Product recommendation method and device based on terminal intelligence, computer equipment and medium

Abstract

The invention discloses a product recommendation method and device based on end intelligence, computer equipment and a medium, and relates to the technical field of data processing.The method comprises the steps that operation behaviors of a target user in a target time period are recorded through an end intelligent system; based on the operation behavior, returning M target products through a background recommendation system; sorting the M target products, selecting N target products with the highest similarity with the operation behavior, and recommending the N target products to the target user; wherein M and N represent positive integers, and M is greater than N. Through the method provided by the invention, the specific target product can be recommended to the target user based on the operation behavior of the target user in the specific time period (target time period), so that more accurate and personalized product recommendation can be realized.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a product recommendation method, device, computer equipment and non-volatile computer-readable storage medium based on terminal intelligence. Background Art

[0002] At present, in the context of digital marketing and advertising, the click-through rate of the operation position is one of the key indicators to measure the effectiveness of advertising. The level of click-through rate is directly related to the attractiveness of the advertising content and the effectiveness of information dissemination. For insurance companies, in the advertising promotion of non-auto insurance products, a high click-through rate can not only increase users' awareness of the product, but also directly affect the sales performance of the product.

[0003] Currently, traditional product recommendation methods usually rely on historical data and static user portraits to recommend products to users. This method generates a recommendation list by analyzing the user's past browsing history, purchase history, and preference settings. However, this method has obvious limitations:

[0004] 1. Data timeliness: The data of traditional product recommendation methods is often delayed. It may be based on user behavior from one day or even several days ago and cannot reflect the latest changes in user interests in real time.

[0005] 2. User interest drift: User interests and needs may change rapidly over time and in different environments, and traditional product recommendation methods cannot capture such changes in a timely manner, resulting in a decrease in the relevance of product recommendations.

[0006] 3. Lack of real-time features: Real-time data generated by users during browsing and interaction, such as page dwell time, mouse hover position, scrolling speed, etc. These micro behaviors can provide clues to users' immediate interests, but traditional product recommendation methods often ignore these real-time features;

[0007] 4. Insufficient personalized recommendations: Due to the lack of analysis of real-time user behavior, traditional product recommendation methods are difficult to achieve highly personalized recommendations and cannot meet users' expectations for the accuracy of recommended content.

[0008] Based on this, how to provide a product recommendation method, device, computer equipment and non-volatile computer-readable storage medium based on end intelligence, which can recommend specific target products to the target user based on the target user's operating behavior within a specific time period to achieve more accurate and personalized product recommendations, is a problem that needs to be urgently solved by technical personnel in this field. Summary of the invention

[0009] In view of the above-mentioned deficiencies in the prior art, the object of the present invention is to provide an end-intelligent product recommendation method, apparatus, computer device and non-volatile computer-readable storage medium that can be used in financial technology or other related fields, aiming to solve the problem of how to recommend specific target products to a target user based on the target user's operating behavior within a specific time period, so as to achieve more accurate and personalized product recommendations.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] A product recommendation method based on terminal intelligence, comprising:

[0012] The target user's operation behavior within the target time period is recorded through the terminal intelligent system;

[0013] Based on the operation behavior, M target products are returned through the background recommendation system;

[0014] Sorting the M target products, selecting N target products with the highest similarity to the operation behavior and recommending them to the target user;

[0015] Wherein, M and N both represent positive integers, and M is greater than N.

[0016] In a further technical solution, the product recommendation method based on terminal intelligence, wherein the operation behavior of the target user within the target time period is recorded by the terminal intelligence system, includes:

[0017] The end intelligent system records the behavior feature text of the target user's operation behavior within the target time period;

[0018] The behavior feature text is converted into a corresponding behavior feature vector through a word vectorization system.

[0019] In a further technical solution, the product recommendation method based on terminal intelligence, wherein the returning M target products through the background recommendation system based on the operation behavior, includes:

[0020] Based on the behavior feature vector, returning product feature texts of M target products through the backend recommendation system;

[0021] Each of the product feature texts is converted into a corresponding product feature vector through the word vectorization system.

[0022] In a further technical solution, the product recommendation method based on terminal intelligence, wherein the M target products are sorted, and N target products having the highest similarity with the operation behavior are selected and recommended to the target user, include:

[0023] Sorting the product feature vectors corresponding to the M target products by using a cosine similarity algorithm;

[0024] Selecting N product feature vectors having the highest similarity to the behavior feature vector corresponding to the operation behavior from among the M product feature vectors;

[0025] The target products corresponding to the N product feature vectors are recommended to the target users.

[0026] In a further technical solution, the product recommendation method based on terminal intelligence, wherein the target product corresponding to the N product feature vectors is recommended to the target user, comprises:

[0027] The target products corresponding to the N product feature vectors are displayed in a target area to recommend them to the target users.

[0028] In a further technical solution, the product recommendation method based on end intelligence, wherein the operation behavior of the target user within the target time period is recorded by the end intelligence system, wherein the target time period includes the current day.

[0029] In a further technical solution, the product recommendation method based on terminal intelligence, wherein the operation behavior of the target user within the target time period is recorded by the terminal intelligence system, wherein the operation behavior includes one or more of the following behaviors:

[0030] a. Views and clicks on waterfall articles;

[0031] b. Waterfall video browsing and clicking;

[0032] c. Waterfall product browsing and clicking;

[0033] d. Browsing and clicking on hot-selling products;

[0034] e. Article, video, and product search.

