Commodity recommendation method, system, terminal and computer readable storage medium

Abstract

This invention discloses a product recommendation method, system, terminal, and storage medium. The method includes: obtaining the main target customer group for the product based on the product's consumption characteristic data; inputting the purchasing behavior data of the main target customer group into a trained scenario mining model to obtain the product's scenario elements, the weights of the scenario elements, the element features of the scenario elements, and the scores of the element features; determining the current target scenario elements and current target element features of the product based on the scenario information of the current target customer and the product's scenario elements; calculating the current recommendation score of the product based on the weights of the scenario elements, the scores of the element features, the current target scenario elements, and the target element features; and generating a product recommendation list for the target customer based on the current recommendation score. This invention solves the problem that existing operator recommendation systems recommend increasingly more content to customers, making it impossible to accurately identify the current needs of customers.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a product recommendation method, system, terminal, and computer-readable storage medium. Background Technology

[0002] In the context of personalized and precision marketing, traditional marketing models are facing significant challenges. With the digital transformation of enterprises, traditional marketing can no longer meet increasingly personalized consumer demands. To cater to the consumption psychology of as many customers as possible and enhance the user experience, operators need to build precision marketing capabilities and provide more personalized services. Currently, operators mainly define customer tags based on customers' historical online and offline behavior data and match products according to these tags to achieve precision marketing. However, as the number of customer tags increases, existing operator recommendation systems are recommending more and more content to customers, making it difficult to accurately identify current customer needs or rank recommendations based on the customer's context. Summary of the Invention

[0003] The main objective of this invention is to propose a product recommendation method, system, terminal, and computer-readable storage medium, aiming to solve the problem that existing operators are recommending more and more content to customers, making it impossible to accurately identify the current needs of customers.

[0004] To achieve the above objectives, the present invention provides a product recommendation method, comprising the following steps:

[0005] Based on the consumer characteristic data corresponding to the product, identify the main target customer group for the product;

[0006] The purchase behavior data of the main marketing customer groups are input into the trained scene mining model to perform scene mining, so as to obtain the scene elements of the product, the weight of the scene elements, the element features of the scene elements, and the scores of the element features.

[0007] Obtain scenario information of the current customer to be marketed to;

[0008] Based on the scene information and the scene elements of the product, determine the current target scene element and the current target element features of the product;

[0009] The current recommendation score of the product is calculated based on the weight of the scene element, the score of the element feature, the current target scene element of the product, and the current target element feature.

[0010] Based on the current recommendation rating of the products, a product recommendation list is generated for the customer to be promoted to.

[0011] Optionally, the step of inputting the purchase behavior data of the main marketing customer group into a trained scene mining model to perform scene mining, in order to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features, includes:

[0012] Feature extraction is performed on the purchasing behavior data of the main marketing customer groups to obtain the purchasing behavior characteristics of the main marketing customer groups;

[0013] The purchase behavior features are input into a trained scene mining model to perform scene mining, and the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features are obtained.

[0014] Optionally, the product recommendation method further includes:

[0015] The system retrieves the latest purchasing behavior data of the main marketing customer groups at preset intervals to adjust the model parameters of the trained scenario mining model and optimize it.

[0016] Optionally, before the step of inputting the purchase behavior data of the main marketing customer group into the trained scene mining model to perform scene mining to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features, the following method is further included:

[0017] Acquire purchase behavior data of the main marketing customer groups as training positive samples, and acquire non-purchase behavior data of the main marketing customer groups as training negative samples;

[0018] The scene mining model to be trained is trained using training positive samples and training negative samples to obtain a trained scene mining model.

[0019] Optionally, before the step of inputting the purchase behavior features into a trained scene mining model to perform scene mining and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features, the following method is further included:

[0020] Normalize the purchasing behavior characteristics of the main marketing customer groups;

[0021] The step of inputting the purchase behavior features into a trained scene mining model to perform scene mining and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features includes:

[0022] The normalized purchase behavior features are input into the trained scene mining model to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

[0023] Optionally, the step of obtaining the main marketing customer group of the product based on the consumption characteristic data corresponding to the product includes:

[0024] Based on preset feature filtering rules, extract the first customer salient feature from the consumption feature data of the customers who purchased the product;

[0025] Based on the first customer's significant characteristics, the main target customer group for the product is selected.

