Advertisement delivery method and system based on user big data analysis

Abstract

The present invention proposes an advertising delivery method and system based on user big data analysis. User information and advertising product information are obtained from a background database in real time; the user information includes user IP information, geographic location information and user identity information; the advertising product information includes product type information, product sales information, product browsing information and product search information; the advertising product information in the area where the user is located is obtained based on the user IP information and geographic location information, and the user identity information corresponding to the advertising product is obtained based on the user behavior data. By analyzing user big data, the user's interests, preferences, behavioral characteristics and other information can be deeply understood, thereby achieving accurate target user positioning. Advertisements can be delivered to users who are truly potentially interested, thereby improving the effect and conversion rate of advertising reach.

Description

Technical Field

[0001] The present invention proposes an advertisement delivery method and system based on user big data analysis, belonging to the technical field of big data analysis and advertisement delivery. Background Art

[0002] With the rapid development of the Internet, the application of big data technology is becoming more and more extensive. As an important means of marketing, how to reach target users more accurately is the focus of advertisers and advertising platforms. Traditional advertising methods are often based on simple user tags, lacking in-depth analysis and mining of user behavior, resulting in poor advertising effects. Therefore, developing an advertising method based on user big data analysis to improve the accuracy and effectiveness of advertising is an urgent problem that the current advertising industry needs to solve.

[0003] Traditional advertising methods are usually based on simple user tags, such as geographic location, age, gender, etc. These tags often lack in-depth analysis and mining of user behavior, resulting in poor advertising results. In addition, traditional advertising methods lack the acquisition and analysis of user identity information, and cannot deliver more accurate advertising based on user identity information. Summary of the invention

[0004] The present invention provides an advertising delivery method and system based on user big data analysis, which is used to solve the problem that the existing advertising delivery method cannot accurately deliver advertisements according to the characteristics of a specific group of people and cannot timely adjust the delivery strategy according to market feedback and actual conditions:

[0005] The present invention proposes an advertisement delivery method based on user big data analysis, the method comprising:

[0006] S1: Acquire user information and advertisement product information from the backend database in real time; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information;

[0007] S2: Obtaining the advertisement product information in the area where the user is located according to the user IP information and the geographic location information, and obtaining the user identity information corresponding to the advertisement product according to the user behavior data;

[0008] S3: Analyze the advertisement product information and the user identity information by using a machine learning algorithm combined with a deep learning algorithm, and establish user profiles of users of different genders and ages in the area according to the analysis results;

[0009] S4: establishing a product recommendation model according to the user portrait, and placing advertisement products according to the product recommendation model;

[0010] S5: Monitor the advertising effect in real time and continuously optimize the recommendation model based on the monitoring results.

[0011] Furthermore, the user information and advertisement product information are obtained in real time from the backend database; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information; including:

[0012] S11: Connecting to the backend database through the API interface, and monitoring the data changes of the user information and the advertising product information table through the monitoring mechanism provided by the backend database;

[0013] S12: When data changes are detected, corresponding events are triggered, where the data changes include insertion, update, modification, and deletion;

[0014] S13: Use Kafka to publish the data change status information in the database to the message queue in the form of messages;

[0015] S14: Every message received is processed and analyzed in real time by using machine learning algorithms;

[0016] S15: updating relevant user information and advertisement product information in real time according to the result of data processing; and storing the result of real-time updating relevant user information and advertisement product information in a database.

[0017] Furthermore, the method uses a machine learning algorithm to process and analyze each received message in real time; including:

[0018] S141: Divide the received message data set into small data blocks according to certain rules, and distribute the data blocks to available computing nodes;

[0019] S142: Each Map task obtains a data block. For each message, the Map module converts it into a key-value pair, where the key represents a feature of the message, including a keyword and a timestamp, and the value represents the message itself or a related message.

