each node being visited will converge to A number, the weight of each node in the final recommendation list is the access probability of the node, that is, the recommendation probability of each node, so as to obtain the correlation between the user 1 node and all other nodes. For products with connected edges, a recommendation list is generated according to the level of relevance. In the embodiment of the present application, a multi-layer graph model of preset product recommendation unique to each user is constructed, and only one node is reserved in the user layer, so that each search and iteration during product recommendation are performed only between necessary nodes. Compared with the traditional PersonalRank algorithm structure that includes multiple user nodes and runs among all user nodes, the embodiment of the present application can reduce the overall running time complexity and memory usage of the model, improve search efficiency, and at the same time recommend preset products. The layers contained in the multi-layer graph model are expanded to include at least three layers, and the next layer nodes are determined through the association relationship between adjacent layers, and so on, until the preset product recommends multi-layer graph model construction and search Completed, can use the association relationship between multiple layers, can handle the association analysis between complex and diverse entities, and achieve multi-layer recommendation effects. Compared with the traditional PersonalRank algorithm structure, which only contains two layers, only one can be achieved. Entity recommendation, the processing method for product recommendation provided by the embodiment of the present application can realize the recommendation of multi-layer entities, and thus is more applicable to the business scenario of multi-layer entity recommendation, thereby not only realizing the structure of the multi-layer graph model for preset product recommendation Smaller and smaller storage space can improve the search efficiency of product recommendation, and realize the recommendation of multi-layer entities, which can improve the diversity and accuracy of product recommendation.

S13. According to the recommendation probability and based on the preset probability usage mode, use the recommendation probability corresponding to each node of different types to perform product recommendation.

Specifically, according to the recommendation probability, multiple recommendation probabilities that satisfy the preset probability conditions can be obtained as target probabilities, and the multiple target probabilities are respectively the recommendation probabilities corresponding to different types of nodes, and the preset probability is used based on the usage mode of the preset probability. describe the target probability for product recommendation.

Further, based on the preset probability usage mode, the recommendation probabilities corresponding to different types of nodes are used for product recommendation, including:

Based on the preset probability screening conditions, the respective recommended probabilities of different types of nodes are selected as the target probability, the node corresponding to the target probability is taken as the target node, and the recommended product is displayed on the preset terminal according to the target node;

Alternatively, based on preset probability screening conditions, the respective recommendation probabilities of different types of nodes are screened as target probabilities, and the target probabilities are displayed on the preset terminal, so that relevant personnel can recommend products according to the recommended probabilities.

Specifically, after obtaining the recommendation probability of each node, based on the preset probability screening conditions, the respective recommendation probabilities of different types of nodes are screened as the target probability. The recommendation probability is used as the target probability. For example, according to the correlation of the recommendation probability from high to low, the recommendation probability with high correlation is determined as the target probability, so as to screen out the respective recommendation probability of different types of nodes as the target probability, so as to realize the multi-entities. Recommendation. In product recommendation, the maximum recommendation probability in each layer is generally used as the target probability. For example, in the insurance field, when marketing non-auto insurance products, for the application scenario where the recommendation probability is used for offline marketing as an insurance agent, multiple entities involved in non-auto insurance products and the corresponding target probability can be displayed to the terminal. In order for the insurance agent to take the entities of each node according to the target probability, determine the optimal non-auto insurance product marketing method for offline marketing of non-auto insurance products. For example, please continue to refer to Fig. 2, for user 1, the specific content described by several sales channels and each insurance category and their recommendation probability can be displayed, so that the insurance agent can choose a suitable product for user 1 according to the recommendation probability. Sales channels and products in the insurance category are used as marketing tools for non-auto insurance products. In the application scenarios of online sales such as e-commerce or movie tickets, based on the preset probability screening conditions, the recommendation probability of different types of nodes is selected as the target probability, and the node corresponding to the target probability is used as the target node, especially the product map. For the products in the layer that the user has not purchased (that is, the products that are not directly connected to the user in the graph structure), the recommended product list is generated according to the level of correlation (that is, the recommendation probability is from large to small), and the products that have been purchased by the user are generated. The same or similar recommended product list is displayed, and the recommended product is displayed on the preset terminal, so that the recommended product is recommended to the user, and the recommendation of multiple entities can also be implemented.