[0026] Optionally, the step of obtaining the main marketing customer group of the product based on the consumption characteristic data corresponding to the product includes:

[0027] The consumption characteristic data of the customers who purchased the product are input into the trained feature extraction model to obtain the second customer salient features;

[0028] Based on the second set of significant customer characteristics, the main target customer groups for the products are selected.

[0029] To achieve the above objectives, the present invention also provides a product recommendation system, the system comprising:

[0030] The customer group mining module is used to identify the main marketing customer groups of a product based on the consumption characteristic data corresponding to the product.

[0031] The scene element mining module is used to input the purchase behavior data of the main marketing customer groups into the trained scene mining model to perform scene mining, so as to obtain the scene elements of the product, the weight of the scene elements, the element features of the scene elements, and the scores of the element features.

[0032] The customer scenario acquisition module is used to acquire scenario information of the current customer to be marketed to;

[0033] The scene matching module is used to determine the current target scene element and current target element features of the product based on the scene information and the scene elements of the product.

[0034] The recommendation rating calculation module is used to calculate the current recommendation rating of the product based on the weight of the scene element, the score of the element feature, the current target scene element and the current target element feature of the product.

[0035] The recommendation list generation module is used to generate a product recommendation list for the customer to be promoted to, based on the current recommendation rating of the product.

[0036] To achieve the above objectives, the present invention also provides a terminal, the terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the product recommendation method as described above.

[0037] To achieve the above objectives, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the product recommendation method described above.

[0038] This invention proposes a product recommendation method, system, terminal, and computer-readable storage medium. The method involves: identifying the main target customer group for a product based on its corresponding consumer characteristic data; inputting the purchasing behavior data of this main target customer group into a trained scenario mining model to perform scenario mining, thereby obtaining scenario elements, weights of scenario elements, element features, and scores of element features; acquiring scenario information of the current customer to be recommended to; determining the current target scenario element and current target element features of the product based on the scenario information and the product's scenario elements; calculating the current recommendation score of the product based on the weights of the scenario elements, the scores of the element features, and the current target scenario elements and current target element features; and generating a product recommendation list for the customer to be recommended to based on the current recommendation score. Therefore, the product recommendation list built based on the mining results of the target customer group's scenario, scenario weights, elements in the scenario, and the corresponding scores of the elements can better meet the needs of the customer's current scenario, improving the targeting and accuracy of product recommendations to users. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the hardware operating environment involved in the embodiments of the present invention;

[0040] Figure 2 This is a flowchart illustrating the first embodiment of the product recommendation method of the present invention;

[0041] Figure 3 This is a detailed flowchart of step S20 in the first embodiment of the product recommendation method of the present invention;

[0042] Figure 4 This is a detailed flowchart of step S10 in the second embodiment of the product recommendation method of the present invention;

[0043] Figure 5 This is a detailed flowchart of step S10 in the third embodiment of the product recommendation method of the present invention;

[0044] Figure 6 This is a schematic diagram of the functional modules of the product recommendation system of the present invention.

[0045] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0046] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0047] Please refer to Figure 1 , Figure 1 This is a schematic diagram of the hardware structure of the terminal provided in various embodiments of the present invention. The terminal includes components such as a communication module 01, a memory 02, and a processor 03. Those skilled in the art will understand that... Figure 1 The terminal shown may also include more or fewer components than illustrated, or combine certain components, or have different component arrangements. The processor 03 is connected to both the memory 02 and the communication module 01. The memory 02 stores a computer program, which is simultaneously executed by the processor 03.

[0048] The communication module 01 can connect to external devices via a network. The communication module 01 can receive data from external devices and can also send data, instructions, and information to the external devices, which can be electronic devices such as mobile phones, tablets, laptops, and desktop computers.