[0020] S143: The Map module outputs the generated intermediate results to the distributed file system in the form of key-value pairs;

[0021] S144: The Shuffle module partitions and sorts the intermediate results output by the Map stage according to the key, and sends the data with the same key to the same Reduce task node;

[0022] S145: Each Reduce module receives the partitioned intermediate result data from the Shuffle module;

[0023] S146: The Reduce module processes each received key-value pair, including aggregation, screening and statistics, to obtain a final output result;

[0024] S147: The Reduce module outputs the processed results to the database.

[0025] Furthermore, the user identity information includes user registration information, purchase preferences and user behavior data; the user registration information includes user gender, age, user name and mobile phone number; the purchase preferences include category preferences, brand preferences, price sensitivity, special demand media influence and regional cultural differences; the user behavior data includes user browsing behavior, click behavior, user purchase history and search behavior.

[0026] Furthermore, the advertising product information and the user identity information are analyzed by combining a machine learning algorithm with a deep learning algorithm, and user portraits of users of different genders and ages in the area are established according to the analysis results; including:

[0027] S31: Acquire a data set of advertisement product information and user identity information through a database; and pre-process the data set;

[0028] S32: extracting features from the user identity information, wherein the features include key features and main features, wherein the key features include user gender and age, and the main features include purchase preferences and user behavior data;

[0029] S33: Associating the advertisement product information with the user identity information, and connecting the two data sets through common attributes to form an integrated data set; the common attributes include the product ID and the user ID;

[0030] S34: Divide the entire data set into a training set and a test set; the training set is used to train the model, and the test set is used to evaluate the model performance;

[0031] S35: Train the model and tune parameters using the training set, and use the test set to evaluate and verify the trained model;

[0032] The performance and generalization ability of the model are evaluated by calculating indicators such as accuracy, recall, and F1 value.

[0033] S36: predict and classify new user data according to the trained model, divide users into different age groups of different genders according to the features, and establish user portraits;

[0034] S37: Visualize the user portrait results through visualization tools.

[0035] Furthermore, the step of establishing a product recommendation model according to the user portrait and placing advertisement products according to the product recommendation model includes:

[0036] Through collaborative filtering algorithms, the correlation and similarity between users and advertised products are calculated, and user portrait data is matched with the features of advertised products.

[0037] According to the matching degree between the user portrait and the advertisement product, a candidate advertisement product list is generated for each user; the candidate product list can be sorted according to the predicted user's preference degree for the advertisement product to determine the most suitable advertisement product.

[0038] Develop advertising delivery mechanisms and strategies based on advertising budget, delivery channels and marketing goals.

[0039] Furthermore, the real-time monitoring of the advertising effect and the continuous optimization of the recommendation model according to the monitoring effect include:

[0040] Determine key indicators for evaluating the effectiveness of advertising; the key indicators include click-through rate, conversion rate and return on investment;

[0041] The data monitoring module collects relevant data in the process of advertising delivery in real time, including advertising exposure rate, click volume and conversion events;

[0042] Based on the collected data, the data is analyzed through deep learning algorithms to obtain information on advertising delivery and generate monitoring reports;

[0043] Based on the monitoring results, adjust and optimize the selection and processing of advertising product features, adjust the parameters of the recommendation model, and evaluate and try different recommendation algorithms or improve existing algorithms;

[0044] Compare different optimization strategies through A / B testing methods; randomly divide a part of users into an experimental group and a control group, apply different models or strategies to advertise respectively, and evaluate the effect of the optimization strategy by comparing the changes in indicators of the two groups of data;

[0045] Establish a continuous monitoring mechanism, regularly check and evaluate the effectiveness of advertising, and perform iterative optimization based on feedback data; timely adjust the recommendation model and advertising strategy according to changes in different periods and needs.