In this embodiment of the present application, by retaining only one node in the user layer describing the user ID, and extending the layer to include at least three layers, a corresponding preset product recommendation multi-layer graph model is constructed for each user, Compared with the traditional PersonalRank algorithm structure, which contains multiple user nodes, the behavior generated by all users is stored in only one complex graph structure. Each iteration must iterate the entire graph structure, which takes a long time and requires memory usage. The graph structure of the multi-layer graph model for preset product recommendation provided by the embodiment of the present application reduces the complexity, so that the storage space occupied by the model is smaller, so that each search and iteration during product recommendation is only relevant to It reduces the overall running time complexity and memory usage of the model, and improves the search efficiency. At the same time, because the layers included in the multi-layer graph model recommended by the preset product are expanded to include at least three layers, And there is an association relationship between multiple layers, so that the recommendation algorithm based on PersonalRank can count the recommendation probability of each node, so as to achieve a multi-layer recommendation effect. Compared with the traditional PersonalRank algorithm structure, it only contains two layers and can only To implement a recommendation of an entity, the processing method for product recommendation provided by the embodiment of the present application can implement the recommendation of multiple layers of entities, and is more suitable for the business scenario of recommendation of multiple layers of entities, thereby not only realizing the multi-layer graph model of preset product recommendation. The structure is simple and the storage space is small, thus improving the search efficiency of product recommendation, and because it can implement multi-layer entity recommendation based on nodes of multiple layers, thus improving the diversity, accuracy and recommendation of product recommendations. efficiency.

Please refer to FIG. 3. FIG. 3 is a schematic diagram of the first sub-flow of the processing method for product recommendation provided by the embodiment of the present application. As shown in FIG. 3, in this embodiment, before acquiring the user ID of the target user, a include:

S111. Acquire an initial user object for product recommendation in response to a start-up instruction for product recommendation;

S112, judging whether there is historical transaction data of the initial user object;

S113, if the historical transaction data exists, use the initial user object as a target user;

S114. If the historical transaction data does not exist, do not use the initial user object as a target user.

Specifically, if it is monitored that a user logs into an online website or APP, or if an insurance agent clicks the preset evaluation button to learn about the product recommendation method for a target insured user, the logged-in user or the target insured user can be used as the The initial user object for product recommendation, in response to the user logging in to the online website or APP, or receiving the insurance agent clicking the preset evaluation button, to obtain the initial user object for product recommendation, and according to the initial user object, you can go to the database to query Whether the initial user object has historical transaction data, if there is the historical transaction data, determine that the initial user object is an old user, take the initial user object as the target user, and obtain the user ID of the target user, if If the historical transaction data does not exist, and the initial user object is not used as the target user, it can be determined that the initial user object is a new user. In the application scenario, if the historical transaction data does not exist, product recommendation cannot be performed according to the processing method for product recommendation provided in the embodiment of the present application. Products with larger sales.

Please refer to FIG. 4. FIG. 4 is a schematic diagram of the second sub-flow of the product recommendation processing method provided by the embodiment of the present application. As shown in FIG. 4, in this embodiment, the user is obtained according to the user identifier. The preset product recommended multi-layer graph model corresponding to the logo, including:

S114, obtaining user target data corresponding to the user ID according to the user ID, and obtaining an initial preset product recommendation multi-layer graph model;

S115. Fill the user target data into the initial preset product recommendation multi-layer graph model, and obtain the preset product recommendation multi-layer graph model corresponding to the user ID.

