Product information pushing method and device based on artificial intelligence, equipment and medium

Abstract

The invention discloses a product information pushing method and device based on artificial intelligence, equipment and a medium, and relates to the technical field of artificial intelligence and financial science and technology, and the method preferentially screens out users with low business product receiving degree when facing a large customer group, achieves targeted recommendation of business products, and improves the user experience. Compared with the prior art, the method has the advantages that computing resources, storage resources and the like for pushing product information are saved, the efficiency of pushing the product information for a large customer group is improved to a certain extent, product preference analysis can be quickly performed on users who are easier to accept insurance products in an insurance recommendation scene, and the user experience is improved according to the preference degree of the users for browsed products. And the target insurance product meeting the user demand is screened out, so that the insurance product recommendation effect on a large customer group is improved on the whole.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and financial technology, and in particular to an artificial intelligence-based product information push method, device, equipment and medium. Background Art

[0002] With the development of the global economy and rising demands for a better quality of life, financial products such as insurance and fund products are attracting increasing attention. In particular, non-motor insurance, as a crucial component of the insurance industry, has become a new growth area. Non-motor insurance covers a wide range of areas, including but not limited to health insurance, life insurance, homeowners insurance, and accident insurance, meeting the needs of different groups at different stages of life. With the continuous improvement of the social security system, public awareness and acceptance of non-motor insurance are increasing year by year, and market demand is showing a rapid growth trend.

[0003] Traditional product information push methods mostly rely on pre-defined customer groups and then use fixed marketing strategies to push product information to users within these groups. This approach struggles with large customer groups, especially when pushing information to a wide variety of insurance products. Furthermore, personalized recommendations for each user require significant computing resources, resulting in low product information push efficiency. Overall, current product information push methods are ineffective when recommending products and pushing information to large customer groups. Summary of the Invention

[0004] The present invention provides an artificial intelligence-based product information push method, device, equipment and medium to solve the problem of poor effect when recommending products and pushing information to a large customer base.

[0005] In a first aspect, a method for pushing product information based on artificial intelligence is provided, the method comprising: Obtaining the user's personal information and the user's behavioral data associated with the business product; Inputting the behavior data into a customer acquisition maturity evaluation model to obtain the customer acquisition maturity corresponding to the user, wherein the customer acquisition maturity is used to represent the user's acceptance of the business product; If the customer retention maturity corresponding to the user is greater than or equal to a preset threshold, determining the user's preference for the business products involved in the product browsing records based on the personal information and the product browsing records in the behavioral data; According to the preference level, target business products that need to be recommended to the user are screened from the business products involved in the product browsing records.

[0006] In a second aspect, a product information push device based on artificial intelligence is provided, the device comprising: An acquisition module, configured to acquire a user's personal information and behavioral data associated with the user and the business product; A customer maturity assessment module, configured to input the behavioral data into a customer maturity assessment model to obtain the customer maturity corresponding to the user, wherein the customer maturity is used to represent the user's acceptance of the business product; A product preference analysis module is configured to determine the user's preference for the business products involved in the product browsing records based on the personal information and the product browsing records in the behavioral data if the customer retention maturity corresponding to the user is greater than or equal to a preset threshold; The screening module is used to screen target business products that need to be recommended to the user from the business products involved in the product browsing records according to the preference degree, so as to push product information of the target business products to the user terminal.

[0007] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned artificial intelligence-based product information push method are implemented.

[0008] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned artificial intelligence-based product information push method are implemented.