[0035] A product recommendation device based on terminal intelligence, comprising:

[0036] The operation recording module is used to record the operation behavior of the target user within the target time period through the terminal intelligent system;

[0037] A product return module, used to return M target products through a background recommendation system based on the operation behavior;

[0038] A product recommendation module is used to sort the M target products, select the N target products with the highest similarity to the operation behavior and recommend them to the target user;

[0039] Wherein, M and N both represent positive integers, and M is greater than N.

[0040] A computer device, wherein the computer device comprises at least one processor; and

[0041] a memory communicatively connected to the at least one processor; wherein,

[0042] The memory stores a computer program that can be executed by the at least one processor. When the computer program is executed by the at least one processor, the end-intelligence-based product recommendation method as described in any one of the above items can be implemented.

[0043] A non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, it can implement any of the above-mentioned end-intelligence-based product recommendation methods.

[0044] Compared with the prior art, the present invention provides a product recommendation method, device, computer equipment and non-volatile computer-readable storage medium based on end intelligence, wherein the method comprises: recording the operation behavior of the target user in the target time period through the end intelligence system; based on the operation behavior, returning M target products through the background recommendation system; sorting the M target products, selecting the N target products with the highest similarity to the operation behavior and recommending them to the target user; wherein M and N both represent positive integers, and M is greater than N. In this way, the method of the present invention can recommend specific target products to the target user based on the operation behavior of the target user in a specific time period (target time period), thereby achieving more accurate and personalized product recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A flowchart of a product recommendation method based on terminal intelligence provided in an embodiment of the present invention.

[0047] Figure 2 A schematic diagram of functional modules of a product recommendation device based on terminal intelligence provided in an embodiment of the present invention.

[0048] Figure 3 A schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0050] In the description of the present invention, the terms "include", "including", "have", "contain", etc. are all open terms, which mean including but not limited to. The descriptions with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", etc. mean that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. The order of steps involved in each embodiment is used to schematically illustrate the implementation of the present application, and the order of steps therein is not limited and can be appropriately adjusted as needed.

[0051] Various non-limiting embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] At present, in the context of digital marketing and advertising, the click-through rate of the operation position is one of the key indicators to measure the effectiveness of advertising. The level of click-through rate is directly related to the attractiveness of the advertising content and the effectiveness of information dissemination. For insurance companies, in the advertising promotion of non-auto insurance products, a high click-through rate can not only increase users' awareness of the product, but also directly affect the sales performance of the product.

[0053] Currently, traditional product recommendation methods usually rely on historical data and static user portraits to recommend products to users. This method generates a recommendation list by analyzing the user's past browsing history, purchase history, and preference settings. However, this method has obvious limitations:

[0054] 1. Data timeliness: The data of traditional product recommendation methods is often delayed. It may be based on user behavior from one day or even several days ago and cannot reflect the latest changes in user interests in real time.

[0055] 2. User interest drift: User interests and needs may change rapidly over time and in different environments, and traditional product recommendation methods cannot capture such changes in a timely manner, resulting in a decrease in the relevance of product recommendations.

[0056] 3. Lack of real-time features: Real-time data generated by users during browsing and interaction, such as page dwell time, mouse hover position, scrolling speed, etc. These micro behaviors can provide clues to users' immediate interests, but traditional product recommendation methods often ignore these real-time features;

[0057] 4. Insufficient personalized recommendations: Due to the lack of analysis of real-time user behavior, traditional product recommendation methods are difficult to achieve highly personalized recommendations and cannot meet users' expectations for the accuracy of recommended content.

[0058] Based on this, how to provide a product recommendation method, device, computer equipment and non-volatile computer-readable storage medium based on end intelligence, which can recommend specific target products to the target user based on the target user's operating behavior within a specific time period to achieve more accurate and personalized product recommendations, is a problem that needs to be urgently solved by technical personnel in this field.

[0059] Therefore, to solve the above problem, please refer to Figure 1 The embodiment of the present invention provides a product recommendation method based on terminal intelligence, wherein the method comprises the steps of:

[0060] S100, recording the operation behavior of the target user within the target time period through the terminal intelligent system;

[0061] S200, based on the operation behavior, returning M target products through the background recommendation system;

[0062] S300, sorting the M target products, selecting N target products with the highest similarity to the operation behavior and recommending them to the target user;

[0063] Wherein, M and N both represent positive integers, and M is greater than N.

[0064] Specifically, in the present invention, on-device machine learning refers to the application and deployment of machine / deep learning algorithm models on end devices. The "end device" here is relative to cloud services, and can be a mobile phone or an IoT device.

[0065] Furthermore, in the product recommendation method based on terminal intelligence, the step S100 of recording the operation behavior of the target user within the target time period by the terminal intelligence system includes:

[0066] The end intelligent system records the behavior feature text of the target user's operation behavior within the target time period;

[0067] The behavior feature text is converted into a corresponding behavior feature vector through a word vectorization system.

[0068] In the specific implementation, in this embodiment, the behavior feature text of the target user's operation behavior within the target time period is recorded by the terminal intelligent system, and the behavior feature text is converted into a corresponding behavior feature vector by the word vectorization system. The specific implementation process is as follows:

[0069] 1) Behavioral characteristic text recording the operation behavior

[0070] The end intelligent system captures and records a series of operation behaviors of the target user within the target time period. These operation behaviors may include but are not limited to the target user's interactive actions such as browsing, clicking, searching, and purchasing. For each operation behavior, the end intelligent system extracts the corresponding behavior feature text. The behavior feature text is text information describing the user's operation behavior, which may include the object of the operation (such as article, video, product, etc.), the type of operation (such as click, browse, etc.), the time and duration of the operation, etc.