[0046] The present invention proposes an advertisement delivery system based on user big data analysis, the system comprising:

[0047] Data acquisition module: acquires user information and advertisement product information from the backend database in real time; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information;

[0048] Area information determination module: obtains the advertisement product information in the area where the user is located according to the user IP information and geographic location information, and obtains the user identity information corresponding to the advertisement product according to the user behavior data;

[0049] User portrait building module: analyzes the advertising product information and the user identity information through a machine learning algorithm combined with a deep learning algorithm, and builds user portraits of users of different genders and ages in the area based on the analysis results;

[0050] Advertisement product recommendation module: establish a product recommendation model based on the user portrait, and place advertisement products based on the product recommendation model;

[0051] Model optimization module: monitor the advertising effect in real time and continuously optimize the recommendation model based on the monitoring results.

[0052] An electronic device proposed in the present invention includes a memory, a processor, and a computer program stored in and executable on the memory, wherein the processor executes the program to implement any of the above-mentioned advertising delivery methods based on user big data analysis.

[0053] The present invention proposes a non-temporary computer-readable storage medium, on which a computer program is stored. The program is executed by a processor to implement any of the above-mentioned advertising delivery methods based on user big data analysis.

[0054] The invention has the beneficial effects that, by analyzing user big data, the user's interests, preferences, behavioral characteristics and other information can be deeply understood, thereby achieving accurate target user positioning. Advertisements can be delivered to users who are truly potentially interested, thereby improving the effect and conversion rate of advertising reach; by deeply analyzing user big data, user groups interested in advertising can be identified, thereby optimizing advertising content, delivery channels and timing. In this way, the click-through rate, conversion rate and return on investment of advertisements can be improved, and the effect and benefit of advertising delivery can be improved; by analyzing user big data, the personalized needs and preferences of users can be understood, thereby adjusting the advertising content and recommendation strategy according to the personalized characteristics of users. Personalized advertisements can better attract the user's attention, increase the user's participation and willingness to buy; based on the analysis of user big data, the advertising delivery effect can be monitored in real time, and optimized and adjusted according to the monitoring results. In this way, problems and improvement space can be quickly discovered, and the effect and delivery efficiency of advertisements can be improved; through analysis based on user big data, more personalized and relevant advertising content and recommendations can be provided to users; the user's sense of interference with advertisements can be reduced, and user satisfaction and brand awareness can be improved; through accurate target user positioning and optimized delivery strategies, the waste of advertising resources can be avoided and unnecessary advertising delivery costs can be reduced. At the same time, improving the click-through rate and conversion rate of ads can achieve better advertising results with less advertising costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a step diagram of an advertisement delivery method based on user big data analysis according to the present invention;

[0056] Figure 2 This is a module diagram of an advertising delivery system based on user big data analysis described in the present invention. DETAILED DESCRIPTION

[0057] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0058] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0060] One embodiment of the present invention, as Figure 1 As shown, an advertisement delivery method based on user big data analysis includes:

[0061] S1: Acquire user information and advertisement product information from the backend database in real time; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information;

[0062] S2: Obtaining the advertisement product information in the area where the user is located according to the user IP information and the geographic location information, and obtaining the user identity information corresponding to the advertisement product according to the user behavior data;

[0063] S3: Analyze the advertisement product information and the user identity information by using a machine learning algorithm combined with a deep learning algorithm, and establish user profiles of users of different genders and ages in the area according to the analysis results;

[0064] S4: establishing a product recommendation model according to the user portrait, and placing advertisement products according to the product recommendation model;

[0065] S5: Monitor the advertising effect in real time and continuously optimize the recommendation model based on the monitoring results.