[0049] Memory 02 can be used to store software programs and various data. Memory 02 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (inputting the purchasing behavior data of the main marketing customer group into a trained scene mining model to perform scene mining to obtain scene elements of the product, the weight of the scene elements, the element features of the scene elements, and the scores of the element features), etc.; the data storage area may store data or information created based on the use of the terminal, etc. In addition, memory 02 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0050] Processor 03 is the control center of the terminal. It connects various parts of the terminal via various interfaces and lines. By running or executing software programs and / or modules stored in memory 02, and by calling data stored in memory 02, it performs various functions and processes data, thereby providing overall monitoring of the terminal. Processor 03 may include one or more processing units; preferably, processor 03 may integrate an application processor and a modem processor. The application processor mainly handles the operating system, user interface, and applications, while the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into processor 03.

[0051] although Figure 1Not shown, but the above terminal may also include a circuit control module, which is used to connect to the mains power supply to realize power control and ensure the normal operation of other components.

[0052] Those skilled in the art will understand that Figure 1 The terminal structure shown does not constitute a limitation on the terminal and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0053] Based on the above hardware structure, various embodiments of the method of the present invention are proposed.

[0054] Reference Figure 2 In a first embodiment of the product recommendation method of the present invention, the product recommendation method includes the following steps:

[0055] Step S10: Based on the consumer characteristic data corresponding to the product, obtain the main marketing customer group of the product;

[0056] In this solution, the terminal executing the product recommendation method can obtain consumption characteristic data of customers who have purchased products from local devices or other devices, such as from a database server. This consumption characteristic data can include customer information and user behavior information related to product usage. Customer information includes age, gender, education level, income level, etc. User behavior information includes the number of times, frequency, and usage patterns of product usage. For example, if the product is a service package, the user behavior information would be the customer's usage pattern of the service package. The terminal executing the product recommendation method extracts several significant consumption characteristics from this data and then uses these characteristics to infer the main marketing customer group corresponding to the product. For example, if the product is a voice upgrade from 8 to 18 yuan service package, the significant characteristics of this service package include: the current customer's basic voice module subscription is 8 yuan; the customer's usual voice consumption demand is low; and the customer has exceeded the voice usage limit at least once in the past three months without exceeding 10 yuan. Customers with these specific characteristics belong to the main marketing customer group corresponding to the voice upgrade from 8 to 18 yuan service package.

[0057] Step S20: Input the purchase behavior data of the main marketing customer group into the trained scene mining model to perform scene mining, so as to obtain the scene elements of the product, the weight of the scene elements, the element features of the scene elements, and the scores of the element features.

[0058] After identifying the primary target customer group for a product, the terminal executing the product recommendation method can obtain purchase behavior data corresponding to the purchase of the product by the primary target customer group from local storage or other devices. This purchase behavior data includes information such as the time and channel through which the customer purchased the product, whether the purchase involved performance evaluations, and whether there were any incentives. The terminal executing the product recommendation method then uses a pre-trained scenario mining model based on a factorization machine algorithm to extract scenario elements and feature characteristics that influence customer purchases from this purchase behavior data. Corresponding weights are assigned to the extracted scenario elements, and corresponding scores are assigned to the feature characteristics.

[0059] like Figure 3 As shown, step S20 specifically includes the following steps:

[0060] Step S21: Extract features from the purchase behavior data of the main marketing customer groups to obtain the purchase behavior characteristics of the main marketing customer groups.

[0061] Step S22: Input the purchase behavior features into the trained scene mining model to perform scene mining, and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

[0062] The terminal will extract the purchase behavior characteristics of the main marketing customer groups from the purchase behavior data of the main marketing customer groups. For example, if the purchase behavior data of a user who purchased the voice 8 to 18 service package is that he purchased the voice 8 to 18 service package through online channels in the middle of the month, then the purchase behavior characteristics extracted from this data will be two features: purchase time - middle of the month and purchase channel - online channel.

[0063] The terminal will input the purchasing behavior characteristics of the main marketing customer groups into the trained scene mining model. The FM (Factorization Machine) algorithm is a machine learning algorithm based on matrix factorization proposed by Steffen Rendle. It can perform regression and binary classification prediction. Its characteristic is that it considers the interaction between features and is a non-linear model. The expression of the FM algorithm is as follows:

[0064]

[0065] W0 is the bias term, n represents the number of features in the sample, and X... i Let Wi and W be the values ​​of the i-th feature. j These are model parameters.

