Product recommendation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a product recommendation method and device, electronic equipment and a storage medium, belongs to the technical field of artificial intelligence, and is suitable for financial science and technology scenes. The method comprises the following steps: acquiring user information of candidate objects; the user information comprises age attributes; classifying the candidate objects based on age attributes to obtain an object group; each object group comprises at least two alternative objects; for each object group, obtaining product operation behavior data of each alternative object; performing product popularity calculation on the product operation behavior data of each alternative object to obtain a group product popularity sequence of each object group and a user product popularity sequence of each alternative object; performing product screening based on the user information, the group product popularity sequence and the user product popularity sequence to obtain a target recommended product sequence of each alternative object; and performing product recommendation on each alternative object based on the target recommendation product sequence. According to the embodiment of the invention, the product recommendation accuracy can be improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and is applicable to the financial technology field, particularly to a product recommendation method and apparatus, electronic device, and storage medium. Background Technology

[0002] Product recommendation is an artificial intelligence technology that uses a customer's personal information and product-related data to filter and push products that the customer may be interested in. Product recommendation can be applied to multiple scenarios. For example, in fintech, it can recommend relevant financial products based on the customer's personal information and product-related data for property insurance, life insurance, and investment products; in health insurance, it can recommend relevant health insurance products based on the customer's personal information and product-related data for health insurance, critical illness insurance, and surgical insurance.

[0003] Currently, when recommending products to customers, sales agents typically select matching products from a product list based on the customer's personal information. However, in real-world scenarios, a communication gap may exist between the sales agent and the customer, making it difficult for the agent to accurately grasp the customer's needs and preferences. This can lead to situations where recommended products do not match the customer's needs, resulting in low accuracy in product recommendations.

[0004] Therefore, improving the accuracy of product recommendations has become an urgent technical problem to be solved. Summary of the Invention

[0005] The main objective of this application is to provide a product recommendation method, apparatus, electronic device, and storage medium, which aims to improve the accuracy of product recommendations.

[0006] To achieve the above objectives, a first aspect of this application proposes a product recommendation method, the method comprising:

[0007] Obtain user information for candidate candidates; wherein, the user information includes age attributes;

[0008] The candidate objects are classified based on the age attribute to obtain object groups; each object group includes at least two candidate objects.

[0009] For each of the aforementioned target groups, obtain product operation behavior data for each of the aforementioned candidate objects;

[0010] The product popularity is calculated for the product operation behavior data of each candidate object to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object;

[0011] Based on the user information, the product popularity sequence of the group, and the product popularity sequence of the user, product screening is performed to obtain the target recommended product sequence for each of the candidate objects;

[0012] Based on the target recommended product sequence, product recommendations are made to each of the candidate objects.

[0013] In some embodiments, the group product popularity sequence includes a group click product sequence and a group search product sequence, and the user product popularity sequence includes a user click product sequence and a user search product sequence; the step of filtering products based on the user information, the group product popularity sequence, and the user product popularity sequence to obtain a target recommended product sequence for each candidate includes:

[0014] The difference between the group click product sequence and the user click product sequence is calculated to obtain the click product attention sequence for each candidate object;

[0015] The difference between the group search product sequence and the user search product sequence is calculated to obtain the search product attention sequence for each of the candidate objects;

[0016] Based on the clicked product attention sequence and the searched product attention sequence, product filtering is performed to obtain the target attention product sequence for each of the candidate objects;

[0017] Based on the user information and the target product sequence, product prediction is performed to obtain the target recommended product sequence for each of the candidate objects.

[0018] In some embodiments, the step of calculating the difference between the group click product sequence and the user click product sequence to obtain the click product attention sequence for each candidate object includes:

[0019] Obtain the name of each product in the sequence of products clicked by the user and the user click order number corresponding to the product name;

[0020] Based on the product name, the product sequence of the group clicks is filtered to obtain the group click sorting number corresponding to the product name;

[0021] The difference between the user click sorting number and the group click sorting number is calculated to obtain the click sorting difference data.

[0022] Based on the click ranking difference data, each product name is sorted to obtain the click product attention sequence for each candidate object.

[0023] In some embodiments, the target product focus sequence includes multiple user-focused products; the step of performing product prediction based on the user information and the target product focus sequence to obtain the target recommended product sequence for each of the candidate objects includes:

[0024] Feature extraction is performed on the user information to obtain user feature data;

[0025] The user characteristic data is used to predict purchasing power to obtain user purchasing power prediction data.

[0026] Obtain the product spending data of the products that the user is interested in;

[0027] Product matching data is obtained by matching the user purchasing power prediction data and the product spending data.

[0028] Based on the product matching data, the products that the user is interested in are filtered to obtain the target recommended product sequence for each of the candidate objects.