[0009] In the solution implemented by the above-mentioned artificial intelligence-based product information push method, device, equipment and medium, first, based on the behavioral data of different users, the customer maturity of different users is evaluated, and then the users are screened using the customer maturity. For users whose customer maturity is greater than or equal to the preset threshold, product preference analysis is performed. In this way, when facing a large customer base, users with a low degree of acceptance of business products can be preferentially screened out to achieve targeted recommendations of business products, saving computing resources and storage resources for product information push and even the entire product recommendation process, and to a certain extent improving the efficiency of product information push for large customer groups. The present invention can quickly analyze the product preferences of users who are more receptive to business products, and screen out target business products that meet user needs based on the user's preference for the browsed business products, thereby improving the overall effect of product information push and even product recommendation for large customer groups. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0011] Figure 1 This is a schematic diagram of an application environment of a product information push method based on artificial intelligence in one embodiment of the present invention; Figure 2 This is a flow chart of a method for pushing product information based on artificial intelligence in one embodiment of the present invention; Figure 3 This is a flow chart of a method for pushing product information based on artificial intelligence in another embodiment of the present invention; Figure 4 This is a flow chart of a method for pushing product information based on artificial intelligence in another embodiment of the present invention; Figure 5 This is a structural diagram of a product information push device based on artificial intelligence in one embodiment of the present invention; Figure 6 It is a structural diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0012] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0013] The product information push method based on artificial intelligence provided by the embodiment of the present invention can be applied in Figure 1 In an application environment, a user terminal communicates with a server via a network. The server can collect user behavior data associated with business products on the user terminal and personal information provided by the user terminal. After determining the target business product to be pushed to the user based on the user's behavior data and personal information, the server pushes product information of the target business product to the user via the user terminal.

[0014] When recommending insurance products in a financial scenario, the artificial intelligence-based product information push method provided in this application can be applied to screen out users with a high degree of acceptance of insurance products from a large customer base and push insurance product information that meets their needs to them.

[0015] This method can cover different types of insurance products within the insurance business within a single information push process. That is, within a single information push process, this method adaptively selects insurance products that best suit the needs of different users from among different types of insurance products. When dealing with a large customer base, there's no need to segment the customer base. The present invention provides an artificial intelligence-based product information push method that predicts a user's acceptance of insurance products based on their behavioral data (it should be noted that insurance products here do not refer to specific products such as health insurance or life insurance, but rather a general term for insurance products). If the user's acceptance of insurance products meets a preset standard, the method further determines the user's preference for each insurance product in their product browsing history based on their personal information data and product browsing history. This allows targeted insurance products that best suit the user's needs to be selected based on their preference. This method allows users with a high acceptance level to be quickly pushed insurance products that meet their needs, while temporarily suspending AI-based product information push for users with a low acceptance level, thereby improving user satisfaction. This approach not only conserves computing and storage resources for pushing insurance product information and even recommending insurance products, but also avoids the inconvenience of marketing to users who don't have a strong desire to purchase insurance.

[0016] The user terminal may include, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. The server may be implemented as a standalone server or a server cluster consisting of multiple servers. The present invention is described in detail below using specific embodiments.

[0017] See also Figure 2 As shown, Figure 2 A flowchart of a method for pushing product information based on artificial intelligence provided by an embodiment of the present invention includes the following steps: S201, obtaining user's personal information and behavior data associated with the user and business products.

[0018] Personal information may include but is not limited to age, gender, occupation, etc.; Behavioral data associated with business products may include but is not limited to users' product browsing history, click behavior, historical purchase records, participation in activities, etc.

[0019] For example, by collaborating with various social platforms to place marketing advertisements for insurance products on the platforms, the server can extract the user information associated with each user ID from the platform data as the personal information in S201 above. Furthermore, the server can extract from the platform data the browsing history of insurance products, click behavior on pages related to insurance products, past purchase history for insurance products, and participation in activities related to insurance products for each user ID. Activity participation can be characterized by information such as the number of participations and the length of time spent on the activity interface.

[0020] S202: Input the behavior data into a customer acquisition maturity evaluation model to obtain the customer acquisition maturity corresponding to the user, where the customer acquisition maturity is used to represent the user's acceptance of the business product.

[0021] The customer retention maturity assessment model can be constructed based on a deep neural network model. By training the customer retention maturity assessment model, the customer retention maturity assessment model can analyze the user's acceptance of business products from the user's historical behavior data.