[0071] 2) Convert the behavior feature text into a behavior feature vector:

[0072] Next, the collected behavior feature text is converted into a numerical vector, i.e., a behavior feature vector, by using a word vectorization system. Specifically, the following steps are involved:

[0073] Text preprocessing: Clean the behavior feature text, including removing stop words and punctuation marks, and performing stem extraction or word form restoration;

[0074] Tokenization: Break the cleaned text into individual words or phrases;

[0075] Vectorization: Use models such as Word2Vec, GloVe, and BERT to map the word segmentation results into a vector space to generate a numerical vector that can represent the original text information;

[0076] These behavioral feature vectors can capture the semantic features and patterns of user behaviors, providing a basis for subsequent data analysis, user behavior prediction, personalized recommendation and other applications;

[0077] Through the above process, the complex operation behaviors of target users can be converted into concise and easy-to-process vector representations, thereby providing a data basis for personalized services and decision support while protecting user privacy;

[0078] Furthermore, the following is a basic introduction to the terminal intelligent system and the word vectorization system in the present invention:

[0079] 1) Terminal Intelligent System

[0080] Edge intelligence is a type of intelligent processing technology that runs directly on user devices. It enables devices to independently collect, process and analyze data without relying on cloud servers. Key features of edge intelligence include:

[0081] Instant response: Process data locally, reducing the time waiting for cloud response;

[0082] Data privacy: Because data processing is done locally, user privacy can be better protected;

[0083] Self-sufficient: Able to perform some basic intelligent tasks even without an Internet connection;

[0084] Personalized services: can provide customized services and experiences based on user behavior on the device;

[0085] 2) Word vectorization system

[0086] The word vectorization system is a core technology in natural language processing (NLP). It converts words or phrases in a text into vectors in numerical form. These vectors can represent the semantic features of the words. The main advantages of word vectorization include:

[0087] Semantic information: word vectors can capture the semantic relationship between words, and similar words are closer in the vector space;

[0088] Dimensionality reduction: Mapping text data from high-dimensional space to low-dimensional space simplifies the calculation process;

[0089] Pre-trained model: A word vector model trained using large-scale text data, which can be used for a variety of language processing tasks;

[0090] Widely used: Word vectorization is the basis of many NLP tasks, such as sentiment analysis, machine translation, text classification, etc.

[0091] From the above introduction, we can see that the end intelligent system and word vectorization system are key technologies for realizing intelligent applications and services. They play an important role in providing a fast, personalized and privacy-friendly user experience.

[0092] Furthermore, in the product recommendation method based on end intelligence, the step S100 records the operation behavior of the target user within the target time period through the end intelligence system, wherein the target time period includes the current day.

[0093] In specific implementation, in this embodiment, the target time period can be preset, such as the past few hours, the past few days (such as 3 days), etc.; this embodiment takes the target time period as the current day as an example. Specifically, the purpose of recording the target user's operation behavior on the current day through the terminal intelligent system mainly includes the following points:

[0094] 1) Real-time behavior analysis: Capturing the target user's daily operation behavior provides data support for real-time analysis and recommendation, helping to understand the target user's immediate interests and needs, and achieving the purpose of real-time recommendation;

[0095] 2) Personalized recommendations: Based on the target user’s activities on the day, personalized content or service recommendations are provided to improve user satisfaction and engagement;

[0096] 3) Trend monitoring: By monitoring the target user’s behavior on the day, we can discover popular trends or rapid changes in user preferences, thereby making timely recommendation decisions;

[0097] 4) User experience optimization: Analyze the target user's daily operation behavior, identify potential problems in the user interface and process, and then optimize the user experience;

[0098] 5) Instant feedback loop: Utilize the target user’s behavior data of the day to quickly collect feedback and adjust the recommendation algorithm or service strategy to achieve more accurate personalized services;

[0099] 6) Behavior prediction: By analyzing the target user's behavior on the day, predict their future behavior patterns, and provide a basis for long-term user relationship management and marketing activities;

[0100] By recording the user's daily operating behavior through the end-to-end intelligent system, enterprises or service providers can better understand user needs, provide more timely and personalized services, and optimize the promotion strategy of products or services.

[0101] Furthermore, in the product recommendation method based on terminal intelligence, in step S100, the operation behavior of the target user in the target time period is recorded by the terminal intelligence system, wherein the operation behavior includes one or more of the following behaviors:

[0102] a. Views and clicks on waterfall articles;

[0103] b. Waterfall video browsing and clicking;

[0104] c. Waterfall product browsing and clicking;

[0105] d. Browsing and clicking on hot-selling products;

[0106] e. Article, video, and product search.

[0107] In the specific implementation, in this embodiment, one or more of the above operation behaviors of the target user in the target time period are recorded by the terminal intelligent system, and the purposes are mainly as follows:

[0108] 1) User interest identification: Analyze the target user's browsing and clicking behavior on articles, videos and products in the waterfall flow to identify the target user's interests and preferences;

[0109] 2) Content optimization: Based on the target users’ interaction with the waterfall content, optimize the product recommendation algorithm to recommend products that are more in line with the target users’ interests;

[0110] 3) Product promotion effect evaluation: Evaluate the effectiveness of marketing strategies and product recommendations by recording the target users’ responses to hot-selling products;

[0111] 4) Search behavior analysis: Study the search habits and query content of target users to understand their specific needs and search intentions;

[0112] 5) Personalized recommendations: Use the target user’s operational behavior data to generate personalized recommendations to improve the target user’s engagement and satisfaction;

[0113] By recording one or more of the above-mentioned operation behaviors of the target user within the target time period through the end intelligent system, the enterprise or service provider can better understand the user needs, provide more timely and personalized services, and optimize the promotion strategy of products or services.