[0066] The working principle of the above technical solution is as follows: through the background database, the user's IP information, geographic location information and user identity information, as well as the type information, sales information, browsing information and search information of the advertising products are obtained in real time; according to the user's IP information and geographic location information, the user's area is determined, and the advertising product information in the area is obtained from the database; according to the user's behavior data, such as browsing history, search history, etc., the association information between the user and the advertising product, that is, the user identity information corresponding to the advertising product, is obtained; the advertising product information and user identity information are analyzed using machine learning algorithms and deep learning algorithms to understand the interests, preferences and consumption behaviors of users of different genders and different age groups, and establish user portraits; according to the user portraits, a product recommendation model is established, which can recommend advertising products according to the user's characteristics and historical behaviors; according to the product recommendation model, the advertising products are delivered. Through precise positioning and personalized recommendations, advertisements are delivered to the target user group to improve the conversion rate and user participation of advertisements; the delivery effect of advertisements is monitored in real time, including indicators such as click-through rate and conversion rate. Data analysis is performed based on the monitoring results to evaluate the effect of advertisements and continuously optimize the recommendation model.

[0067] The effects of the above technical solutions are as follows: by obtaining user information and advertising product information in real time through the background database, the timeliness and accuracy of the data can be ensured, providing a basis for subsequent analysis and recommendation; through user IP information and geographic location information, advertising products can be accurately delivered to users in the corresponding area, improving the coverage and accuracy of advertising; by combining deep learning algorithms to analyze advertising product information and user identity information, it is possible to explore user behavior patterns and preferences, and further understand user needs and interests; through the analysis results, user portraits of users of different genders and ages can be established to help advertisers better understand the target user groups and thus optimize advertising delivery strategies; based on user portraits, a product recommendation model is established, which can accurately recommend advertising products suitable for users based on their interests and preferences, thereby improving the click-through rate and conversion rate of advertisements; by real-time monitoring of advertising delivery effects, problems and optimization space can be discovered in a timely manner, thereby improving the effectiveness and ROI (return on investment) of advertising; based on the monitoring results, the recommendation model is continuously optimized, and the recommendation algorithm and parameters are adjusted to improve the effectiveness and accuracy of advertising delivery.

[0068] In one embodiment of the present invention, the user information and the advertisement product information are obtained in real time from the background database; the user information includes the user IP information, the geographic location information and the user identity information; the advertisement product information includes the product type information, the product sales information, the product browsing information and the product search information; including:

[0069] S11: Connecting to the backend database through the API interface, and monitoring the data changes of the user information and the advertising product information table through the monitoring mechanism provided by the backend database;

[0070] S12: When data changes are detected, corresponding events are triggered, where the data changes include insertion, update, modification, and deletion;

[0071] S13: Use Kafka to publish the data change status information in the database to the message queue in the form of messages;

[0072] S14: Every message received is processed and analyzed in real time by using machine learning algorithms;

[0073] S15: updating relevant user information and advertisement product information in real time according to the result of data processing; and storing the result of real-time updating relevant user information and advertisement product information in a database.

[0074] The working principle of the above technical solution is as follows: connect with the backend database through the API interface, and use the monitoring mechanism provided by the backend database to monitor the data changes of the user information table and the advertising product information table in real time; when data changes are detected, the system will trigger corresponding events, including data insertion, update, modification and deletion operations; use Kafka as a message queue, and publish the data change information in the database to the message queue in the form of messages; Kafka has the characteristics of high throughput and low latency, and can effectively process a large number of messages; after receiving the data change information in the message queue, the system uses machine learning algorithms to process and analyze each message in real time. This can include steps such as data cleaning, feature extraction, and model training to obtain more valuable information; according to the results of data processing, the system updates the relevant user information and advertising product information in real time, and stores the updated results in the backend database for subsequent query and use.