[0022] In some embodiments, the collected behavioral data may be preprocessed before being input into the customer maturity assessment model. For example, the preprocessing process may include, but is not limited to, cleaning, padding, converting, removing invalid or duplicate data, filling missing values, normalizing numerical data, and encoding categorical data.

[0023] In some embodiments, the customer maturity assessment model can be trained in the following manner: Extracting time series data corresponding to preset features from the historical behavior data of different sample users, wherein the preset features include the activity level of the sample users within a preset time period and / or the frequency of visits of the sample users to business product-related pages; The customer maturity assessment model is trained using the time series data, and the customer maturity assessment model includes an LSTM model.

[0024] For example, use the TensorFlow framework to build an LSTM model and train it using historical behavior data. During model training, you can select a loss function and optimizer as needed, such as the cross-entropy loss function and the Adam optimizer. This application does not limit this.

[0025] The LSTM model can also be replaced with other machine learning models such as Lightgbm (an ensemble learning framework). This implementation utilizes a customer retention maturity assessment model built on the LSTM model to monitor changes in user online behavior in real time, extracting time-series changes in user behavior, and predicting each user's customer retention maturity. When a user meets the customer retention maturity criteria, the model promptly analyzes their product preferences and provides them with product information tailored to their needs.

[0026] S203: If the customer retention maturity corresponding to the user is greater than or equal to a preset threshold, determine the user's preference for the business products involved in the product browsing records based on the personal information and the product browsing records in the behavioral data.

[0027] After obtaining the customer acquisition maturity of each user, the customer acquisition maturity of each user is compared with the preset threshold corresponding to the customer acquisition maturity. In the next step of product preference analysis, users whose customer acquisition maturity is less than the preset threshold are no longer considered, and only users whose customer acquisition maturity is greater than or equal to the preset threshold are considered.

[0028] When analyzing user product preferences, each business product involved in the user's product browsing history can be analyzed based on the user's personal information. This process can be achieved through a deep neural network, which can output the probability of the user purchasing each business product.

[0029] For example, in the context of insurance recommendations, the product information for each insurance product may include one or more of the following: insurance type, age limit, premium amount, payment method (lump sum or installment payment, etc.), compensation conditions, insurance term, basic coverage, etc.; the user's personal information may include one or more of the following: age, family status, financial status, occupational characteristics, health status, etc. When determining the user's preference for each insurance product, a deep learning model can be used to evaluate the degree of compatibility between the product information of the insurance product and the user's personal information. The higher the degree of compatibility, the greater the user's preference for the insurance product. Conversely, the lower the degree of compatibility, the less the user's preference for the insurance product.

[0030] S204 , based on the preference level, select target business products that need to be recommended to the user from the business products involved in the product browsing records, and push product information of the target business products to the user terminal.

[0031] After determining the user's preference for each product in the product browsing history, these products can be ranked by preference, with the top N% of products selected as target products. These target products are then recommended to the user. These target products can be recommended to the user by pushing their product information to the user's terminal.

[0032] In summary, the artificial intelligence-based product information push method proposed in the embodiment of the present application obtains the user's personal information and behavioral data associated with the business products, first uses the customer maturity assessment model to analyze the user's acceptance of the business products from the user's behavioral data, and filters out users with low acceptance, that is, retains users whose customer maturity is greater than or equal to the preset threshold. For these users, based on the user's personal information and product browsing records in the behavioral data, analyze and determine the user's preference for each business product in the product browsing records, and finally, based on the preference level, filter out the target business products from the business products involved in the product browsing records. In this way, when facing a large customer base, users with low acceptance of business products can be screened out first, and targeted push of business product information can be achieved, saving computing resources and storage resources for artificial intelligence-based product information push and even the entire product recommendation process, etc., and to a certain extent improving the efficiency of product information push and even product recommendation for large customer groups. In the insurance recommendation scenario, the present invention can quickly analyze the product preferences of users who are more receptive to insurance products, and screen out target insurance products that meet user needs based on the user's preference for the browsed insurance products, thereby improving the overall effect of insurance product recommendation for large customer groups.