[0029] In some embodiments, the step of calculating product popularity based on the product operation behavior data of each candidate object to obtain a group product popularity sequence for each object group and a user product popularity sequence for each candidate object includes:

[0030] Feature extraction is performed on the product operation behavior data to obtain initial behavior data;

[0031] The initial behavioral data is classified to obtain target behavioral data; wherein, the behavioral types of the target behavioral data include click behavior and search behavior;

[0032] For each of the aforementioned behavior types, keywords are extracted from the target behavior data to obtain product keyword data;

[0033] For each of the target groups, the product keyword data is aggregated and sorted based on the behavior type to obtain the group product popularity sequence for each target group;

[0034] For each candidate, the product keyword data is aggregated and sorted based on the behavior type to obtain the user product popularity sequence for each candidate.

[0035] In some embodiments, the step of extracting features from the product operation behavior data to obtain initial behavior data includes:

[0036] The product operation behavior data is subjected to outlier detection and processing to obtain the raw behavior data;

[0037] The original behavioral data is standardized to obtain standard behavioral data;

[0038] The standard behavioral data is subjected to feature filtering to obtain filtered behavioral data;

[0039] The initial behavioral data is obtained by constructing features from the filtered behavioral data.

[0040] In some embodiments, the target recommended product sequence includes multiple candidate recommended products; the step of recommending products to each candidate based on the target recommended product sequence includes:

[0041] Obtain product information of the recommended alternative products;

[0042] Based on the product information, the candidate recommended products are combined to obtain the target recommended product combination;

[0043] Based on the target recommended product portfolio, product recommendations are made to each of the candidate items.

[0044] To achieve the above objectives, a second aspect of this application provides a product recommendation device, the device comprising:

[0045] The user information acquisition module is used to acquire user information of candidate objects; wherein, the user information includes age attributes;

[0046] The user classification module is used to classify the candidate objects based on the age attribute to obtain object groups; each object group includes at least two candidate objects;

[0047] The operation behavior acquisition module is used to acquire product operation behavior data of each of the candidate objects for each of the object groups.

[0048] The product popularity calculation module is used to calculate the product popularity of the product operation behavior data of each candidate object, and obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object.

[0049] The target product filtering module is used to filter products based on the user information, the group product popularity sequence, and the user product popularity sequence to obtain a target recommended product sequence for each candidate.

[0050] The target product recommendation module is used to recommend products to each of the candidate objects based on the target recommended product sequence.

[0051] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0052] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0053] The product recommendation method, apparatus, electronic device, and storage medium proposed in this application obtain user information including age attributes of candidate objects, and classify the candidate objects based on age attributes to obtain object groups. Each object group includes at least two candidate objects. Because different age groups have different consumption preferences, grouping can improve the accuracy and targeting of product recommendations. Next, product operation behavior data of each candidate object in each object group is obtained, which can truly reflect the products that the candidate objects are interested in. Product popularity is calculated on the product operation behavior data of each candidate object to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object. Based on user information, group product popularity sequence, and user product popularity sequence, product screening is performed to obtain the target recommended product sequence for each candidate object. This can take into account both individual user characteristics and group commonalities, thereby improving recommendation accuracy. Finally, products are recommended to each candidate object based on the target recommended product sequence, which can improve user satisfaction and product recommendation success rate, and achieve more effective product promotion and user service. Attached Figure Description

[0054] Figure 1 This is a flowchart of the product recommendation method provided in the embodiments of this application;

[0055] Figure 2 yes Figure 1 The flowchart of step S104 in the process;

[0056] Figure 3 yes Figure 2 The flowchart of step S201 in the text;

[0057] Figure 4 yes Figure 1 The flowchart of step S105 in the process;

[0058] Figure 5 yes Figure 4 The flowchart of step S401 in the text;

[0059] Figure 6 yes Figure 4 The flowchart of step S404 in the document;

[0060] Figure 7 yes Figure 1 The flowchart of step S106 in the process;

[0061] Figure 8 This is a schematic diagram of the product recommendation device provided in the embodiments of this application;

[0062] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0066] First, let's analyze some of the terms used in this application:

[0067] Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0068] Product recommendation is an artificial intelligence technology that uses a customer's personal information and product-related data to filter and push products that the customer may be interested in. Product recommendation can be applied to multiple scenarios. For example, in fintech, it can recommend relevant financial products based on the customer's personal information and product-related data for property insurance, life insurance, and investment products; in health insurance, it can recommend relevant health insurance products based on the customer's personal information and product-related data for health insurance, critical illness insurance, and surgical insurance.

[0069] Currently, when recommending products to customers, sales agents typically select matching products from a product list based on the customer's personal information. However, in real-world scenarios, while sales agents possess extensive sales experience, a communication gap may exist between them and the customer. This can lead to the sales agent being unable to accurately grasp the customer's needs and preferences, potentially resulting in recommended products that do not match the customer's requirements, leading to low accuracy in product recommendations.

[0070] Based on this, embodiments of this application provide a product recommendation method and apparatus, an electronic device, and a storage medium, aiming to improve the accuracy of product recommendations.

[0071] The product recommendation method, apparatus, electronic device, and storage medium provided in this application are specifically illustrated through the following embodiments. First, the XX method in the embodiments of this application is described.