[0114] Furthermore, in the product recommendation method based on terminal intelligence, the step S200, based on the operation behavior, returns M target products through the background recommendation system, including:

[0115] Based on the behavior feature vector, returning product feature texts of M target products through the backend recommendation system;

[0116] Each of the product feature texts is converted into a corresponding product feature vector through the word vectorization system.

[0117] In the specific implementation, in this embodiment, based on the behavior feature vector, the product feature texts of M target products are returned through the background recommendation system, and each of the product feature texts is converted into a corresponding product feature vector through the word vectorization system. The specific implementation process is as follows:

[0118] 1) Product feature text matching

[0119] The backend recommendation system selects M (e.g., 20) target products from the product library based on the target user's behavioral feature vectors, and extracts the feature texts of these products, such as product descriptions, features, user reviews, etc.

[0120] 2) Product feature vector calculation

[0121] The word vectorization system is also used to convert the text descriptions of the M target products into corresponding product feature vectors. This step ensures the vectorization representation of the product features, which facilitates the subsequent similarity calculation.

[0122] Through the above process, it is possible to provide personalized target product recommendations for target users, while continuously learning and adapting to the user's latest preferences to improve the accuracy of recommendations and user satisfaction;

[0123] Furthermore, the following is a basic introduction to the background recommendation system in the present invention:

[0124] The backend recommendation system is a service that runs on the server side and predicts and recommends products or content that users may be interested in by analyzing the target user's operation behavior data. Its characteristics include the following:

[0125] 1) Centralized processing: The backend recommendation system processes data in the cloud, utilizing the computing power of the server;

[0126] 2) User behavior learning: Analyze users’ historical behaviors and preferences, including browsing, clicking, purchasing, etc.

[0127] 3) Personalized recommendation: Generate a personalized recommendation list based on learning results to improve user experience;

[0128] 4) Real-time update: Continuously update the recommendation algorithm to reflect the user's latest interests and behaviors;

[0129] The backend recommendation system is an important tool for improving user satisfaction and commercial benefits, providing customized content and product recommendations through intelligent analysis.

[0130] Furthermore, the product recommendation method based on terminal intelligence, wherein the step S300, sorting the M target products, selecting the N target products with the highest similarity to the operation behavior and recommending them to the target user, comprises:

[0131] Sorting the product feature vectors corresponding to the M target products by using a cosine similarity algorithm;

[0132] Selecting N product feature vectors having the highest similarity to the behavior feature vector corresponding to the operation behavior from among the M product feature vectors;

[0133] The target products corresponding to the N product feature vectors are recommended to the target users.

[0134] In specific implementation, in this embodiment, the product feature vectors corresponding to the M target products are sorted by a cosine similarity algorithm, and the N product feature vectors with the highest similarity to the behavior feature vector corresponding to the operation behavior are selected from the M product feature vectors, and the target products corresponding to the N product feature vectors are recommended to the target user. The specific implementation process is as follows:

[0135] 1) Product feature vector sorting

[0136] Use the cosine similarity algorithm to reorder the product feature vectors corresponding to the M target products. The cosine similarity algorithm evaluates their similarity by measuring the angle between two vectors.

[0137] 2) Similarity calculation

[0138] Calculate the similarity between each product feature vector and the behavior feature vector corresponding to the user's operation behavior;

[0139] 3) Highly similar product selection

[0140] From M (e.g., 20) product feature vectors, select N (e.g., 3) product feature vectors with the highest similarity to the user behavior feature vector; these N vectors represent the products that best match the current interests of the target user;

[0141] 4) Product recommendations

[0142] Recommend these N target products with high similarity to the target user (e.g., send them directly to the user or display them to the user in a specific area). These recommendations are based on the user's real-time behavior and preferences, aiming to provide a personalized user experience.

[0143] Through the above process, it can be ensured that the recommendation process can dynamically respond to the user's latest behavior and update the recommendation list in a timely manner to improve user engagement and satisfaction; through this method, the user's interest changes can be captured more accurately, thereby providing more relevant and attractive product recommendations.

[0144] Furthermore, in the product recommendation method based on terminal intelligence, the step of recommending the target products corresponding to the N product feature vectors to the target user includes:

[0145] The target products corresponding to the N product feature vectors are displayed in a target area to recommend them to the target users.

[0146] In the specific implementation, in this embodiment, the target products corresponding to the N product feature vectors are displayed in the target area to be recommended to the target user, wherein the specific implementation process is as follows:

[0147] 1) Target product selection

[0148] Based on the user's behavior feature vector, calculate and select N product feature vectors with high similarity, and select the corresponding target product;

[0149] 2) Placement selection

[0150] Choose a high-traffic and high-visibility placement (target area), such as the "Hot Products" placement, which is an area specially designed to attract user attention and promote clicks and purchases;

[0151] 3) Recommended product display

[0152] Display these N selected target products in the selected target area, and use the advantages of the location to increase the exposure of these products;

[0153] 4) User Interaction

[0154] When the target users browse the app or website, they will notice these specially recommended products, which may arouse their interest and lead to further clicks or purchases;

[0155] 5) Effect monitoring

[0156] Real-time monitoring of user responses to products displayed in the target area, including key indicators such as click-through rate and conversion rate, to evaluate the effectiveness of recommendations;

[0157] 6) Dynamic Adjustment

[0158] Based on user feedback and business goals, the recommendation strategy can be adjusted dynamically, including changing products, adjusting recommendation algorithms, or updating display content in target areas.

[0159] Through the above process, not only can personalized product recommendations be provided, but it can also ensure that the recommended content receives the greatest user attention in the best display position, thereby improving the effectiveness of the recommendation and business results.