[0033] Based on the above embodiments, Figure 3 As shown, the artificial intelligence-based product information push method proposed in this application may further include the following steps after the above step S204: S301: Generate marketing copy based on the personal information and product information of the target business product.

[0034] S302: Push the marketing copy to the user terminal.

[0035] In some embodiments, a preset promotion template can be obtained, and the user's personal information and product information of the target business product can be updated to the promotion template to generate a customized push page for the user, that is, the above-mentioned marketing copy.

[0036] In other embodiments, a prompt may be created by combining the user's personal information and the product information of the target business product, and then a large model may be used to generate marketing copy.

[0037] When pushing marketing copy to users, it can be achieved through various forms such as text messages, pop-ups, etc.

[0038] For example, by utilizing the interface provided by the WeChat open platform, we can realize the automation and personalized push of marketing copy, while monitoring the push effect and feeding it back to the data collection module to form a closed loop of continuous optimization.

[0039] The embodiment of the present application combines the user's personal information and the target business products screened for the user to generate customized and personalized marketing copy for the user, thereby improving the effect of product information push based on artificial intelligence.

[0040] Based on the above embodiments, Figure 4 As shown, the above step S203 of "determining the user's preference for each business product in the product browsing history based on the personal information and the product browsing history in the behavior data" can be achieved by the following process: S401, obtaining product information of the business product involved in the product browsing record; S402: Input the personal information and the product information of the business product into a purchase probability prediction model to predict the probability value of the user purchasing the business product, and use the probability value as the preference level.

[0041] The personal information and the product information of each business product in the product browsing record are input into a purchase probability prediction model to obtain a probability value of the user purchasing each business product in the product browsing record.

[0042] The purchase probability prediction model may include but is not limited to an input layer, an intermediate layer, a neural collaborative filtering layer, an output layer, etc.; The input layer consists of two parts: one is used to extract user feature vectors from personal information, and the other is used to extract product feature vectors from the product information of each business product; The input layer inputs the user feature vector and the product feature vector into the middle layer. In the embodiment of the present application, the middle layer of the purchase probability prediction model includes multiple fully connected layers. For ease of distinction, the fully connected layer set in the middle layer is recorded as the first fully connected layer. The number of neurons in the first fully connected layer decreases along the direction of information transmission. For example, the first fully connected layer connected to the input layer is FC1, and the first fully connected layer connected to the neural collaborative filtering layer is FCn. Then the number of neurons in FC1 > the number of neurons in FC2 > ... > the number of neurons in FCn, where n is an integer greater than 2.

[0043] In this way, the sparse features of the input layer can be expressed as dense vectors in the intermediate layers.

[0044] The Neural Collaborative Filtering Layer (NCF Layer) contains a multi-layer neural network structure, which is used to capture the complex interactive relationship between users and business products, or to capture the correlation between users' personal information and product information of business products.

[0045] The output layer may include two layers, one of which is a fully connected layer. For ease of distinction, the fully connected layer set in the output layer is referred to as the second fully connected layer; the other layer is an activation function, such as a sigmoid function. The activation function maps the features output by the above intermediate layer into a probability value, which is used to represent the probability of a user purchasing a certain business product.

[0046] The number of neurons in the second fully connected layer is the same as the number of neurons in the last first fully connected layer on the middle layer. Combined with the above example, the number of neurons in the second fully connected layer is the same as the number of neurons in FCn.

[0047] In the embodiment of the present application, a random deactivation mechanism (dropout mechanism) is added to the second fully connected layer to randomly deactivate neurons in the layer, thereby reducing the risk of overfitting.

[0048] It should be noted that the purchase probability prediction model used in the embodiments of the present application is a pre-built model that is trained based on a large amount of sample data (the sample data may include personal information of different users, historical product browsing records, historical product purchase records, etc.) in specific application scenarios (such as insurance product recommendation scenarios) to determine the model with the optimal hyperparameter configuration.

[0049] Among them, various model training methods such as semi-supervised learning and unsupervised learning can be used for model training, which will not be repeated here.