[0066] The trained scenario mining model analyzes the correlation between these purchasing behavior features and the likelihood of user purchases to obtain the scenario elements of the product, the weights of the scenario elements, the element features of the scenario elements, and the scores of the element features. Among multiple scenario elements, the larger the weight of the scenario element, the greater the correlation between the scenario element and the user's successful purchase, and the smaller the weight, the smaller the correlation. Similarly, for the scores of multiple element feature values ​​within the same scenario element, the larger the score, the greater the correlation between the element feature and the user's successful purchase, and the smaller the score, the smaller the correlation.

[0067] In this embodiment, the training process of the trained scene mining model is as follows:

[0068] Step S80: Obtain purchase behavior data of the main marketing customer group as training positive samples, and obtain non-purchase behavior data of the main marketing customer group as training negative samples.

[0069] Step S90: Train the scene mining model to be trained based on the training positive samples and training negative samples to obtain the trained scene mining model.

[0070] The terminal acquires purchase behavior data from the main marketing customer group as positive training samples, while the negative training samples are non-purchase behavior data from the main marketing customer group. This includes data on customers browsing or being recommended products but not purchasing them, such as the time the customer browsed or was recommended a product, the channel used to browse the product, the recommendation channel, whether the product was involved in performance evaluations, and whether there were any rewards involved. The terminal iteratively inputs the positive and negative samples into the scene mining model using a pre-set loss function. It then iteratively solves the loss function of the scene mining model using stochastic gradient descent to achieve the model parameters that meet a pre-set convergence condition. This pre-set convergence condition can be the number of iteration stops or a loss function threshold. The model parameters that meet the pre-set convergence condition are used as the final parameters of the trained scene mining model, thus completing the training of the scene mining model.

[0071] Due to the differences in purchase behavior characteristics across different types, directly inputting the purchase behavior characteristics extracted by the terminal into the trained scene mining model may affect the weights and scores of the model's output. To prevent the differences in purchase behavior characteristics across different types from affecting the weights and scores, in one embodiment, step S22 is preceded by:

[0072] Step S23: Normalize the purchasing behavior characteristics of the main marketing customer groups;

[0073] Step S22 includes:

[0074] Step S221: Input the normalized purchase behavior features into the trained scene mining model to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

[0075] Before the terminal inputs the purchase behavior features into the trained scene mining model to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features, the terminal will normalize the extracted purchase behavior features of the main marketing customer groups to obtain normalized purchase behavior features. Then, the normalized purchase behavior features are used as input data to the trained scene mining model to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

[0076] To ensure that recommendations based on the mined scenes, scene weights, elements within the scenes, and their corresponding scores better meet the latest customer needs, this method also performs real-time optimization of the trained scene mining model. The specific optimization process of the scene mining model includes:

[0077] Step S70: Obtain the latest purchase behavior data of the main marketing customer groups at preset intervals in order to adjust the model parameters of the trained scenario mining model and optimize the trained scenario mining model.

[0078] The terminal will acquire the latest purchase behavior data of the main marketing customer groups at preset intervals. For example, for mobile phone plans, the latest purchase behavior data of customers can be collected every month or half a year. The latest purchase behavior data is analyzed to extract the latest purchase behavior characteristics of customers. The latest purchase behavior characteristics are input into the scenario mining model to adjust the model parameters of the scenario mining model and update the scenario mining model regularly.

[0079] Step S30: Obtain scenario information of the current customer to be marketed to;

[0080] Step S40: Based on the scene information and the scene elements of the product, determine the current target scene element and the current target element features of the product;

[0081] The terminal obtains the current scenario information of the customer to be marketed to through front-end data, and the user's channel is online. Based on the current scenario information of the customer to be marketed to and the product scenario elements determined by the scenario mining model, the terminal determines the current target scenario elements and current target element characteristics of the product. For example, if the current time is April 15th, and the customer to be marketed to is browsing products through the APP, the front-end will record the user's APP login data. The terminal can use this data to determine the user's current scenario information, including the time of April 15th. Therefore, the terminal will determine the current target scenario elements of the product, including time and channel, and determine the current target element characteristics, including mid-month and online channel.