[0072] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0073] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0074] The product recommendation method provided in this application relates to the field of artificial intelligence technology. The product recommendation method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the product recommendation method, but is not limited to the above forms.

[0075] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0076] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0077] Figure 1 This is an optional flowchart of the product recommendation method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S106.

[0078] Step S101: Obtain user information of the candidate; wherein, the user information includes age attribute;

[0079] Step S102: Classify the candidate objects based on the age attribute to obtain object groups; each object group includes at least two candidate objects.

[0080] Step S103: For each target group, obtain product operation behavior data for each candidate target.

[0081] Step S104: Calculate the product popularity of the product operation behavior data of each candidate object to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object.

[0082] Step S105: Based on user information, group product popularity sequence and user product popularity sequence, product screening is performed to obtain the target recommended product sequence for each candidate.

[0083] Step S106: Recommend products to each candidate based on the target recommended product sequence.

[0084] Steps S101 to S106, as illustrated in this embodiment, involve acquiring user information containing age attributes for candidate objects and classifying them based on these attributes to obtain object groups. Each object group includes at least two candidate objects. Since different age groups have different consumption preferences, grouping can improve the accuracy and targeting of product recommendations. Next, product operation behavior data for each candidate object within each object group is acquired, accurately reflecting the products the candidate objects are interested in. Product popularity is calculated for each candidate object's product operation behavior data, resulting in a group product popularity sequence for each object group and a user product popularity sequence for each candidate object. Based on user information, the group product popularity sequence, and the user product popularity sequence, product filtering is performed to obtain a target recommended product sequence for each candidate object. This approach considers both individual user characteristics and group commonalities, improving recommendation accuracy. Finally, product recommendations are made to each candidate object based on the target recommended product sequence, increasing user satisfaction and product recommendation success rate, and achieving more effective product promotion and user service.

[0085] In step S101 of some embodiments, the candidate object refers to a potential user who may become a recipient of product recommendations. User information is a collection of data on various aspects of the candidate object, including but not limited to: age attribute, gender attribute, income attribute, city of residence attribute, industry attribute, education attribute, etc. User information is obtained through various data collection methods, such as user registration information, third-party data authorization, etc.

[0086] In step S102 of some embodiments, the object group is a collection of candidate objects with the same or similar age attributes. Specifically, multiple age ranges can be set, such as 18-25 years old, 26-35 years old, 36-45 years old, etc. By using range classification, candidate objects are classified into different ranges based on age attributes, resulting in multiple object groups. This allows for the analysis of common needs at the group level, facilitating the subsequent development of recommendation strategies for different candidate objects and helping to improve recommendation efficiency.

[0087] In step S103 of some embodiments, for each candidate object in each object group, product operation behavior data of the candidate object is obtained. The product operation behavior data is a record of various operations generated by the candidate object in the past preset time period when using the target platform, such as clicking, searching, purchasing, and collecting. It can be stored in the form of logs. The product operation behavior data truly reflects the candidate object's attention to and demand for the product, providing a basis for subsequent calculation of product popularity and helping to explore potential user needs.

[0088] Specifically, the preset time period needs to be set according to the actual application scenario, such as one week, two weeks, one month, etc. The target platform is the application platform that provides services to customers, such as insurance applications, financial management applications, insurance websites, and healthcare platforms.

[0089] Please see Figure 2 In some embodiments, step S104 may include, but is not limited to, steps S201 to S205:

[0090] Step S201: Extract features from product operation behavior data to obtain initial behavior data;

[0091] Step S202: Classify the initial behavioral data to obtain target behavioral data; wherein, the behavioral types of the target behavioral data include click behavior and search behavior;

[0092] Step S203: For each behavior type, extract keywords from the target behavior data to obtain product keyword data;

[0093] Step S204: For each target group, aggregate and sort the product keyword data based on behavior type to obtain the group product popularity sequence for each target group.

[0094] Step S205: For each candidate object, aggregate and sort the product keyword data based on behavior type to obtain the user product popularity sequence for each candidate object.

[0095] Steps S201 to S205, as illustrated in this embodiment, extract features from product operation behavior data to accurately focus on product-related behavioral information, forming initial behavioral data. Next, the initial behavioral data is categorized to obtain target behavioral data; the target behavioral data includes click behavior and search behavior, which can more intuitively reflect the degree of attention of candidate users to the product. Then, for each behavioral type, keywords are extracted from the target behavioral data to obtain product keyword data. For each target group, the product keyword data is aggregated and sorted based on behavioral type to obtain a group product popularity sequence for each target group, clearly showing the differences in product preferences among different groups. For each candidate user, the product keyword data is aggregated and sorted based on behavioral type to obtain a user product popularity sequence for each candidate user, enabling a deeper understanding of individual user interests and achieving personalized recommendations.

[0096] Please see Figure 3 In some embodiments, step S201 may include, but is not limited to, steps S301 to S304:

[0097] Step S301: Perform outlier detection and processing on the product operation behavior data to obtain the original behavior data;

[0098] Step S302: Standardize the raw behavioral data to obtain standard behavioral data;

[0099] Step S303: Perform feature filtering on the standard behavioral data to obtain filtered behavioral data;

[0100] Step S304: Feature construction is performed on the filtered behavioral data to obtain initial behavioral data.