Specifically, when recommending products to a certain user, based on the data corresponding to the user, construct a multi-layer graph model of preset product recommendation corresponding to the user, that is, pre-build a graph-based initial preset product recommendation multi-layer graph model , the initial preset product recommendation multi-layer graph model can be limited to an empty or a model frame structure with an initial value, and the initial preset product recommended multi-layer graph model can only limit how many layers it contains, and according to the specific The business determines which business associations exist between adjacent layers, thereby determining which data each layer is used to store the user, obtains the user target data corresponding to the user ID, and obtains the initial preset based on the graph After the product recommendation multi-layer graph model, the user target data is filled into the initial preset product recommendation multi-layer graph model, that is, the specific entities (specific data) contained in the user target data are filled into the initial preset product The corresponding nodes of each layer included in the multi-layer graph model are recommended, so as to obtain the preset product recommended multi-layer graph model corresponding to the user ID. For example, for the preset product recommended multi-layer graph model corresponding to Zhang San, If there are 3 kinds of products that Zhang San has purchased, the purchased insurance product layer included in the recommended multi-layer graph model for this preset product will build 3 nodes to store 3 kinds of entities. Layer graph model, if Li Si has purchased 5 kinds of products, this preset product recommends that the purchased insurance product layer included in the multi-layer graph model will build 5 nodes to store 5 kinds of entities, because it is only necessary to confirm the user When product recommendation is performed, the preset product recommendation multi-layer graph model corresponding to the user ID is constructed. Or adopting a preset product recommendation multi-layer graph model with a fixed structure can reduce the space used by the preset product recommended multi-layer graph model for storage resources and save the usage rate of storage resources.

In one embodiment, the first layer of the preset product recommendation multi-layer graph model is the user layer, and the second layer is the product identifier corresponding to the product described by the user layer that the user has purchased, and The second layer to the nth layer are product element layers with business associations between adjacent layers, where n≥3, please refer to FIG. 5 , which is a product provided by this embodiment of the application The third sub-flow schematic diagram of the recommended processing method is shown in Figure 5. In this embodiment, according to all the nodes, the recommendation algorithm based on PersonalRank is used to obtain the recommendation probability of each of the nodes, including:

S51. Connect the user nodes included in the user layer with each node included in the second layer, and connect the nodes with business associations from the second layer to the nth layer. Edge connection to obtain several paths between the respective nodes of the first layer to the nth layer;

S52. According to each of the paths, start walking from one end of the path until the other end of the path is completed, and count the probability of each node being visited, and walk through multiple rounds until each node is visited. The probability that each of the nodes is visited tends to be stable, and the recommendation probability of each of the nodes relative to the user node is obtained.

Specifically, when recommending products to the user based on the products that the user has purchased, the first layer of the multi-layer graph model for preset product recommendation is set as the user layer, and the second layer is the user described by the user layer. The product identifier corresponding to the purchased product, and the second to nth layers are product element layers with business correlations between adjacent layers, where n≥3, for example, please continue Referring to FIG. 2, in FIG. 2, the nodes of the user layer of the first layer are used to describe the user 1, and the second layer describes the insurance products that the user 1 has purchased, including the purchased insurance product 1 and the purchased insurance product 2. The third layer is used to describe the purchased insurance product 1 can be sold through sales channel 1, the purchased insurance product 2 can be sold through sales channel 1 and sales channel 2, and the fourth layer is used to describe the sales channel 1. Insurance products of insurance category 1 and insurance category 3 can be sold, so the second to fourth layers are product element layers with business correlations between adjacent layers.