[0082] Step S50: Calculate the current recommendation score of the product based on the weight of the scene element, the score of the element feature, the current target scene element of the product, and the current target element feature.

[0083] Step S60: Generate a product recommendation list for the customer to be promoted to based on the current recommendation rating of the product.

[0084] After the terminal determines the current target scene elements and element features of the product, it will query the weights of the current target scene elements and the scores of the current element features based on the weights of the scene elements and the scores of the current element features determined by the previous scene mining model. Then, it will input the weights of the current target scene elements and the scores of the current element features into the scoring calculation formula to calculate the current recommended score of the product. The scoring calculation formula is C=w1*z1+w2z2+...+w n z n , where w n z represents the weight of the nth target scene element of the product. n This is the score corresponding to the feature of the target element in the nth scene element of the current product.

[0085] After calculating the current recommendation rating of each product, the terminal will select a preset number of products to add to the product recommendation list in descending order of their current recommendation ratings, forming a product recommendation list for customers to be promoted. The products in this product recommendation list are arranged in descending order of their current recommendation ratings.

[0086] This example identifies the main target customer group for a product based on its corresponding consumer characteristic data. The purchasing behavior data of this main target customer group is then input into a trained scenario mining model to perform scenario mining, obtaining scenario elements, their weights, features, and scores. The example also obtains scenario information for the current customer to be marketed to. Based on this scenario information and the product's scenario elements, the example determines the product's current target scenario elements and features. Based on the weights of the scenario elements, the scores of the features, and the current target scenario elements and features, the example calculates the product's current recommendation score. Finally, based on the product's current recommendation score, a product recommendation list is generated for the customer to be marketed to. This results in a product recommendation list built based on the mining results of the target customer group's scenario, scenario weights, elements within the scenario, and their corresponding scores, which better matches the needs of the customer's current scenario, improving the targeting and accuracy of product recommendations to users.

[0087] Further, please refer to Figure 4 , Figure 4 To provide a second embodiment of the product recommendation method of this application based on the first embodiment, in this embodiment, step S10 includes:

[0088] Step S11: Extract the first customer salient feature from the consumption feature data of the purchasing customer corresponding to the product according to the preset feature filtering rules;

[0089] Step S12: Based on the first customer's significant characteristics, filter the main marketing customer groups for the product.

[0090] In this embodiment, after the terminal collects the consumption characteristic data of purchasing customers, it cleans the collected consumption characteristic data, extracts consumption characteristic fields, and determines the customer characteristics corresponding to the consumption characteristic fields. For example, if the extracted consumption characteristic field is "monthly voice usage overage fee of 8 yuan," and the preset voice usage overage fee range corresponding to the customer characteristic of "overage fee not exceeding 10 yuan" is 0 to 10 yuan, then one characteristic of this customer can be determined as "overage fee not exceeding 10 yuan." The terminal extracts all customer characteristics of all purchasing customers, and then filters out multiple customer salient characteristics from the customer characteristics according to preset filtering rules. For example, the extracted customer characteristics are sorted according to the frequency of occurrence of the customer characteristics, and the customer characteristics ranked higher than a preset rank are selected as the first customer salient characteristics. Another example is that the preset weight corresponding to each type of customer characteristic is multiplied by the frequency of occurrence of the customer characteristic to obtain the salient score of the customer characteristic, and then the customer characteristics are sorted according to the score, and the customer characteristics ranked higher than a preset rank are selected as the first customer salient characteristics. The terminal obtains the first customer salient characteristics and filters out customers with these first customer salient characteristics to form the main marketing customer group of the product.

[0091] This embodiment uses manually set screening rules to more accurately extract significant customer characteristics, thereby more precisely identifying the main marketing customer group for the product based on these characteristics.

[0092] Further, please refer to Figure 5 , Figure 5 To provide a third embodiment of the product recommendation method of this application based on the first embodiment, in this embodiment, step S10 includes:

[0093] Step S13: Input the consumption characteristic data of the purchasing customer corresponding to the product into the trained feature extraction model to obtain the second customer salient features;

[0094] Step S14: Based on the second customer's significant characteristics, screen the main marketing customer groups for the product.