[0101] Steps S301 to S304, as illustrated in this embodiment, involve first detecting and processing outliers in the product operation behavior data to remove interfering data, ensuring data quality and providing accurate and reliable raw behavior data for subsequent analysis. Next, the raw behavior data is standardized to eliminate the influence of differences in feature units and numerical ranges, resulting in standard behavior data that facilitates unified processing and analysis. Then, the standard behavior data features are filtered to remove irrelevant or redundant features, reducing data dimensionality and computational complexity, resulting in filtered behavior data. Finally, the filtered behavior data features are constructed to generate more valuable new features, yielding initial behavior data that helps improve the accuracy of subsequent product recommendations.

[0102] In step S301 of some embodiments, outliers refer to data that deviates significantly from normal data, possibly due to data entry errors, system malfunctions, or special accidental events. Specifically, statistical methods (such as the 3σ principle, where values ​​exceeding three standard deviations from the mean are considered outliers if the data follows a normal distribution; box plots), machine learning algorithms (such as the Isolation Forest algorithm), etc., can be used to identify outliers from massive amounts of product operation behavior data. Detected outliers are then corrected (e.g., replaced with the mean or zero), deleted, or retained (in special but research-valuable cases, such as extreme user behavior research) based on the actual situation. By detecting and processing outliers in product operation behavior data, data quality can be improved, preventing outliers from misleading subsequent analysis and thus improving the accuracy of product recommendations.

[0103] In step S302 of some embodiments, the original behavioral data after outlier processing may have significant differences in units and numerical ranges. Therefore, it is necessary to standardize the original behavioral data to convert data with different units to the same scale, resulting in standard behavioral data. For example, the age and annual income of the candidates need to be standardized to unify the units. After standardization, the data of each feature has the same units and similar numerical ranges, which facilitates subsequent comparison and analysis.

[0104] Specifically, methods such as Z-score standardization (which transforms the data into a distribution with a mean of 0 and a standard deviation of 1) and Min-Max standardization (which linearly transforms the data to the [0,1] interval) can be used to eliminate differences in the dimensions and numerical ranges between different features.

[0105] In step S303 of some embodiments, the standardized product operation behavior data contains multiple features, but not all features are useful for product recommendation. Therefore, it is necessary to perform feature filtering on the standardized behavior data to select data composed of features that are highly relevant to product recommendation.

[0106] Specifically, preset filtering fields can be determined based on the actual application scenario, and then feature filtering can be performed on the standard behavioral data based on the preset filtering fields to obtain the filtered behavioral features. Alternatively, methods such as filtering (e.g., filtering based on the correlation coefficient between features and the target variable), wrapping methods (e.g., recursive feature elimination, filtering by continuously trying different feature combinations to evaluate model performance), and embedding methods (e.g., decision tree algorithms automatically select important features during training) can be used to select key features from the standard behavioral data to obtain the filtered behavioral data.

[0107] It should be noted that the preset filter fields are fields that are pre-set by business agents or R&D personnel based on business scenarios and have a significant impact on improving the conversion rate of product recommendations. Specific settings need to be tailored to the actual application scenario and are not limited to these.

[0108] In step S304 of some embodiments, the key feature data retained after feature filtering may still be insufficient to comprehensively and accurately describe the behavior of the candidate objects. Therefore, it is necessary to further construct features from the filtered behavioral data to generate new valuable feature data that can more richly and comprehensively reflect the behavioral characteristics of the candidate objects.

[0109] Specifically, multiple fields in the filtering behavior data can be combined to form new features. For example, the age range of candidates can be combined with their occupation, or the age range can be combined with their gender. In addition, mathematical operations (such as adding, subtracting, multiplying, and dividing two features), statistical methods (such as calculating the mean, median, and variance of features), and business logic (such as constructing user shopping preference features based on the categories of goods browsed and the time of purchase) can be used to generate new features from the filtering behavior data.

[0110] It is understood that the embodiments shown in steps S303 to S304 above are for feature engineering of standard behavioral data to obtain initial behavioral data that can characterize the features of candidate objects, which helps to understand the behavioral characteristics of candidate objects more accurately and provides data support for subsequent product recommendations.

[0111] In step S202 of some embodiments, after obtaining the initial behavior data, the initial behavior data is classified according to the behavior type to obtain the target behavior data, wherein the behavior type includes click behavior and search behavior.

[0112] In step S203 of some embodiments, for each behavior type, keywords are extracted from the target behavior data according to a preset product keyword library to obtain product keyword data corresponding to each behavior type.

[0113] For example: the product keywords corresponding to the click behavior of object 1 include product 1, product 2, product 3, and product 5; the product keywords corresponding to the search behavior of object 1 include product 2, product 3, and product 5.