[0160] It can be seen from the above method embodiments that the product recommendation method based on end intelligence provided by the present invention at least includes: recording the operation behavior of the target user in the target time period through the end intelligence system; based on the operation behavior, returning M target products through the background recommendation system; sorting the M target products, selecting the N target products with the highest similarity to the operation behavior and recommending them to the target user; wherein M and N both represent positive integers, and M is greater than N; and so on. In this way, the method of the present invention can recommend specific target products to the target user based on the operation behavior of the target user in a specific time period (target time period), thereby achieving more accurate and personalized product recommendations.

[0161] Specifically, the end-intelligence-based product recommendation method provided in an embodiment of the present invention can capture the real-time operation behavior of the target user by introducing the real-time operation of the target user as an input parameter into the product recommendation system, so that the insurance company can realize real-time recommendation of non-auto insurance products, solving the problem of low real-time performance of existing recommendation methods.

[0162] It should be understood that, although the present application provides method operation steps as described in the embodiments or flowcharts, more or less operation steps may be included based on conventional or non-creative labor, and these operation steps are not necessarily performed in sequence according to the embodiment or flowchart. The order of steps listed in the embodiment or flowchart is only one way of executing the order of many steps, and does not represent the only execution order. It should be noted that there is not necessarily a certain order between the above steps. A person of ordinary skill in the art can understand from the description of the embodiment of the present invention that in different embodiments, the above steps may have different execution orders, that is, they may be executed in parallel, or they may be executed in exchange, etc. Moreover, at least a part of the steps in the embodiment or flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but may be executed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but may be executed in turn, alternately or synchronously with other steps or at least a part of the sub-steps or stages of other steps.

[0163] Based on the above method embodiment, please refer to Figure 2 Another embodiment of the present invention further provides a product recommendation device based on terminal intelligence, wherein the device includes:

[0164] The operation recording module 11 is used to record the operation behavior of the target user within the target time period through the terminal intelligent system;

[0165] A product return module 12, used to return M target products through a background recommendation system based on the operation behavior;

[0166] A product recommendation module 13 is used to sort the M target products, select the N target products with the highest similarity to the operation behavior and recommend them to the target user;

[0167] Wherein, M and N both represent positive integers, and M is greater than N.

[0168] Specifically, in the present invention, on-device machine learning refers to the application and deployment of machine / deep learning algorithm models on end devices. The "end device" here is relative to cloud services, and can be a mobile phone or an IoT device.

[0169] Furthermore, in the device for recommending products based on terminal intelligence, the step of recording the operation behavior of the target user within the target time period by the terminal intelligence system includes:

[0170] The end intelligent system records the behavior feature text of the target user's operation behavior within the target time period;

[0171] The behavior feature text is converted into a corresponding behavior feature vector through a word vectorization system.

[0172] In the specific implementation, in this embodiment, the behavior feature text of the target user's operation behavior within the target time period is recorded by the terminal intelligent system, and the behavior feature text is converted into a corresponding behavior feature vector by the word vectorization system. The specific implementation process is as follows:

[0173] 1) Behavioral characteristic text recording the operation behavior

[0174] The end intelligent system captures and records a series of operation behaviors of the target user within the target time period. These operation behaviors may include but are not limited to the target user's interactive actions such as browsing, clicking, searching, and purchasing. For each operation behavior, the end intelligent system extracts the corresponding behavior feature text. The behavior feature text is text information describing the user's operation behavior, which may include the object of the operation (such as article, video, product, etc.), the type of operation (such as click, browse, etc.), the time and duration of the operation, etc.

[0175] 2) Convert the behavior feature text into a behavior feature vector:

[0176] Next, the collected behavior feature text is converted into a numerical vector, i.e., a behavior feature vector, by using a word vectorization system. Specifically, the following steps are involved:

[0177] Text preprocessing: Clean the behavior feature text, including removing stop words and punctuation marks, and performing stem extraction or word form restoration;

[0178] Tokenization: Break the cleaned text into individual words or phrases;

[0179] Vectorization: Use models such as Word2Vec, GloVe, and BERT to map the word segmentation results into a vector space to generate a numerical vector that can represent the original text information;

[0180] These behavioral feature vectors can capture the semantic features and patterns of user behaviors, providing a basis for subsequent data analysis, user behavior prediction, personalized recommendation and other applications;

[0181] Through the above process, the complex operation behaviors of target users can be converted into concise and easy-to-process vector representations, thereby providing a data basis for personalized services and decision support while protecting user privacy;

[0182] Furthermore, the following is a basic introduction to the terminal intelligent system and the word vectorization system in the present invention:

[0183] 1) Terminal Intelligent System

[0184] Edge intelligence is a type of intelligent processing technology that runs directly on user devices. It enables devices to independently collect, process and analyze data without relying on cloud servers. Key features of edge intelligence include:

[0185] Instant response: Process data locally, reducing the time waiting for cloud response;

[0186] Data privacy: Because data processing is done locally, user privacy can be better protected;

[0187] Self-sufficient: Able to perform some basic intelligent tasks even without an Internet connection;

[0188] Personalized services: can provide customized services and experiences based on user behavior on the device;

[0189] 2) Word vectorization system

[0190] The word vectorization system is a core technology in natural language processing (NLP). It converts words or phrases in a text into vectors in numerical form. These vectors can represent the semantic features of the words. The main advantages of word vectorization include:

[0191] Semantic information: word vectors can capture the semantic relationship between words, and similar words are closer in the vector space;

[0192] Dimensionality reduction: Mapping text data from high-dimensional space to low-dimensional space simplifies the calculation process;

[0193] Pre-trained model: A word vector model trained using large-scale text data, which can be used for a variety of language processing tasks;

[0194] Widely used: Word vectorization is the basis of many NLP tasks, such as sentiment analysis, machine translation, text classification, etc.