When the product recommendation of multi-layer entities is carried out based on purchased products, since the second layer is the product purchased by the user, that is, the user node and each node of the second layer have interacted with each other. All nodes are correlated with user nodes. First, connect the user nodes included in the user layer with each node included in the second layer. The edge connections can be undirected edges. For example, in Figure 2 The user 1 in Figure 2 is edge-connected with the purchased insurance product 1, and the user 1 in Figure 2 is edge-connected with the purchased insurance product 2, and the second layer to the nth layer has business associations The nodes of the relationship are edge-connected to obtain several paths between the respective nodes of the first layer to the nth layer. For example, the purchased insurance product 1 in Figure 2 is connected with the sales channel 1, and The purchased insurance product 2 in Figure 2 is edge-connected with the sales channel 1, and the purchased insurance product 2 in Figure 2 is edge-connected with the sales channel 2, and then the sales channel 1 in Figure 2 is connected to the insurance categories 1 and 2 respectively. Insurance category 3 is edge-connected, and sales channel 2 is edge-connected with insurance category 2 and insurance category 3 to form a path: user 1 - purchased insurance product 1 - sales channel 1 - insurance category 1; user 1-Purchased insurance product 1-Sales channel 1-Insurance category 3; User 2-Purchased insurance product 2-Sales channel 1-Insurance category 3; User 1-Purchased insurance product 2-Sales channel 2-Insurance category Class 2 and other paths, then according to the connected paths, and then according to each of the paths, start walking from one end of the path, for example, you can start walking from the user node of the path until the other end of the path is completed. The node of the fourth layer, and count the probability of each node being visited, and after multiple rounds of walking, until the probability of each node being visited converges and tends to be stable, the importance of the node to the user will converge to a certain value, which is used to describe the importance of each node, and obtain the recommendation probability of each node relative to the user node, which can improve the search efficiency, and because it can be based on multiple layers The node implements the recommendation of multi-layer entities, thereby improving the diversity, accuracy and recommendation efficiency of product recommendations.

Please refer to FIG. 6 . FIG. 6 is a schematic diagram of the fourth sub-flow of the processing method for product recommendation provided by the embodiment of the present application. As shown in FIG. 6 , in this embodiment, different types of The respective recommendation probability of the node is used as the target probability, including:

S61. According to the successive association relationships of nodes corresponding to different layers in the business, the layer corresponding to the starting node on the business is used as the starting layer, and the recommended probabilities contained in the starting layer are arranged in ascending order. Sorting in the smallest order to obtain the starting layer probability sequence, and obtaining several recommended probabilities at the top of the starting layer probability sequence as the starting target probability;

S62. According to the initial target probability, determine a number of the highest recommended probabilities of the node existing paths with the initial target probability in each other layer as the corresponding associated target probability, and compare the initial target probability with the The associated target probability is used as the target probability of different types of nodes filtered out.

Specifically, since different nodes have successive association relationships in the business, such as the relationship between determination and determination, and the logical relationship of the order before and after, for example, please continue to refer to Figure 2, sales channel 1 can only sell insurance category 1 and insurance category 3 , sales channel 2 can only sell insurance category 2 and insurance category 3. When the sales channel is used as the pre-order node, the subsequent corresponding insurance category nodes can be determined, so that according to the order of the nodes corresponding to different layers in the business association relationship, take the layer corresponding to the starting node on the business as the starting layer, and sort the recommended probabilities contained in the starting layer in descending order to obtain the starting layer probability sequence, And obtain the first several recommendation probabilities in the starting layer probability sequence as the starting target probability. For example, you can obtain the highest recommendation probability in the starting layer probability sequence as the starting target probability, or you can obtain The top three recommended probabilities in the probability sequence of the initial layer are used as the initial target probability to recommend three products, and then according to the initial target probability, determine the difference between the initial target probability and the initial target probability in each other layer. At least one highest recommendation probability of the node existence path is used as the corresponding associated target probability, and the initial target probability and the associated target probability are used as the respective target probability of the selected nodes of different types. For example, please continue to refer to Figure 2. When recommending sales channels and insurance categories corresponding to sales channels, if sales channel 1 is the highest recommendation probability, the recommendation probability of sales channel 1 can be obtained as the initial target probability, and the The recommendation probability of insurance category 1 that has a path with sales channel 1 is taken as the associated target probability, the recommendation probability of sales channel 1 is taken as the target probability of the sales channel node, and the recommended probability of insurance category 1 is taken as the insurance category node. Target probability, or obtain the respective recommendation probability of insurance category 1 and insurance category 3 that have a path with sales channel 1 as the associated target probability, and use the recommended probability of sales channel 1 as the target probability of the sales channel node, and take the insurance category as the target probability. The respective recommendation probabilities of category 1 and insurance category 3 are used as the target probability of insurance category nodes, so as to realize the recommendation of associated multi-entity nodes, and to achieve the diversity, accuracy and recommendation efficiency of product recommendations.

It should be noted that, for the recommended processing methods for products described in the above embodiments, the technical features contained in different embodiments can be recombined as required to obtain a combined embodiment, but all of them are within the protection claimed in this application. within the range.