[0114] It should be noted that the preset product keyword library includes keywords related to the names of products or services offered by the target platform, including but not limited to: health insurance products (such as medical insurance, critical illness insurance, outpatient insurance, and surgical insurance), life insurance products (such as whole life insurance, participating insurance, and annuity insurance), property insurance products (such as home insurance, car insurance, fire insurance, and theft insurance), investment and wealth management products (such as stock funds, bond funds, and index funds), accident insurance products (accident injury insurance, accident medical insurance, and accident disability insurance), savings products (such as time deposits, education savings, and retirement savings), loan products (home loans, car loans, education loans, and personal loans), retirement planning products (such as personal retirement accounts, deferred annuities, and occupational annuities), and insurance service products (such as insurance comparison tools, insurance calculators, insurance recommendations, policy management, and claims services). Specifically, the settings need to be tailored to the actual application scenario and are not limited to this.

[0115] In step S204 of some embodiments, for each object group, the product keyword data of all candidate objects in the object group are aggregated and sorted according to behavior type to obtain the group product popularity sequence. Therefore, the group product popularity sequence includes the group click product sequence and the group search product sequence. The group click product sequence is obtained by aggregating and sorting the product keyword data corresponding to the click behavior, and the group search product sequence is obtained by aggregating and sorting the product keyword data corresponding to the search behavior.

[0116] In step S205 of some embodiments, for each candidate object, the product keyword data of the candidate object is aggregated and sorted according to the behavior type to obtain the user product popularity sequence of the candidate object. Therefore, the user product popularity sequence includes the user click product sequence and the user search product sequence. The user click product sequence is obtained by aggregating and sorting the product keyword data corresponding to the click behavior, and the user search product sequence is obtained by aggregating and sorting the product keyword data corresponding to the search behavior.

[0117] Specifically, the sorting can be done by numerical values ​​from largest to smallest, or by numerical values ​​from smallest to largest, and is not limited to this.

[0118] Please see Figure 4 In some embodiments, step S105 may include, but is not limited to, steps S401 to S404:

[0119] Step S401: Calculate the difference between the group click product sequence and the user click product sequence to obtain the click product attention sequence for each candidate object;

[0120] Step S402: Calculate the difference between the group search product sequence and the user search product sequence to obtain the search product attention sequence for each candidate object;

[0121] Step S403: Based on the clicked product attention sequence and the searched product attention sequence, filter products to obtain the target attention product sequence for each candidate.

[0122] Step S404: Based on user information and the target product sequence, perform product prediction to obtain the target recommended product sequence for each candidate.

[0123] Steps S401 to S404, as illustrated in this embodiment, involve calculating the differences between the group click product sequence and the user click product sequence. This accurately identifies user click preferences and group differences, yielding a click product interest sequence for each candidate and clarifying user click needs. Calculating the differences between the group search product sequence and the user search product sequence allows for the mining of user search preferences, obtaining a search product interest sequence for each candidate and understanding user search intent. Product filtering based on the click product interest sequence and search product interest sequence comprehensively considers click and search interest, resulting in a target interest product sequence more closely aligned with the interests of each candidate. Finally, combining user information with the target interest product sequence for product prediction fully considers user characteristics, yielding a target recommended product sequence that accurately meets the personalized needs of candidate users, improving product recommendation accuracy and user satisfaction.

[0124] In step S401 of some embodiments, by performing difference calculations on the product sequence clicked by the group and the product sequence clicked by the user, such as comparing the frequency and order of products in the two sequences, the parts in the product sequence clicked by the user that are different from those in the product sequence clicked by the group are identified, and the clicked product attention sequence of each candidate object is obtained. This can accurately locate the personalized click preferences of the candidate object, avoid imposing the general behavior of the group on the candidate object, and provide a basis for subsequent accurate recommendations.

[0125] Please see Figure 5 In some embodiments, step S401 may also include, but is not limited to, steps S501 to S504:

[0126] Step S501: Obtain each product name in the product sequence clicked by the user and the user click sort number corresponding to the product name;

[0127] Step S502: Filter the product sequence of the group clicks based on the product name to obtain the group click sorting number corresponding to the product name;

[0128] Step S503: Calculate the difference between the user click sorting number and the group click sorting number to obtain the click sorting difference data;

[0129] Step S504: Sort each product name based on the click sorting difference data to obtain the click product attention sequence for each candidate object.

[0130] Steps S501 to S504, as illustrated in this embodiment, accurately determine the order of product preferences among potential users by obtaining the names of each product in the user-clicked product sequence and their corresponding user click ranking numbers. By filtering the group's clicked product sequence based on product names and obtaining the group click ranking numbers corresponding to those product names, the general level of attention to the products within the target group can be understood. Next, the difference between the user click ranking numbers and the group click ranking numbers is calculated to obtain click ranking difference data, clearly showing the difference between user attention and group attention. Finally, each product name is sorted based on the click ranking difference data to obtain the clicked product attention sequence for each potential user. This integrates individual preferences and group trends, helping to more accurately grasp the unique needs of potential users and providing a strong basis for personalized recommendations.