[0050] This implementation provides a purchase probability prediction model, which extracts user feature vectors from the user's personal information and product feature vectors from the product information of the business products, and then uses multiple fully connected layers with a decreasing number of neurons set in the middle layer to express the user feature vectors and product feature vectors as dense vectors, and inputs the dense vectors into the neural collaborative filtering layer, thereby fully capturing the potential associations and interactions between the business products browsed by the user and the user's personal information, improving the model's expressive power, and enhancing the accuracy of product preference analysis. In addition, before mapping the output features to probability values, a fully connected layer with a random deactivation mechanism is used to process the feature vectors output by the neural collaborative filtering layer to reduce the risk of overfitting. This ensures the expressiveness of the purchase probability prediction model in practical applications. When applied in the insurance product recommendation scenario, the model can have the expressiveness of the model training and testing process when performing product preference analysis on different types of insurance products and different customer groups, ensuring the accuracy of product preference analysis.

[0051] Based on the above embodiment, the number of neurons in two adjacent first fully connected layers is in multiples. For example, the first first fully connected layer has 256 neurons, the second first fully connected layer has 128 neurons, and the third first fully connected layer has 64 neurons.

[0052] The exponential reduction in the number of neurons in two adjacent fully connected layers helps reduce the number of parameters, lowering model complexity and reducing computing resource consumption. Furthermore, as the number of neurons decreases, the network can gradually extract more critical and abstract features from the original input user and product features. This feature extraction process helps the neural collaborative filtering layer better explore the correlation between user and product features, thereby improving the accuracy of product preference analysis.

[0053] The number of neurons in each first fully connected layer is reduced exponentially, making it easier to integrate sparse features.

[0054] Based on the above embodiment, in step S203, "determining the user's preference for the business product involved in the product browsing record based on the personal information and the product browsing record in the behavior data" can also be implemented in the following manner: Obtain product information of the business product involved in the product browsing record; input the personal information and the product information of the business product into a purchase probability prediction model to predict the probability value of the user purchasing the business product.

[0055] It should be noted that the implementation method of predicting the probability value of a user purchasing a business product can refer to the description of steps S401-S402 of the above embodiment, and will not be repeated here.

[0056] Determine the proportion of the number of views corresponding to the business product in the total number of views corresponding to the product browsing records; multiply the probability value and the proportion by the corresponding preset weights respectively, and sum them up to obtain the sum value as the preference degree.

[0057] After using the purchase probability prediction model to predict the probability value of a user purchasing a certain business product, this embodiment can further analyze the user's product browsing history to determine the proportion of the number of times the business product was browsed in the product browsing history to the total number of views corresponding to the product browsing history; Then, using pre-set weights, the probability value and proportion of the purchase probability corresponding to the business product are combined to obtain the user's preference for the business product. The weights can be obtained by training and analyzing sample data from a large number of specific application scenarios (such as insurance recommendation scenarios and other financial product recommendation scenarios), or can be flexibly adjusted as needed, and this application does not impose any restrictions.

[0058] In this embodiment, while considering the degree of match between the user's personal information and the product information of the business product (i.e., the probability value of the user purchasing the business product), the business products that the user frequently browses are also determined from the user's product browsing history. When determining the user's preference value for the business product, both the theoretical information match degree and the user's actual purchase intention are considered. In this way, when using the preference value to determine the target business product, it is easier to screen out products that meet the user's needs.

[0059] Based on the above embodiment, in step S204, target business products are screened from product browsing records according to the degree of preference. There are also the following implementation methods: From all the business products involved in the product browsing records, the business product with the greatest preference is selected as the target business product.

[0060] In this way, more targeted and accurate recommendations can be made to users, avoiding excessive redundant information in the pushed information.

[0061] In addition, before referencing the product browsing history to perform product preference analysis, the user's product purchase history can be combined to exclude products purchased by the user from all business products involved in the product browsing history.