[0195] From the above introduction, we can see that the end intelligent system and word vectorization system are key technologies for realizing intelligent applications and services. They play an important role in providing a fast, personalized and privacy-friendly user experience.

[0196] Furthermore, the product recommendation device based on end intelligence, wherein the operation behavior of the target user within a target time period is recorded by the end intelligence system, wherein the target time period includes the current day.

[0197] In specific implementation, in this embodiment, the target time period can be preset, such as the past few hours, the past few days (such as 3 days), etc.; this embodiment takes the target time period as the current day as an example. Specifically, the purpose of recording the target user's operation behavior on the current day through the terminal intelligent system mainly includes the following points:

[0198] 1) Real-time behavior analysis: Capturing the target user's daily operation behavior provides data support for real-time analysis and recommendation, helping to understand the target user's immediate interests and needs, and achieving the purpose of real-time recommendation;

[0199] 2) Personalized recommendations: Based on the target user’s activities on the day, personalized content or service recommendations are provided to improve user satisfaction and engagement;

[0200] 3) Trend monitoring: By monitoring the target user’s behavior on the day, we can discover popular trends or rapid changes in user preferences, thereby making timely recommendation decisions;

[0201] 4) User experience optimization: Analyze the target user's daily operation behavior, identify potential problems in the user interface and process, and then optimize the user experience;

[0202] 5) Instant feedback loop: Utilize the target user’s behavior data of the day to quickly collect feedback and adjust the recommendation algorithm or service strategy to achieve more accurate personalized services;

[0203] 6) Behavior prediction: By analyzing the target user's behavior on the day, predict their future behavior patterns, and provide a basis for long-term user relationship management and marketing activities;

[0204] By recording the user's daily operating behavior through the end-to-end intelligent system, enterprises or service providers can better understand user needs, provide more timely and personalized services, and optimize the promotion strategy of products or services.

[0205] Furthermore, in the product recommendation device based on terminal intelligence, the operation behavior of the target user within the target time period is recorded by the terminal intelligence system, wherein the operation behavior includes one or more of the following behaviors:

[0206] a. Views and clicks on waterfall articles;

[0207] b. Waterfall video browsing and clicking;

[0208] c. Waterfall product browsing and clicking;

[0209] d. Browsing and clicking on hot-selling products;

[0210] e. Article, video, and product search.

[0211] In the specific implementation, in this embodiment, one or more of the above operation behaviors of the target user in the target time period are recorded by the terminal intelligent system, and the purposes are mainly as follows:

[0212] 1) User interest identification: Analyze the target user's browsing and clicking behavior on articles, videos and products in the waterfall flow to identify the target user's interests and preferences;

[0213] 2) Content optimization: Based on the target users’ interaction with the waterfall content, optimize the product recommendation algorithm to recommend products that are more in line with the target users’ interests;

[0214] 3) Product promotion effect evaluation: Evaluate the effectiveness of marketing strategies and product recommendations by recording the target users’ responses to hot-selling products;

[0215] 4) Search behavior analysis: Study the search habits and query content of target users to understand their specific needs and search intentions;

[0216] 5) Personalized recommendations: Use the target user’s operational behavior data to generate personalized recommendations to improve the target user’s engagement and satisfaction;

[0217] By recording one or more of the above-mentioned operation behaviors of the target user within the target time period through the end intelligent system, the enterprise or service provider can better understand the user needs, provide more timely and personalized services, and optimize the promotion strategy of products or services.

[0218] Furthermore, the product recommendation device based on terminal intelligence, wherein the returning M target products through the background recommendation system based on the operation behavior includes:

[0219] Based on the behavior feature vector, returning product feature texts of M target products through the backend recommendation system;

[0220] Each of the product feature texts is converted into a corresponding product feature vector through the word vectorization system.

[0221] In the specific implementation, in this embodiment, based on the behavior feature vector, the product feature texts of M target products are returned through the background recommendation system, and each of the product feature texts is converted into a corresponding product feature vector through the word vectorization system. The specific implementation process is as follows:

[0222] 1) Product feature text matching

[0223] The backend recommendation system selects M (e.g., 20) target products from the product library based on the target user's behavioral feature vectors, and extracts the feature texts of these products, such as product descriptions, features, user reviews, etc.

[0224] 2) Product feature vector calculation

[0225] The word vectorization system is also used to convert the text descriptions of the M target products into corresponding product feature vectors. This step ensures the vectorization representation of the product features, which facilitates the subsequent similarity calculation.

[0226] Through the above process, it is possible to provide personalized target product recommendations for target users, while continuously learning and adapting to the user's latest preferences to improve the accuracy of recommendations and user satisfaction;

[0227] Furthermore, the following is a basic introduction to the background recommendation system in the present invention:

[0228] The backend recommendation system is a service that runs on the server side and predicts and recommends products or content that users may be interested in by analyzing the target user's operation behavior data. Its characteristics include the following:

[0229] 1) Centralized processing: The backend recommendation system processes data in the cloud, utilizing the computing power of the server;

[0230] 2) User behavior learning: Analyze users’ historical behaviors and preferences, including browsing, clicking, purchasing, etc.

[0231] 3) Personalized recommendation: Generate a personalized recommendation list based on learning results to improve user experience;

[0232] 4) Real-time update: Continuously update the recommendation algorithm to reflect the user's latest interests and behaviors;

[0233] The backend recommendation system is an important tool for improving user satisfaction and commercial benefits, providing customized content and product recommendations through intelligent analysis.