Please refer to FIG. 7 , which is a schematic block diagram of a processing apparatus for product recommendation provided by an embodiment of the present application. Corresponding to the above-mentioned processing method for product recommendation, an embodiment of the present application further provides a processing device for product recommendation. As shown in FIG. 7 , the product recommendation processing apparatus includes a unit for executing the above-mentioned product recommendation processing method, and the product recommendation processing apparatus may be configured in a computer device. Specifically, please refer to FIG. 7 , the product recommendation processing device 70 includes a first obtaining unit 71 , a second obtaining unit 72 and a probability using unit 73 .

The first obtaining unit 71 is configured to obtain the user ID of the target user, and according to the user ID, obtain a preset product recommendation multi-layer graph model corresponding to the user ID, and the preset product recommendation multi-layer graph The model includes at least three layers, each layer includes several nodes, and there is a business relationship between adjacent layers, including a user layer, and the user layer only includes one node as a user node, so The user node is used to describe the user identifier;

The second obtaining unit 72 is configured to obtain the recommendation probability of each of the nodes according to the recommendation algorithm based on PersonalRank according to all the nodes;

The probability using unit 73 is configured to, according to the recommendation probability and based on the preset probability usage mode, use the recommendation probability corresponding to each node of different types to perform product recommendation.

In one embodiment, the product recommendation processing device 70 further includes:

The response unit is used to respond to the start-up instruction for product recommendation, and obtain the initial user object of product recommendation;

a judging unit for judging whether there is historical transaction data of the initial user object;

A determining unit, configured to use the initial user object as a target user if the historical transaction data exists.

In one embodiment, the first obtaining unit 71 includes:

a first obtaining subunit, configured to obtain user target data corresponding to the user ID according to the user ID, and obtain a graph-based initial preset product recommendation multi-layer graph model;

The filling subunit is used for filling the user target data into the initial preset product recommendation multi-layer graph model to obtain the preset product recommendation multi-layer graph model corresponding to the user ID.

In one embodiment, the first layer of the preset product recommendation multi-layer graph model is the user layer, and the second layer is the product identifier corresponding to the product described by the user layer that the user has purchased, and The second layer to the nth layer are product element layers with business associations between adjacent layers, where n≥3, the second obtaining unit 72 includes:

The first connection sub-unit is used to connect the user node included in the user layer with each node included in the second layer, and connect the second layer to the nth layer with business information. The nodes of the associated relationship are connected by edges, and several paths between the respective nodes of the first layer to the nth layer are obtained;

The second connection sub-unit is used to walk from one end of the path until the other end of the path is completed according to each path, and count the probability of each node being visited, and after a number of Walk in turns until the probability that each of the nodes is visited converges and tends to be stable, and the recommendation probability of each of the nodes relative to the user node is obtained.

In one embodiment, the probability using unit 73 includes:

a first screening subunit, configured to screen respective recommended probabilities of different types of nodes as target probabilities based on preset probability screening conditions;

The first display subunit is configured to use the node corresponding to the target probability as the target node, and display the recommended product on the preset terminal according to the target node.

In one embodiment, the probability using unit 73 includes:

a first screening subunit, configured to screen respective recommended probabilities of different types of nodes as target probabilities based on preset probability screening conditions;

The second display subunit is used to display the target probability on the preset terminal, so that the relevant personnel can recommend the product according to the recommendation probability.

In one embodiment, the first screening subunit includes:

The second acquisition sub-unit is used for taking the layer corresponding to the starting node on the business as the starting layer according to the successive associations of the nodes corresponding to different layers in the business, and taking the starting layer containing the The recommendation probabilities are sorted in descending order to obtain a starting layer probability sequence, and several top recommended probabilities in the starting layer probability sequence are obtained as starting target probabilities;

The determining subunit is configured to determine, according to the initial target probability, several highest recommended probabilities of the existence paths of the nodes with the initial target probability in each other layer as the corresponding associated target probability, and use the initial target probability as the corresponding associated target probability. The probability and the associated target probability are taken as the respective target probability of the selected nodes of different types.