[0075] The effect of the above technical solution is: by obtaining user information in real time, including IP information, geographic location information and user identity information, the system can push personalized advertisements based on this information. For example, according to the characteristics and preferences of the user's region, the system can display locally relevant advertising products to the user, improve the click-through rate and conversion rate of advertisements, and thus increase the sales of merchants and advertisers; by monitoring the data changes in the advertising product information table, the system can obtain the sales information, browsing information and search information of the product in real time. Real-time updated information can help merchants better understand market demand and user behavior, optimize product pricing strategies, promotional activities and recommendation strategies, and improve sales performance and user satisfaction; by using machine learning algorithms to process and analyze each received message in real time, the system can mine potential patterns and trends hidden in massive data. It helps to predict user behavior and interests, gain insight into market trends in advance, and thus formulate more accurate marketing strategies for merchants, reduce market risks and improve competitiveness; by adopting a combination of message queues and background databases, the system can process and store large amounts of data with high throughput and low latency. This enables the system to cope with high-concurrency data access and updates, and ensure the reliability and consistency of the data; due to the real-time nature of the technical solution, merchants can instantly obtain user feedback and market data, and adjust advertising push and product strategies in a timely manner to adapt to changing market demands and user preferences, thereby improving the accuracy and agility of business decisions.

[0076] In one embodiment of the present invention, the real-time processing and analysis of each received message is performed by using a machine learning algorithm; comprising:

[0077] S141: Divide the received message data set into small data blocks according to certain rules, and distribute the data blocks to available computing nodes;

[0078] S142: Each Map task obtains a data block. For each message, the Map module converts it into a key-value pair, where the key represents a feature of the message, including a keyword and a timestamp, and the value represents the message itself or a related message.

[0079] S143: The Map module outputs the generated intermediate results to the distributed file system in the form of key-value pairs;

[0080] S144: The Shuffle module partitions and sorts the intermediate results output by the Map stage according to the key, and sends the data with the same key to the same Reduce task node;

[0081] S145: Each Reduce module receives the partitioned intermediate result data from the Shuffle module;

[0082] S146: The Reduce module processes each received key-value pair, including aggregation, screening and statistics, to obtain a final output result;

[0083] S147: The Reduce module outputs the processed results to the database.

[0084] The working principle of the above technical solution is as follows: the received message data set will be divided into small data blocks according to certain rules, and these data blocks will be distributed to the available computing nodes. This is done to decompose large-scale data sets into smaller tasks for parallel processing and to improve computing efficiency; the Map task on each computing node will obtain a data block, and for each message, the Map module will convert it into a key-value pair. The key represents the characteristics of the message, such as keywords and timestamps, while the value represents the message itself or other information related to it. The Map module outputs the generated intermediate results in the form of key-value pairs to the distributed file system for subsequent processing; the Shuffle module partitions and sorts the intermediate results output by the Map stage according to the key. It sends data with the same key to the same Reduce task node for subsequent processing and aggregation operations; each Reduce task node receives the partitioned intermediate result data from the Shuffle stage. The Reduce module processes each received key-value pair, including aggregation, filtering, and statistics operations, to obtain the final output result; the Reduce module outputs the processed results to the database for subsequent data analysis and use.

[0085] The effect of the above technical solution is: by dividing the large-scale message data set into small data blocks and distributing them to multiple computing nodes for parallel processing, the processing speed and computing efficiency can be improved. Each computing node can independently process the data block assigned to it, thereby reducing the computing load of a single node and speeding up the processing speed of the entire system; real-time processing and analysis of each received message can be achieved. The Map stage converts and extracts features for each message, and the Reduce stage aggregates, filters, and counts the messages, and finally obtains real-time output results. This is very important for application scenarios that require rapid response and timely decision-making; distributed computing and MapReduce frameworks have good scalability, and computing nodes can be flexibly added according to demand to adapt to the growing amount of message data and processing requirements. By adding computing nodes, the system can smoothly expand processing capabilities and ensure efficient and reliable processing services; because the data is divided and distributed to multiple computing nodes for processing, even if a computing node fails, the entire system can still operate normally and will not affect the work of other nodes. This distributed fault tolerance ensures the high availability and stability of the system; through the MapReduce framework, the system can perform multi-dimensional processing and analysis on the received message data, including feature extraction, aggregation, screening, statistics and other operations. These processing and analysis operations can help users understand and mine the potential information and patterns in the message data, thereby supporting more in-depth data analysis and application development.