[0131] In step S501 of some embodiments, the user click sorting number is a number assigned after sorting the candidate objects according to the number of times they clicked the product corresponding to the product name. The larger the number, the more times the candidate object clicked the product.

[0132] In step S502 of some embodiments, the group click sorting number is a number assigned after sorting the number of clicks on the product corresponding to the product name by all candidate objects, wherein the larger the number, the more times the product is clicked.

[0133] In step S503 of some embodiments, the difference between the user's click sorting number and the group's click sorting number is calculated to obtain click sorting difference data, which is used to quantify the difference between individual users and groups in the order of product attention.

[0134] For example: For a certain product, the product name is Product 1. The user click ranking number of Product 1 in the user click product sequence of candidate object A is 2. The group click ranking number of Product 1 in the group click product sequence of object group is 7. Calculate the difference between the user click ranking number and the group click ranking number, that is, 2 minus 7, which is -5. Then take the absolute value, which is 5. That is, the click ranking difference data is 5.

[0135] In step S504 of some embodiments, each product name is sorted from largest to smallest according to the value of the sorting difference to obtain the clicked product attention sequence of each candidate object, which can reflect the unique attention of users to products relative to the group and provide a more accurate basis for personalized recommendations.

[0136] In some embodiments, step S402 may include, but is not limited to, the following steps:

[0137] Get the name of each product in the user's search product sequence and the user's search sort number corresponding to the product name;

[0138] Based on the product name, the product sequence of the group search is filtered to obtain the group search ranking number corresponding to the product name;

[0139] The search ranking difference data is obtained by calculating the difference between the user search ranking number and the group search ranking number.

[0140] The search ranking difference data is used to sort each product name to obtain the search product attention sequence for each candidate.

[0141] By obtaining the names of each product in a user's search product sequence and their corresponding user search ranking numbers, we can accurately grasp the order of product preferences among potential users. Filtering the group's search product sequence based on product names yields the corresponding group search ranking numbers, revealing the general level of interest in products within the target group. Next, calculating the difference between the user search ranking number and the group search ranking number provides search ranking difference data, clearly showing the difference between user and group attention. Finally, sorting each product name based on the search ranking difference data yields the search product attention sequence for each potential user. This comprehensive approach, combining individual preferences and group trends, helps to more accurately grasp the unique needs of potential users, providing strong evidence for personalized recommendations.

[0142] Specifically, the specific implementation of step S402 is basically the same as the specific implementation shown in steps S501 to S504, and will not be repeated here.

[0143] In step S403 of some embodiments, the clicked product attention sequence and the searched product attention sequence can be filtered based on a preset ranking threshold. Products that meet the preset ranking threshold are selected to obtain the target attention product sequence for each candidate. If two identical products exist, one of them is retained. The preset ranking threshold needs to be set according to the actual application scenario, such as setting the preset ranking threshold to 5 or 6, etc., and is not limited to this.

[0144] In some embodiments, the target product series includes multiple user-focused products.

[0145] Please see Figure 6 In some embodiments, step S404 includes, but is not limited to, steps S601 to S605:

[0146] Step S601: Extract features from user information to obtain user feature data;

[0147] Step S602: Perform purchasing power prediction on user characteristic data to obtain user purchasing power prediction data;

[0148] Step S603: Obtain product spending data for products that users are interested in;

[0149] Step S604: Match user purchasing power prediction data and product spending data to obtain product matching data;

[0150] Step S605: Based on product matching data, filter the products that the user is interested in to obtain the target recommended product sequence for each candidate.

[0151] Steps S601 to S605, as illustrated in this embodiment, involve extracting features from user information to obtain user feature data, which accurately outlines user profiles. Purchasing power prediction is then performed on the user feature data to obtain user purchasing power prediction data. Next, product spending data for each product in the target product series is obtained. The user purchasing power prediction data and product spending data are matched to obtain product matching data, determining whether the product matches the user's purchasing power. Finally, the product matching data is used to filter the user's target products, resulting in a target recommended product series for each candidate, achieving precise recommendations and improving the effectiveness and accuracy of product recommendations.

[0152] In step S601 of some embodiments, user characteristic data includes, but is not limited to, the age, gender, income, place of residence, industry, education, etc. of the candidate, which can accurately depict the user profile, provide a solid foundation for subsequent purchasing power prediction and product recommendation, and improve the accuracy and targeting of the recommendation.

[0153] In step S602 of some embodiments, a pre-trained autoregressive model is used to predict the purchasing power of user feature data to obtain user purchasing power prediction data. This allows for an early understanding of the purchasing power of potential customers, which helps to recommend products that match their purchasing power and improve conversion rates.

[0154] In step S603 of some embodiments, the product expenditure data includes data related to the cost of purchasing the product that the user is interested in, such as price and taxes. This data is pre-set by the target platform to clarify the actual cost of the product that the user is interested in, providing data for subsequent matching with the user's purchasing power and ensuring that the recommended product is within the affordability range of the candidate.