It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned product recommendation processing device and each unit may refer to the corresponding description in the foregoing method embodiments. For the convenience and brevity of the description, here No longer.

Meanwhile, the division and connection method of each unit in the above-mentioned product recommendation processing device are only used for illustration. In other embodiments, the product recommendation processing device can be divided into different units according to needs, and the product recommendation processing device Each unit in the device adopts different connection sequences and methods to complete all or part of the functions of the processing device recommended by the above product.

The above-mentioned processing apparatus for product recommendation can be implemented in the form of a computer program, and the computer program can be executed on a computer device as shown in FIG. 8 .

Please refer to FIG. 8 , which is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device 500 may be a computer device such as a desktop computer or a server, or may be a component or component in other devices.

Referring to FIG. 8, the computer device 500 includes a processor 502, a memory and a network interface 505 connected through a system bus 501, wherein the memory may include a non-volatile storage medium 503 and an internal memory 504, and the memory may also be volatile Sexual storage medium.

The nonvolatile storage medium 503 can store an operating system 5031 and a computer program 5032 . When the computer program 5032 is executed, it can cause the processor 502 to execute an above-mentioned processing method for product recommendation.

The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

The internal memory 504 provides an environment for running the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a processing method recommended by the product.

The network interface 505 is used for network communication with other devices. Those skilled in the art can understand that the structure shown in FIG. 8 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied. The specific computer device 500 may include more or fewer components than shown, or combine certain components, or have a different arrangement of components. For example, in some embodiments, the computer device may only include a memory and a processor. In such an embodiment, the structures and functions of the memory and the processor are the same as those of the embodiment shown in FIG. 8 , which will not be repeated here.

Wherein, the processor 502 is configured to run the computer program 5032 stored in the memory, so as to realize the following steps: obtaining the user ID of the target user, and obtaining the preset product recommendation corresponding to the user ID according to the user ID A multi-layer graph model, the preset product recommendation multi-layer graph model includes at least three layers, each layer includes several nodes, and there is a business relationship between adjacent layers, including a user layer, And the user layer only includes one node as a user node, and the user node is used to describe the user identification; according to all the nodes, based on the recommendation algorithm of PersonalRank, obtain the recommendation probability of each of the nodes; according to the Recommendation probability, and based on the preset probability usage method, use the recommendation probability corresponding to different types of nodes to recommend products.

In one embodiment, the processor 502 further implements the following steps before acquiring the user identity of the target user:

In response to the start-up instruction for product recommendation, obtain the initial user object for product recommendation;

Determine whether there is historical transaction data of the initial user object;

If the historical transaction data exists, the initial user object is used as the target user.

In one embodiment, the processor 502 specifically implements the following steps when implementing the obtaining of the preset product recommendation multi-layer graph model corresponding to the user ID according to the user ID:

According to the user ID, obtain user target data corresponding to the user ID, and obtain an initial preset product recommendation multi-layer graph model;

Filling the user target data into the initial preset product recommendation multi-layer graph model to obtain the preset product recommendation multi-layer graph model corresponding to the user ID.

In one embodiment, the first layer of the preset product recommendation multi-layer graph model is the user layer, and the second layer is the product identifier corresponding to the product described by the user layer that the user has purchased, and The second layer to the nth layer are product element layers with business associations between adjacent layers, where n≥3, the processor 502 is implementing the above according to all the nodes, In the recommendation algorithm based on PersonalRank, when obtaining the recommendation probability of each node, the following steps are specifically implemented:

Connect the user nodes included in the user layer with each node included in the second layer, and connect the nodes with business associations in the second layer to the nth layer. , get several paths between the respective nodes of the first layer to the nth layer;

According to each of the paths, walk from one end of the path until the other end of the path is completed, and count the probability of each node being visited, and walk through multiple rounds until each node is visited. The probability that the node is visited tends to be stable, and the recommendation probability of each node relative to the user node is obtained.