[0234] Furthermore, the device for recommending products based on terminal intelligence, wherein the step of sorting the M target products and selecting the N target products having the highest similarity to the operation behavior for recommendation to the target user, comprises:

[0235] Sorting the product feature vectors corresponding to the M target products by using a cosine similarity algorithm;

[0236] Selecting N product feature vectors having the highest similarity to the behavior feature vector corresponding to the operation behavior from among the M product feature vectors;

[0237] The target products corresponding to the N product feature vectors are recommended to the target users.

[0238] In specific implementation, in this embodiment, the product feature vectors corresponding to the M target products are sorted by a cosine similarity algorithm, and the N product feature vectors with the highest similarity to the behavior feature vector corresponding to the operation behavior are selected from the M product feature vectors, and the target products corresponding to the N product feature vectors are recommended to the target user. The specific implementation process is as follows:

[0239] 1) Product feature vector sorting

[0240] Use the cosine similarity algorithm to reorder the product feature vectors corresponding to the M target products. The cosine similarity algorithm evaluates their similarity by measuring the angle between two vectors.

[0241] 2) Similarity calculation

[0242] Calculate the similarity between each product feature vector and the behavior feature vector corresponding to the user's operation behavior;

[0243] 3) Highly similar product selection

[0244] From M (e.g., 20) product feature vectors, select N (e.g., 3) product feature vectors with the highest similarity to the user behavior feature vector; these N vectors represent the products that best match the current interests of the target user;

[0245] 4) Product recommendations

[0246] Recommend these N target products with high similarity to the target user (e.g., send them directly to the user or display them to the user in a specific area). These recommendations are based on the user's real-time behavior and preferences, aiming to provide a personalized user experience.

[0247] Through the above process, it can be ensured that the recommendation process can dynamically respond to the user's latest behavior and update the recommendation list in a timely manner to improve user engagement and satisfaction; through this device, the user's interest changes can be captured more accurately, thereby providing more relevant and attractive product recommendations.

[0248] Furthermore, in the device for product recommendation based on terminal intelligence, the step of recommending the target products corresponding to the N product feature vectors to the target user includes:

[0249] The target products corresponding to the N product feature vectors are displayed in a target area to recommend them to the target users.

[0250] In the specific implementation, in this embodiment, the target products corresponding to the N product feature vectors are displayed in the target area to be recommended to the target user, wherein the specific implementation process is as follows:

[0251] 1) Target product selection

[0252] Based on the user's behavior feature vector, calculate and select N product feature vectors with high similarity, and select the corresponding target product;

[0253] 2) Placement selection

[0254] Choose a high-traffic and high-visibility placement (target area), such as the "Hot Products" placement, which is an area specially designed to attract user attention and promote clicks and purchases;

[0255] 3) Recommended product display

[0256] Display these N selected target products in the selected target area, and use the advantages of the location to increase the exposure of these products;

[0257] 4) User Interaction

[0258] When the target users browse the app or website, they will notice these specially recommended products, which may arouse their interest and lead to further clicks or purchases;

[0259] 5) Effect monitoring

[0260] Real-time monitoring of user responses to products displayed in the target area, including key indicators such as click-through rate and conversion rate, to evaluate the effectiveness of recommendations;

[0261] 6) Dynamic Adjustment

[0262] Based on user feedback and business goals, the recommendation strategy can be adjusted dynamically, including changing products, adjusting recommendation algorithms, or updating display content in target areas.

[0263] Through the above process, not only can personalized product recommendations be provided, but it can also ensure that the recommended content receives the greatest user attention in the best display position, thereby improving the effectiveness of the recommendation and business results.

[0264] From the above device embodiments, it can be known that the device for product recommendation based on end intelligence provided by the present invention at least includes: recording the operation behavior of the target user in the target time period through the end intelligence system; based on the operation behavior, returning M target products through the background recommendation system; sorting the M target products, selecting the N target products with the highest similarity to the operation behavior and recommending them to the target user; wherein M and N both represent positive integers, and M is greater than N; and so on. In this way, the device of the present invention can recommend specific target products to the target user based on the operation behavior of the target user in a specific time period (target time period), thereby achieving more accurate and personalized product recommendations.

[0265] Specifically, the end-intelligence-based product recommendation device provided in an embodiment of the present invention can capture the real-time operation behavior of the target user by introducing the real-time operation of the target user as an input parameter into the product recommendation system, so that the insurance company can realize real-time recommendation of non-auto insurance products, solving the problem of low real-time performance of existing recommendation devices.

[0266] Based on the above method embodiment, please refer to Figure 3 Another embodiment of the present invention further provides a computer device, wherein the computer device 10 includes:

[0267] memory 120 and one or more processors 110, Figure 3 A processor 110 is used as an example for description. The processor 110 and the memory 120 may be connected via a communication bus or other means. Figure 3 The example of connecting via a communication bus is taken as an example.

[0268] The processor 110 is used to complete various control logics of the computer device 10. It can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, a discrete gate or transistor logic, a discrete hardware component, or any combination of these components. In addition, the processor 110 can also be any traditional processor, microprocessor or state machine. The processor 110 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a DSP, or any other such configuration.

[0269] The memory 120 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as the computer program corresponding to the end-intelligent product recommendation method in the embodiment of the present invention. The processor 110 executes various functional applications and data processing of the computer device 10 by running the non-volatile software programs, instructions and units stored in the memory 120, that is, implementing the end-intelligent product recommendation method in the above method embodiment.

[0270] The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store an application required for operating the device or at least one function; the data storage area may store data created according to the use of the computer device 10, etc. In addition, the memory 120 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 120 may optionally include a memory remotely arranged relative to the processor 110, and these remote memories may be connected to the computer device 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0271] One or more units are stored in the memory 120. When executed by one or more processors 110, the product recommendation method based on terminal intelligence in any of the above method embodiments can be implemented. For example, the above described Figure 1 The method comprises steps S100 to S300.