[0095] In this embodiment, a trained feature extraction model is installed in the terminal. This model is a classification algorithm based on machine learning, such as Logistic Regression (LR), Decision Tree, or Random Forest. After collecting the consumption characteristic data of purchasing customers, the terminal inputs this data into the trained feature extraction model. The model automatically extracts secondary salient features of the customers. The terminal then uses these secondary salient features to filter out customers who possess them, forming the main target customer group for the product.

[0096] It should be noted that the training samples for the feature extraction model include customer features extracted from the feature data of all customers who have purchased the product as positive samples and customer features extracted from the feature data of all customers who have never purchased the product as negative samples; then the feature extraction model is trained based on the positive and negative samples.

[0097] This embodiment uses a machine learning algorithm model to extract significant customer features more quickly, thereby identifying the main customer base for the product more rapidly and improving overall work efficiency.

[0098] See Figure 6 The present invention also provides a product recommendation system, comprising:

[0099] Customer group mining module 10 is used to obtain the main marketing customer group of the product based on the consumption characteristic data corresponding to the product;

[0100] The scene element mining module 20 is used to input the purchase behavior data of the main marketing customer group into the trained scene mining model to perform scene mining, so as to obtain the scene elements of the product, the weight of the scene elements, the element features of the scene elements, and the scores of the element features.

[0101] The customer scenario acquisition module 30 is used to acquire scenario information of the current customer to be marketed to;

[0102] The scene matching module 40 is used to determine the current target scene element and current target element features of the product based on the scene information and the scene elements of the product;

[0103] The recommendation score calculation module 50 is used to calculate the current recommendation score of the product based on the weight of the scene element, the score of the element feature, the current target scene element and the current target element feature of the product.

[0104] The recommendation list generation module 60 is used to generate a product recommendation list for the customer to be promoted to based on the current recommendation rating of the product.

[0105] Furthermore, the scene element mining module 20 includes:

[0106] The first feature extraction unit 21 is used to extract features from the purchase behavior data of the main marketing customer groups to obtain the purchase behavior features of the main marketing customer groups.

[0107] The scene element mining unit 22 is used to input the purchase behavior features into the trained scene mining model to perform scene mining, and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

[0108] Furthermore, the product recommendation system also includes:

[0109] The optimization module 70 is used to obtain the latest purchase behavior data of the main marketing customer groups at preset intervals in order to adjust the model parameters of the trained scene mining model and optimize the trained scene mining model.

[0110] Furthermore, the product recommendation system also includes:

[0111] The training sample acquisition module 80 is used to acquire purchase behavior data of the main marketing customer groups as training positive samples, and to acquire non-purchase behavior data of the main marketing customer groups as training negative samples.

[0112] Training module 90 is used to train the scene mining model to be trained based on training positive samples and training negative samples, so as to obtain the trained scene mining model.

[0113] Furthermore, the scene element mining module 20 also includes:

[0114] Normalization unit 23 is used to normalize the purchasing behavior characteristics of the main marketing customer groups;

[0115] The scene element mining unit 22 is also used to input the normalized purchase behavior features into the trained scene mining model to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

[0116] Furthermore, the customer segmentation module 10 includes:

[0117] The second feature extraction unit 11 is used to extract the first customer salient feature from the consumption feature data of the purchasing customer corresponding to the product according to the preset feature filtering rules.

[0118] The first customer group mining unit 12 is used to screen the main marketing customer groups of the product based on the first customer's significant characteristics.

[0119] Furthermore, the customer segmentation module 10 includes:

[0120] The third feature extraction unit 13 is used to input the consumption feature data of the purchasing customer corresponding to the product into the trained feature extraction model to obtain the second customer salient feature;

[0121] The second customer group mining unit 14 is used to screen the main marketing customer groups of the product based on the second customer's significant characteristics.