In one embodiment, the processor 502 specifically implements the following steps when implementing the preset probability-based usage mode and using the recommendation probabilities corresponding to different types of nodes to perform product recommendation:

Based on the preset probability screening conditions, the respective recommended probabilities of different types of nodes are screened as the target probability;

The node corresponding to the target probability is used as the target node, and the recommended product is displayed on the preset terminal according to the target node.

In one embodiment, the processor 502 specifically implements the following steps when implementing the preset probability-based usage mode and using the recommendation probabilities corresponding to different types of nodes to perform product recommendation:

Based on the preset probability screening conditions, the respective recommended probabilities of different types of nodes are screened as the target probability;

The target probability is displayed on the preset terminal, so that the relevant personnel can recommend the product according to the recommendation probability.

In one embodiment, the processor 502 specifically implements the following steps when implementing the preset probability-based screening conditions to select the respective recommended probabilities of different types of nodes as target probabilities:

According to the successive associations of nodes corresponding to different layers in the business, the layer corresponding to the starting node on the business is taken as the starting layer, and the recommended probabilities contained in the starting layer are arranged in descending order. Sort the sequence in order to obtain a starting layer probability sequence, and obtain a number of recommended probabilities at the top of the starting layer probability sequence as starting target probabilities;

According to the initial target probability, determine some of the highest recommended probabilities of the node existing paths with the initial target probability in each other layer as the corresponding associated target probability, and compare the initial target probability with the associated target The probability is used as the target probability of the different types of nodes screened out.

It should be understood that, in this embodiment of the present application, the processor 502 may be a central processing unit (Central Processing Unit, CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), dedicated integrated Circuit (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. Wherein, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor or the like.

It can be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program, and the computer program can be stored in a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the flow steps of the above-described method embodiments.

Therefore, the present application also provides a computer-readable storage medium. The computer-readable storage medium may be a non-volatile computer-readable storage medium, or a volatile computer-readable storage medium, and the computer-readable storage medium stores a computer program, and the computer program is executed by the processor causes the processor to perform the following steps:

A computer program product, when run on a computer, causes the computer to execute the steps of the product recommendation processing method described in the above embodiments.

The computer-readable storage medium may be an internal storage unit of the aforementioned device, such as a hard disk or a memory of the device. The computer-readable storage medium may also be an external storage device of the device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card equipped on the device , Flash Card (Flash Card) and so on. Further, the computer-readable storage medium may also include both an internal storage unit of the device and an external storage device.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, for the specific working process of the above-described devices, devices and units, reference may be made to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

The storage medium is a physical, non-transitory storage medium, such as a U disk, a removable hard disk, a read-only memory (Read-Only Memory, ROM), a magnetic disk or an optical disk and other physical storage that can store computer programs. medium.

Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. Interchangeability, the above description has generally described the components and steps of each example in terms of function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each particular application, but such implementations should not be considered beyond the scope of this application.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is only a logical function division, and other division methods may be used in actual implementation. For example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not implemented.

The steps in the method of the embodiment of the present application may be adjusted, combined and deleted in sequence according to actual needs. Units in the apparatus of the embodiment of the present application may be combined, divided, and deleted according to actual needs. In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as an independent product, may be stored in a storage medium. Based on this understanding, the technical solutions of the present application are essentially or part of contributions to the prior art, or all or part of the technical solutions can be embodied in the form of software products, and the computer software products are stored in a storage medium , including several instructions to cause an electronic device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

The above are only specific implementations of the present application, but the scope of protection disclosed in the present application is not limited to this. Any person skilled in the art can easily think of various equivalents within the technical scope disclosed in the present application. Modifications or substitutions of the present application shall be included within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (10)