[0272] Those skilled in the art will understand that Figure 3 The hardware structure diagram shown in the figure is only a schematic diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0273] Based on the above method embodiments, another embodiment of the present invention further provides a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the product recommendation method based on terminal intelligence in any of the above method embodiments can be implemented, for example, the above described Figure 1 The method comprises steps S100 to S300.

[0274] As an example, the non-volatile storage medium can include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable ROM (EEPROM), or a flash memory. Volatile memory can include a random access memory (RAM) as an external cache memory. By way of illustration and not limitation, RAM can be obtained in many forms such as synchronous RAM (SRAM), dynamic RAM, (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synch link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable types of memory.

[0275] Based on the above method embodiments, another embodiment of the present invention further provides a computer program product, the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, the computer program includes program instructions, when the program instructions are executed by a processor, the product recommendation method based on end intelligence in any of the above method embodiments can be implemented, for example, the above described Figure 1 The method comprises steps S100 to S300.

[0276] The embodiments described above are merely illustrative, in which the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0277] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can exist in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0278] Conditional language such as "can," "can," "might," or "may," unless specifically stated otherwise or otherwise understood within the context as used, is generally intended to convey, among other things, that a particular embodiment can include (while other embodiments do not) a particular feature, element, and / or operation. Thus, such conditional language is also generally intended to imply that a feature, element, and / or operation is required for one or more embodiments anyway or that one or more embodiments must include logic for determining, with or without input or prompting, whether such features, elements, and / or operations are included or will be performed in any particular embodiment.

[0279] What has been described herein, in this specification and in the drawings includes examples of a method, device, computer equipment, and non-volatile computer-readable storage medium that can provide a product recommendation method based on end intelligence. Of course, it is not possible to describe every conceivable combination of elements and / or methods for the purpose of describing the various features of the present disclosure, but it can be recognized that many other combinations and permutations of the disclosed features are possible. Therefore, it is obvious that various modifications can be made to the present disclosure without departing from the scope or spirit of the present disclosure, but all these various modifications should fall within the scope of protection of the claims attached to the present invention. In addition, or in the alternative, other embodiments of the present disclosure may be apparent from consideration of the present specification and drawings and from the practice of the present disclosure as presented herein. It is intended that the examples presented in this specification and the drawings are considered to be illustrative and not restrictive in all respects. Although specific terms are employed herein, they are used in a general and descriptive sense and are not used for limiting purposes.

Claims

1. A product recommendation method based on terminal intelligence, characterized in that: include: The target user's operation behavior within the target time period is recorded through the terminal intelligent system; Based on the operation behavior, M target products are returned through the background recommendation system; Sorting the M target products, selecting N target products with the highest similarity to the operation behavior and recommending them to the target user; Wherein, M and N both represent positive integers, and M is greater than N.

2. The product recommendation method based on terminal intelligence according to claim 1, characterized in that: The operation behavior of the target user in the target time period is recorded by the terminal intelligent system, including: The end intelligent system records the behavior feature text of the target user's operation behavior within the target time period; The behavior feature text is converted into a corresponding behavior feature vector through a word vectorization system.

3. The product recommendation method based on terminal intelligence according to claim 2 is characterized in that: Based on the operation behavior, M target products are returned through the background recommendation system, including: Based on the behavior feature vector, returning product feature texts of M target products through the backend recommendation system; Each of the product feature texts is converted into a corresponding product feature vector through the word vectorization system.

4. The product recommendation method based on terminal intelligence according to claim 3 is characterized in that: The sorting of the M target products and selecting the N target products with the highest similarity to the operation behavior to recommend to the target user includes: Sorting the product feature vectors corresponding to the M target products by using a cosine similarity algorithm; Selecting N product feature vectors having the highest similarity to the behavior feature vector corresponding to the operation behavior from among the M product feature vectors; The target products corresponding to the N product feature vectors are recommended to the target users.

5. The product recommendation method based on terminal intelligence according to claim 4 is characterized in that: The recommending the target products corresponding to the N product feature vectors to the target user includes: The target products corresponding to the N product feature vectors are displayed in a target area to recommend them to the target users.

6. The product recommendation method based on terminal intelligence according to any one of claims 1 to 5, characterized in that: The end intelligent system records the operation behavior of the target user within a target time period, wherein the target time period includes the current day.

7. The product recommendation method based on terminal intelligence according to claim 6 is characterized in that: The end intelligent system records the operation behavior of the target user within the target time period, wherein the operation behavior includes one or more of the following behaviors: a. Views and clicks on waterfall articles; b. Waterfall video browsing and clicking; c. Waterfall product browsing and clicking; d. Browsing and clicking on hot-selling products; e. Article, video, and product search.

8. A product recommendation device based on terminal intelligence, characterized in that: include: The operation recording module is used to record the operation behavior of the target user within the target time period through the terminal intelligent system; A product return module, used to return M target products through a background recommendation system based on the operation behavior; A product recommendation module is used to sort the M target products, select the N target products with the highest similarity to the operation behavior and recommend them to the target user; Wherein, M and N both represent positive integers, and M is greater than N.

9. A computer device, characterized in that: The computer device comprises at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor. When the computer program is executed by the at least one processor, the end-intelligence-based product recommendation method as described in any one of claims 1 to 7 can be implemented.

10. A non-volatile computer-readable storage medium, characterized in that: The non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, it can implement the end-intelligence-based product recommendation method as described in any one of claims 1-7.

Images