[0122] The present invention also proposes a computer-readable storage medium having a computer program stored thereon. The computer-readable storage medium may be... Figure 1The memory 02 in the terminal may also be at least one of ROM (Read-Only Memory) / RAM (Random Access Memory), magnetic disk, optical disk, etc. The computer-readable storage medium includes several information to enable the terminal to perform the methods described in the various embodiments of the present invention.

[0123] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0124] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0125] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0126] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A product recommendation method, characterized in that, The product recommendation method includes the following steps: Based on the consumer characteristic data corresponding to the product, identify the main target customer group for the product; Feature extraction is performed on the purchasing behavior data of the main marketing customer groups to obtain the purchasing behavior characteristics of the main marketing customer groups; The purchase behavior features are input into a trained scene mining model to perform scene mining, and the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features are obtained. Obtain scenario information of the current customer to be marketed to; Based on the scene information and the scene elements of the product, determine the current target scene element and the current target element features of the product; The current recommendation score of the product is calculated based on the weight of the scene element, the score of the element feature, the current target scene element of the product, and the current target element feature. Based on the current recommendation rating of the products, a product recommendation list is generated for the customer to be promoted to.

2. The product recommendation method according to claim 1, characterized in that, The product recommendation method also includes: The system retrieves the latest purchasing behavior data of the main marketing customer groups at preset intervals to adjust the model parameters of the trained scenario mining model and optimize it.

3. The product recommendation method according to claim 1, characterized in that, Before the step of inputting the purchase behavior data of the main marketing customer group into the trained scene mining model to perform scene mining and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features, the following steps are included: Acquire purchase behavior data of the main marketing customer groups as training positive samples, and acquire non-purchase behavior data of the main marketing customer groups as training negative samples; The scene mining model to be trained is trained using training positive samples and training negative samples to obtain a trained scene mining model.

4. The product recommendation method according to claim 1, characterized in that, Before the step of inputting the purchase behavior features into the trained scene mining model to perform scene mining and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features, the following steps are included: Normalize the purchasing behavior characteristics of the main marketing customer groups; The step of inputting the purchase behavior features into a trained scene mining model to perform scene mining and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features includes: The normalized purchase behavior features are input into the trained scene mining model to obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

5. The product recommendation method according to any one of claims 1 to 4, characterized in that, The step of obtaining the main target customer group for a product based on its corresponding consumer characteristic data includes: Based on preset feature filtering rules, extract the first customer salient feature from the consumption feature data of the customers who purchased the product; Based on the first customer's significant characteristics, the main target customer group for the product is selected.

6. The product recommendation method according to any one of claims 1 to 4, characterized in that, The step of obtaining the main target customer group for a product based on its corresponding consumer characteristic data includes: The consumption characteristic data of the customers who purchased the product are input into the trained feature extraction model to obtain the second customer salient features; Based on the second set of significant customer characteristics, the main target customer groups for the products are selected.

7. A product recommendation system, characterized in that, The product recommendation system includes: The customer group mining module is used to identify the main marketing customer groups of a product based on the consumption characteristic data corresponding to the product. The scene element mining module is used to input the purchase behavior data of the main marketing customer groups into the trained scene mining model to perform scene mining, so as to obtain the scene elements of the product, the weight of the scene elements, the element features of the scene elements, and the scores of the element features. The customer scenario acquisition module is used to acquire scenario information of the current customer to be marketed to; The scene matching module is used to determine the current target scene element and current target element features of the product based on the scene information and the scene elements of the product. The recommendation rating calculation module is used to calculate the current recommendation rating of the product based on the weight of the scene element, the score of the element feature, the current target scene element and the current target element feature of the product. The recommendation list generation module is used to generate a product recommendation list for the customer to be promoted to based on the current recommendation rating of the product. The scene element mining module includes: The first feature extraction unit is used to extract features from the purchase behavior data of the main marketing customer groups to obtain the purchase behavior characteristics of the main marketing customer groups. The scene element mining unit is used to input the purchase behavior features into the trained scene mining model to perform scene mining, and obtain the scene elements of the product, the weights of the scene elements, the element features of the scene elements, and the scores of the element features.

8. A terminal, characterized in that, The terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the product recommendation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the product recommendation method as described in any one of claims 1 to 6.

Images