1. a processing method of product recommendation, is characterized in that, comprises: Obtain the user ID of the target user, and according to the user ID, obtain a preset product recommendation multi-layer graph model corresponding to the user ID, the preset product recommendation multi-layer graph model includes at least three layers, each A layer includes several nodes, and there is a business relationship between adjacent layers, including a user layer, and the user layer only includes one node as a user node, and the user node is used to describe the user ID ; According to all the nodes, the recommendation algorithm based on PersonalRank is used to obtain the recommendation probability of each of the nodes; According to the recommendation probability and based on the preset probability usage mode, the recommendation probability corresponding to different types of nodes is used to perform product recommendation. 2. The processing method for product recommendation according to claim 1, wherein before the acquiring the user identification of the target user, further comprising: In response to the start-up instruction for product recommendation, obtain the initial user object for product recommendation; Determine whether there is historical transaction data of the initial user object; If the historical transaction data exists, the initial user object is used as the target user. 3 . The processing method for product recommendation according to claim 1 , wherein, acquiring the preset product recommendation multi-layer graph model corresponding to the user ID according to the user ID comprises: 3 . According to the user ID, obtain user target data corresponding to the user ID, and obtain an initial preset product recommendation multi-layer graph model; Filling the user target data into the initial preset product recommendation multi-layer graph model to obtain the preset product recommendation multi-layer graph model corresponding to the user ID. 4. The processing method for product recommendation according to claim 1, wherein the first layer of the preset product recommendation multi-layer graph model is a user layer, and the second layer is described by the user layer. The product identifier corresponding to the product that the user has purchased, and the second layer to the nth layer are product element layers with business associations between adjacent layers, where n≥3, according to All the nodes, based on the recommendation algorithm of PersonalRank, obtain the recommendation probability of each of the nodes, including: Connect the user nodes included in the user layer with each node included in the second layer, and connect the nodes with business associations in the second layer to the nth layer. , get several paths between the respective nodes of the first layer to the nth layer; According to each of the paths, walk from one end of the path until the other end of the path is completed, and count the probability of each node being visited, and walk through multiple rounds until each node is visited. The probability that the node is visited tends to be stable, and the recommendation probability of each node relative to the user node is obtained. 5. The processing method for product recommendation according to claim 1, wherein the product recommendation is performed based on the preset probability usage mode using the respective recommendation probabilities corresponding to different types of nodes, comprising: Based on the preset probability screening conditions, the respective recommended probabilities of different types of nodes are screened as the target probability; The node corresponding to the target probability is used as the target node, and the recommended product is displayed on the preset terminal according to the target node. 6 . The processing method for product recommendation according to claim 1 , wherein the product recommendation is performed based on the preset probability usage mode using the respective recommendation probabilities corresponding to different types of nodes, comprising: 6 . Based on the preset probability screening conditions, the respective recommended probabilities of different types of nodes are screened as the target probability; The target probability is displayed on the preset terminal, so that the relevant personnel can recommend the product according to the recommendation probability. 7. The processing method for product recommendation according to claim 5 or 6, characterized in that, based on a preset probability screening condition, screening the respective recommendation probabilities of different types of nodes as target probabilities, comprising: According to the successive associations of nodes corresponding to different layers in the business, the layer corresponding to the starting node on the business is taken as the starting layer, and the recommended probabilities contained in the starting layer are arranged in descending order. Sort the sequence in order to obtain a starting layer probability sequence, and obtain a number of recommended probabilities at the top of the starting layer probability sequence as starting target probabilities; According to the initial target probability, determine some of the highest recommended probabilities of the node existing paths with the initial target probability in each other layer as the corresponding associated target probability, and compare the initial target probability with the associated target The probability is used as the target probability of the different types of nodes screened out. 8. A processing device for product recommendation, comprising: The first obtaining unit is configured to obtain the user ID of the target user, and according to the user ID, obtain a preset product recommendation multi-layer graph model corresponding to the user ID, and the preset product recommendation multi-layer graph model includes at least Three layers, each layer contains several nodes, there is a business relationship between adjacent layers, including a user layer, and the user layer only includes one node as a user node, the user node used to describe the user ID; A second obtaining unit, configured to obtain the recommendation probability of each of the nodes according to the recommendation algorithm based on PersonalRank according to all the nodes; The probability usage unit is configured to use the recommendation probability corresponding to different types of nodes to perform product recommendation according to the recommendation probability and based on the preset probability usage mode. 9. A computer device, characterized in that the computer device comprises a memory and a processor connected to the memory; the memory is used to store a computer program; the processor is used to run the computer program to execute the The steps of the method of any one of claims 1-7. 10. A computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1-7 can be implemented.

Images