[0086] In one embodiment of the present invention, the user identity information includes user registration information, purchase preferences and user behavior data; the user registration information includes user gender, age, user name and mobile phone number; the purchase preferences include category preference, brand preference, price sensitivity, special demand media influence and regional cultural differences; the user behavior data includes user browsing behavior, click behavior, user purchase history and search behavior.

[0087] The working principle of the above technical solution is: the user identity information includes user registration information, purchase preferences and user behavior data; the user registration information includes user gender, age, user name and mobile phone number; the purchase preferences include category preferences, brand preferences, price sensitivity, special demand media influence and regional cultural differences; the user behavior data includes user browsing behavior, click behavior, user purchase history and search behavior.

[0088] Specific:

[0089] Category Preference: Users may have preferences for certain product categories, such as fashion apparel, electronics, household items, food and beverages, etc. They are more inclined to buy products that they are interested in or use frequently.

[0090] Brand preference: Users may have a favorable impression or trust towards certain specific brands and tend to buy products from these brands. Factors such as brand image, quality, and reputation may influence users’ purchasing decisions.

[0091] Price sensitivity: User sensitivity to price is also an aspect of purchasing preference. Some users prefer to buy lower-priced products, while others value product quality and value more and are willing to pay a higher price for high-quality products.

[0092] Special needs: Users may have special needs, such as requirements for specific functions, materials, specifications or designs. They will pay more attention to whether the product meets their special needs and choose to buy according to their needs.

[0093] Purchasing habits: Users’ purchasing habits and shopping preferences are also part of their purchasing preferences. For example, some users prefer online shopping, while others prefer to go to physical stores; some users are more inclined to buy discounted products, while others focus more on buying full-priced products.

[0094] Media influence and social factors: Users may be influenced by the media, advertising, friends or social circles, and develop an interest in and a liking for certain products or brands, thus forming a purchasing preference.

[0095] Regional and cultural differences: Users in different regions and cultural backgrounds may have different product needs and purchasing preferences. Regional unique culture, customs and habits and other factors will affect users' purchasing decisions and preferences.

[0096] The effect of the above technical solution is: through the analysis of user registration information and purchase preferences, the user's preferences, interests and needs can be understood. Based on this information, the system can provide personalized recommendation services to recommend products or services that meet the user's category preferences, brand preferences and special needs. Personalized recommendations can improve user satisfaction, promote user purchase decisions, and increase sales; through the comprehensive analysis of user registration information, purchase preferences and user behavior data, users can be segmented into different groups or types. For example, users can be divided according to dimensions such as gender, age, regional cultural differences, etc., to understand the preferences and needs of different groups, and to carry out marketing activities and service strategies in a targeted manner. Customer segmentation can improve market accuracy, optimize resource utilization, and enhance user stickiness; by analyzing user behavior data, such as browsing behavior, click behavior and purchase history, the user's behavior trajectory and purchasing habits can be understood. The system can provide personalized guidance and push based on user behavior data, and provide services and products that are more in line with user interests and needs. This helps to improve user retention and loyalty, increase user repurchase and long-term value; by analyzing user registration information, purchase preferences and user behavior data, a large amount of market feedback and user opinions can be obtained. This data can be used for market research to understand user satisfaction, needs and improvement suggestions for products or services. Based on this feedback and opinions, companies can optimize products and improve services to enhance product competitiveness and user experience.

[0097] In one embodiment of the present invention, the advertising product information and the user identity information are analyzed by combining a machine learning algorithm with a deep learning algorithm, and user portraits of users of different genders and ages in the region are established according to the analysis results; including:

[0098] S31: Acquire a data set of advertisement product information and user identity information through a database; and pre-process the data set;

[0099] S32: extracting features from the user identity information, wherein the features include key features and main features, wherein the key features include user gender and age, and the main features include purchase preferences and user behavior data;

[0100] S33: Associating the advertisement product information with the user identity information, and connecting the two data sets through common attributes to form an integrated data set; the common attributes include the product ID and the user ID;

[0101] S34: Divide the entire data set into a training set and a test set; the training set is used to train the model, and the test set is used to evaluate the model performance;

[0102] S35: Train the model and optimize its parameters using the training set, and use the test set to evaluate and verify the trained model; evaluate the performance and generalization ability of the model by calculating indicators such as accuracy, recall, and F1 value;

[0103] S36: predict and classify new user data according to the trained model, divide users into different age groups of different genders according to the features, and establish user portraits;

[0104] S37: Visualize the user portrait results through visualization tools.