[0155] In step S604 of some embodiments, user purchasing power prediction data and product spending data are compared and analyzed by using preset matching rules or matching algorithms to calculate the degree of matching between the two and obtain product matching data. This can quickly filter out products that match the user's purchasing power, narrow the recommendation range, and improve recommendation efficiency and quality.

[0156] Specifically, matching rules can employ a threshold matching method, setting a range for the ratio of predicted user purchasing power data to product spending data. When the ratio falls within this range, the product is considered a match between the user's purchasing power and the product's spending power. Alternatively, a tiered matching method can be used, dividing both predicted user purchasing power data and product spending data into different tiers and then matching based on these tier correspondences. For example, predictive user purchasing power data can be divided into high, medium, and low tiers, and product spending data can also be divided into high, medium, and low tiers. This means high purchasing power corresponds to high and medium spending products, medium purchasing power corresponds to medium and low spending products, and low purchasing power corresponds to low spending products.

[0157] The matching algorithm can be either linear regression or cosine similarity matching.

[0158] In some embodiments, the target recommended product sequence includes multiple alternative recommended products.

[0159] Please see Figure 7 In some embodiments, step S106 may include, but is not limited to, steps S701 to S703:

[0160] Step S701: Obtain product information for the alternative recommended products;

[0161] Step S702: Based on the product information, combine the candidate recommended products to obtain the target recommended product combination;

[0162] Step S703: Recommend products to each candidate based on the target recommended product portfolio.

[0163] Steps S701 to S703, as illustrated in this embodiment, involve obtaining information on candidate recommended products to comprehensively and accurately understand product characteristics. Based on this product information, candidate recommended products are combined to obtain a more suitable and complementary target recommended product combination, thus improving the accuracy of product recommendations. Finally, product recommendations are made to each candidate based on the target recommended product combination, which helps to improve the conversion rate of recommended products.

[0164] In step S701 of some embodiments, the product information of the candidate recommended products covers multiple aspects of the candidate recommended products, including but not limited to the product's basic attributes (such as name), functional features, price, user reviews, sales data, etc. This information is pre-set by the target platform to provide a data foundation for subsequent product combination and accurate recommendations.

[0165] In step S702 of some embodiments, different alternative recommended products are combined and integrated according to preset rules and strategies to form a whole. The products in this whole may have complementary, related or synergistic relationships, which can better meet the diverse needs of the candidates and provide more comprehensive and personalized solutions. Compared with single product recommendations, product combinations can improve users' willingness to purchase and satisfaction.

[0166] In some embodiments, alternative recommended products can also be combined through a business agent.

[0167] In step S703 of some embodiments, after generating the final target recommended product portfolio, the target recommended product portfolio is recommended to each candidate through a preset channel, so that the candidate will have the idea of ​​purchasing / subscribing when they learn about the products in the target recommended product portfolio.

[0168] Pre-set channels can be recommended through the target platform's recommendation page. Alternatively, recommendations can be made to potential candidates via outbound calls or SMS, among other methods.

[0169] Please see Figure 8 This application also provides a product recommendation device that can implement the above-described product recommendation method. The device includes:

[0170] User information acquisition module 801 is used to acquire user information of candidate objects; wherein, user information includes age attribute;

[0171] User classification module 802 is used to classify candidate objects based on age attributes to obtain object groups; each object group includes at least two candidate objects;

[0172] The operation behavior acquisition module 803 is used to acquire product operation behavior data for each candidate object for each object group.

[0173] The product popularity calculation module 804 is used to calculate the product popularity of each candidate object's product operation behavior data, and obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object.

[0174] The target product screening module 805 is used to screen products based on user information, group product popularity sequence, and user product popularity sequence to obtain the target recommended product sequence for each candidate.

[0175] The target product recommendation module 806 is used to recommend products to each candidate object based on the target recommended product sequence.

[0176] The specific implementation of the product recommendation device is basically the same as the specific implementation of the product recommendation method described above, and will not be repeated here.

[0177] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the product recommendation method described above. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0178] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0179] The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0180] The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the product recommendation method of the embodiments of this application.

[0181] The input / output interface 903 is used to implement information input and output;

[0182] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0183] Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904);

[0184] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0185] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described product recommendation method.

[0186] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0187] The product recommendation method, apparatus, electronic device, and storage medium provided in this application obtain user information including age attributes of candidate objects, and classify the candidate objects based on age attributes to obtain object groups. Each object group includes at least two candidate objects. Because different age groups have different consumption preferences, grouping can improve the accuracy and targeting of product recommendations. Next, product operation behavior data of each candidate object in each object group is obtained, which can truly reflect the products that the candidate objects are interested in. Product popularity is calculated on the product operation behavior data of each candidate object to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object. Based on user information, group product popularity sequence, and user product popularity sequence, product screening is performed to obtain the target recommended product sequence for each candidate object. This can take into account both individual user characteristics and group commonalities, improving recommendation accuracy. Finally, product recommendations are made to each candidate object based on the target recommended product sequence, which can improve user satisfaction and product recommendation success rate, and achieve more effective product promotion and user service.