[0062] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0063] In one embodiment, a product information push device based on artificial intelligence is provided, and the product information push device based on artificial intelligence corresponds to the product information push method based on artificial intelligence in the above embodiment. Figure 5 As shown, the artificial intelligence-based product information push device 500 includes an acquisition module 501, a customer maturity assessment module 502, a product preference analysis module 503, and a screening module 504. The functional modules are described in detail as follows: Acquisition module 501, used to acquire user's personal information and behavioral data associated with the user and business products; A customer maturity evaluation module 502 is configured to input the behavior data into a customer maturity evaluation model to obtain the customer maturity corresponding to the user, wherein the customer maturity is used to represent the user's acceptance of the business product; The product preference analysis module 503 is configured to determine the user's preference for the business products involved in the product browsing records based on the personal information and the product browsing records in the behavioral data if the customer retention maturity corresponding to the user is greater than or equal to a preset threshold; The screening module 504 is configured to screen target business products that need to be recommended to the user from the business products involved in the product browsing records according to the preference level, so as to push product information of the target business products to the user terminal.

[0064] In one embodiment, the apparatus 500 is further configured to: Generate marketing copy based on the personal information and product information of the target business product; Push the marketing copy to the user terminal.

[0065] In one embodiment, the product preference analysis module is specifically configured to: Obtain product information of the business products involved in the product browsing record; Inputting the personal information and the product information of the business product into a purchase probability prediction model to predict the probability value of the user purchasing the business product, and using the probability value as the preference level; The purchase probability prediction model includes an input layer, an intermediate layer, a neural collaborative filtering layer, and an output layer. The input layer is used to extract a user feature vector from the personal information and a product feature vector from the product information of the business product. The intermediate layer includes a plurality of first fully connected layers, wherein the number of neurons in the plurality of first fully connected layers decreases along the data transmission direction of the intermediate layer; The output layer includes a second fully connected layer with a random dropout mechanism turned on and an activation function.

[0066] In one embodiment, there is a multiple relationship between the number of neurons in two adjacent first fully connected layers.

[0067] In one embodiment, the product preference analysis module is specifically configured to: Obtain product information of the business products involved in the product browsing record; Inputting the personal information and the product information of the business product into a purchase probability prediction model to predict the probability value of the user purchasing the business product; Determine the proportion of the number of views corresponding to the business product in the total number of views corresponding to the product browsing records; The probability value and the proportion are multiplied by corresponding preset weights respectively, and the sum is calculated to obtain the sum value as the preference degree.

[0068] In one embodiment, the screening module is specifically configured to: The business product with the highest preference degree is selected from all the business products involved in the product browsing records as the target business product.

[0069] In one embodiment, the apparatus is further configured to train the customer maturity assessment model by: Extracting time series data corresponding to preset features from the historical behavior data of different sample users, wherein the preset features include the activity level of the sample users within a preset time period and / or the frequency of visits of the sample users to business product-related pages; The customer maturity assessment model is trained using the time series data, and the customer maturity assessment model includes an LSTM model.

[0070] The present invention provides a product information push device based on artificial intelligence. First, based on the behavioral data of different users, the customer maturity of different users is evaluated, and then the customer maturity is used to screen the users. For users whose customer maturity is greater than or equal to a preset threshold, product preference analysis is performed. In this way, when facing a large customer base, users with a low acceptance of insurance products can be preferentially screened out to achieve targeted recommendations of insurance products, saving computing resources and storage resources for product information push and even the entire product recommendation process, and to a certain extent improving the efficiency of product information push and even product recommendation for large customer groups. In the insurance product recommendation scenario, the present invention can quickly analyze the product preferences of users who are more receptive to insurance products, and screen out target insurance products that meet user needs based on the user's preference for browsed products, thereby improving the overall effect of insurance product recommendation for large customer groups.

[0071] The specific limitations of the AI-based product information push device can be found in the limitations of the AI-based product information push method above and will not be repeated here. Each module in the aforementioned AI-based product information push device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the aforementioned modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so that the processor can call and execute the operations corresponding to each of the aforementioned modules.