[0105] The working principle of the above technical solution is as follows: obtain the data set of advertising product information and user identity information from the database and perform preprocessing. Preprocessing includes data cleaning, missing value processing, feature standardization, etc. to ensure the quality and consistency of the data; extract key features from the user identity information, including user gender and age, as well as main features such as purchase preferences and user behavior data. These features will be used as input for model training; associate the advertising product information and user identity information through common attributes (such as product ID and user ID), and connect the two data sets into an overall data set; the overall data set is divided into a training set and a test set. The training set is used for model training and parameter tuning, and the test set is used to evaluate the performance and generalization ability of the model; the model is trained and parameter tuned by using the training set. Machine learning algorithms and deep learning algorithms will learn the complex relationship between advertising product information and user identity information and build a model. Use the test set to evaluate and verify the trained model. Calculate indicators such as accuracy, recall rate, F1 value, etc. to evaluate the performance and generalization ability of the model. According to the trained model, predict and classify new user data. According to user characteristics, such as gender and age, users are divided into different genders and age groups, and user portraits are created; finally, the user portrait results are visualized through visualization tools. In this way, companies can more intuitively understand the characteristics of users of different genders and age groups to formulate more accurate marketing strategies and personalized services.

[0106] The effects of the above technical solution are: by preprocessing the acquired advertising product information and user identity information data, the quality and consistency of the data can be improved, thereby improving the accuracy of subsequent model training and prediction; extracting key features and main features from user identity information, including gender, age, purchase preferences and user behavior data, etc., these features help to better understand and describe the characteristics and behaviors of users; by associating advertising product information and user identity information through common attributes, the two data sets can be connected into an overall data set, providing basic data for subsequent model training and classification; training and parameter tuning of the overall data set through machine learning algorithms and deep learning algorithms, building a model, and evaluating the performance and generalization ability of the model through the test set. This can improve the accuracy and reliability of the model; based on the trained model, predict and classify new user data, divide users into different genders and age groups, and build user portraits. This can better understand the characteristics and needs of user groups; by visualizing the user portrait results through visualization tools, enterprises can more intuitively understand the characteristics and behavioral preferences of users of different genders and age groups. This provides a basis for formulating precise marketing strategies and personalized services. In summary, this technical solution can help companies gain a deeper understanding of their user groups, and develop more precise marketing strategies and personalized services based on the characteristics of users of different genders and age groups, thereby improving the company's marketing effectiveness and user satisfaction.

[0107] In one embodiment of the present invention, the method of establishing a product recommendation model according to the user portrait and placing advertisement products according to the product recommendation model comprises:

[0108] Through collaborative filtering algorithms, the correlation and similarity between users and advertised products are calculated, and user portrait data is matched with the features of advertised products.

[0109] According to the matching degree between the user portrait and the advertisement product, a candidate advertisement product list is generated for each user; the candidate product list can be sorted according to the predicted user's preference degree for the advertisement product to determine the most suitable advertisement product.

[0110] According to the advertising budget, distribution channels and marketing goals, formulate advertising distribution mechanism and strategy, and allocate the budget for the next advertising distribution cycle. The calculation formula is:

[0111] ;

[0112] in, Indicates advertising campaign In the The time period of the advertising cycle The number of ad impressions received that meet the delivery rules. G represents the total budget of the ad campaign a set by the advertiser in a complete ad delivery cycle. Represents the number of time periods within an advertising delivery cycle. V is a time window set by the demand-side platform. V ≥ 1, which is used to limit the data of the most recent V advertising delivery cycles as the basis for prediction and statistics.