[0188] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0189] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0190] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0191] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0192] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0193] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0194] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0195] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0196] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0197] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0198] The software tools or components not belonging to our company that appear in the embodiments of this application are for illustrative purposes only and do not represent actual use.

[0199] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A product recommendation method, characterized in that, The method includes: Obtain user information for candidate candidates; wherein, the user information includes age attributes; The candidate objects are classified based on the age attribute to obtain object groups; each object group includes at least two candidate objects. For each of the aforementioned target groups, obtain product operation behavior data for each of the aforementioned candidate objects; The product popularity is calculated for the product operation behavior data of each candidate object to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object; Based on the user information, the product popularity sequence of the group, and the product popularity sequence of the user, product screening is performed to obtain the target recommended product sequence for each of the candidate objects; Based on the target recommended product sequence, product recommendations are made to each of the candidate objects.

2. The method according to claim 1, characterized in that, The group product popularity sequence includes a group click product sequence and a group search product sequence, and the user product popularity sequence includes a user click product sequence and a user search product sequence; the product filtering based on the user information, the group product popularity sequence, and the user product popularity sequence to obtain the target recommended product sequence for each candidate includes: The difference between the group click product sequence and the user click product sequence is calculated to obtain the click product attention sequence for each candidate object; The difference between the group search product sequence and the user search product sequence is calculated to obtain the search product attention sequence for each of the candidate objects; Based on the clicked product attention sequence and the searched product attention sequence, product filtering is performed to obtain the target attention product sequence for each of the candidate objects; Based on the user information and the target product sequence, product prediction is performed to obtain the target recommended product sequence for each of the candidate objects.

3. The method according to claim 2, characterized in that, The step of calculating the difference between the group click product sequence and the user click product sequence to obtain the click product attention sequence for each candidate object includes: Obtain the name of each product in the sequence of products clicked by the user and the user click order number corresponding to the product name; Based on the product name, the product sequence of the group clicks is filtered to obtain the group click sorting number corresponding to the product name; The difference between the user click sorting number and the group click sorting number is calculated to obtain the click sorting difference data. Based on the click ranking difference data, each product name is sorted to obtain the click product attention sequence for each candidate object.

4. The method according to claim 2, characterized in that, The target product series includes multiple products that users are interested in; the step of predicting products based on the user information and the target product series to obtain the target recommended product series for each candidate includes: Feature extraction is performed on the user information to obtain user feature data; The user characteristic data is used to predict purchasing power to obtain user purchasing power prediction data. Obtain the product spending data of the products that the user is interested in; Product matching data is obtained by matching the user purchasing power prediction data and the product spending data. Based on the product matching data, the products that the user is interested in are filtered to obtain the target recommended product sequence for each of the candidate objects.

5. The method according to any one of claims 1 to 4, characterized in that, The step of calculating product popularity based on the product operation behavior data of each candidate object to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object includes: Feature extraction is performed on the product operation behavior data to obtain initial behavior data; The initial behavioral data is classified to obtain target behavioral data; wherein, the behavioral types of the target behavioral data include click behavior and search behavior; For each of the aforementioned behavior types, keywords are extracted from the target behavior data to obtain product keyword data; For each of the target groups, the product keyword data is aggregated and sorted based on the behavior type to obtain the group product popularity sequence for each target group; For each candidate, the product keyword data is aggregated and sorted based on the behavior type to obtain the user product popularity sequence for each candidate.

6. The method according to claim 5, characterized in that, The step of extracting features from the product operation behavior data to obtain initial behavior data includes: The product operation behavior data is subjected to outlier detection and processing to obtain the raw behavior data; The original behavioral data is standardized to obtain standard behavioral data; The standard behavioral data is subjected to feature filtering to obtain filtered behavioral data; The initial behavioral data is obtained by constructing features from the filtered behavioral data.

7. The method according to any one of claims 1 to 4, characterized in that, The target recommended product sequence includes multiple candidate recommended products; the step of recommending products to each candidate based on the target recommended product sequence includes: Obtain product information of the recommended alternative products; Based on the product information, the candidate recommended products are combined to obtain the target recommended product combination; Based on the target recommended product portfolio, product recommendations are made to each of the candidate items.

8. A product recommendation device, characterized in that, The device includes: The user information acquisition module is used to acquire user information of candidate objects; wherein, the user information includes age attributes; The user classification module is used to classify the candidate objects based on the age attribute to obtain object groups; each object group includes at least two candidate objects; The operation behavior acquisition module is used to acquire product operation behavior data of each of the candidate objects for each of the object groups. The product popularity calculation module is used to calculate the product popularity of the product operation behavior data of each candidate object, so as to obtain the group product popularity sequence of each object group and the user product popularity sequence of each candidate object. The target product filtering module is used to filter products based on the user information, the group product popularity sequence, and the user product popularity sequence to obtain a target recommended product sequence for each candidate. The target product recommendation module is used to recommend products to each of the candidate objects based on the target recommended product sequence.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images