[0072] In one embodiment, Figure 6 As shown, a computer device 600 is provided, including a memory 601, a processor 602, and a computer program stored in the memory 601 and executable on the processor 602. When the processor executes the computer program, the steps of the artificial intelligence-based product information push method described in any of the above embodiments are implemented.

[0073] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the artificial intelligence-based product information push method described in any of the above embodiments are implemented.

[0074] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant description in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0075] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0076] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0077] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A product information push method based on artificial intelligence, characterized in that: The method comprises: Obtaining the user's personal information and the user's behavior data associated with the business product; Inputting the behavior data into a customer maturity evaluation model to obtain the customer maturity corresponding to the user, wherein the customer maturity is used to characterize the user's acceptance of the business product; If the customer maturity corresponding to the user is greater than or equal to a preset threshold, determining the user's preference for the business product involved in the product browsing record based on the personal information and the product browsing record in the behavior data; According to the degree of preference, a target business product that needs to be recommended to the user is screened out from the business products involved in the product browsing record, so as to push product information of the target business product to the user terminal.

2. The method according to claim 1, characterized in that The determining, based on the personal information and the product browsing records in the behavior data, the user's preference for the business product involved in the product browsing records includes: Obtain product information of the business product involved in the product browsing record; Inputting the personal information and the product information of the business product into a purchase probability prediction model to predict the probability value of the user purchasing the business product, and using the probability value as the preference degree; The purchase probability prediction model includes an input layer, an intermediate layer, a neural collaborative filtering layer, and an output layer, wherein the input layer is used to extract a user feature vector from the personal information and a product feature vector from the product information of the business product; The intermediate layer includes a plurality of first fully connected layers, wherein the number of neurons in the plurality of first fully connected layers decreases along the data transmission direction of the intermediate layer; The output layer includes a second fully connected layer with a random dropout mechanism turned on and an activation function.

3. The method according to claim 2, characterized in that There is a multiple relationship between the numbers of neurons in two adjacent first fully connected layers.

4. The method according to claim 1, characterized in that The determining, based on the personal information and the product browsing records in the behavior data, the user's preference for the business product involved in the product browsing records includes: Obtain product information of the business product involved in the product browsing record; Inputting the personal information and the product information of the business product into a purchase probability prediction model to predict the probability value of the user purchasing the business product; Determine the proportion of the number of views corresponding to the business product in the total number of views corresponding to the product browsing records; The probability value and the proportion are multiplied by the corresponding preset weights respectively, and the sum is taken as the preference degree.

5. The method according to any one of claims 1 to 4, characterized in that: The step of screening out target business products to be recommended to the user from the business products involved in the product browsing records according to the preference degree includes: The business product with the highest preference degree is selected from all the business products involved in the product browsing record as the target business product.

6. The method according to any one of claims 1 to 4, characterized in that: The customer maturity assessment model is trained in the following way: Extracting time series data corresponding to preset features from historical behavior data of different sample users, wherein the preset features include the activity of the sample users within a preset time period and / or the frequency of visits of the sample users to business product-related pages; The customer maturity assessment model is trained using the time series data, and the customer maturity assessment model includes an LSTM model.

7. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Generate marketing copy based on the personal information and product information of the target business product; The marketing copy is pushed to the user terminal.

8. A product information push device based on artificial intelligence, characterized in that: The device comprises: An acquisition module, used to acquire the user's personal information and the behavior data associated with the user and the business product; A customer maturity evaluation module, used to input the behavior data into a customer maturity evaluation model to obtain the customer maturity corresponding to the user, wherein the customer maturity is used to characterize the user's acceptance of the business product; A product preference analysis module, for determining the user's preference for the business product involved in the product browsing record based on the personal information and the product browsing record in the behavior data if the customer maturity corresponding to the user is greater than or equal to a preset threshold; The screening module is used to screen out target business products that need to be recommended to the user from the business products involved in the product browsing records according to the preference degree, so as to push the product information of the target business products to the user terminal.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the product information push method based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the product information push method based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

Images