[0113] The working principle of the above technical solution is: to make product recommendations by calculating the correlation and similarity between users and advertised products through a collaborative filtering algorithm, and to match user portrait data with the characteristics of advertised products. For example, the user's gender, age, purchasing preferences and other characteristics are compared and matched with the attributes of the advertised products to determine the user's preference for the advertised products. Generate a list of candidate advertising products: Based on the degree of matching, a list of candidate advertising products is generated for each user. The advertising products in the list are sorted according to the predicted user's preference for the products to determine the most suitable advertising products. Advertising delivery mechanism and strategy: Develop advertising delivery mechanisms and strategies based on the advertising budget, delivery channels and marketing goals. Among them, when formulating advertising delivery mechanisms and strategies, the following aspects need to be considered:

[0114] Budget allocation: Determine the amount of money to be spent on each advertising channel or media based on the budget of the advertising campaign. This can be done by evaluating and analyzing factors such as the number of audiences, coverage, and advertising pricing of different channels. For example, the budget is RMB 100,000. Allocate 40% of the budget to Internet advertising, 30% to TV advertising, 20% to radio advertising, and 10% to outdoor advertising.

[0115] Delivery time: Determine the best delivery time based on the target audience's behavior habits and data analysis of advertising effectiveness. For example, if the target users are more inclined to browse shopping websites at night, the exposure of the advertisement can be increased at night. For example, through user behavior data analysis, it is found that 8 to 10 pm is the time when the target audience uses smart devices more. Therefore, it is decided to increase the frequency and exposure of advertising from 8 to 10 pm.

[0116] Advertising channel selection: Choose the appropriate advertising channel based on the characteristics of the target audience and the advertising goals. Different channels have different characteristics and audiences. For example, you can reach young people through social media platforms, while TV advertising can cover a wider user group. For example, considering that the target audience is mostly young people, you choose to place advertisements on social media platforms (such as WeChat and Weibo) and cooperate with some technology blogs to push advertising content. In addition, it is decided to place advertisements in TV stations' technology programs to cover a wider user group.

[0117] Positioning and targeting: Accurate positioning and targeting are performed based on user portraits and the characteristics of the advertised products. The targeting function provided by the advertising platform is used to display ads to users who meet the target audience conditions. For example, based on user portraits and the characteristics of the advertised products, ads are targeted to male users between the ages of 18 and 35, who are interested in purchasing electronic products and following technology information. In addition, ads are targeted to users in first- and second-tier cities based on their geographic locations.

[0118] The effects of the above technical solution are as follows: by establishing a user portrait and product recommendation model, personalized advertising product recommendations can be provided to each user based on the user's interests and characteristics. This can increase the user's interest in advertising and willingness to buy, and improve the conversion rate of advertising; by calculating the correlation and similarity between the user and the advertising product, and matching the user portrait and the characteristics of the advertising product, the advertising products related to the user's needs can be more accurately recommended and delivered. This can improve the targeting and accuracy of advertising, reduce the noise of advertising push and the user's disgust; by generating a list of candidate advertising products for each user and sorting them according to the user's preference for the product, the most suitable advertising product can be selected for delivery. This can improve the click-through rate, conversion rate and return rate of advertising, and effectively improve the effect and commercial value of advertising delivery; personalized advertising recommendations can provide product information that is more in line with the user's interests and needs, and provide users with a better shopping experience. Users see more products that meet their preferences in advertisements, which can reduce information overload and advertising interference, and enhance users' recognition and acceptance of advertising; formulate advertising delivery mechanisms and strategies based on advertising budgets, delivery channels and marketing goals. Advertising resources can be reasonably allocated, appropriate delivery channels and timing can be selected, and the effectiveness and cost-effectiveness of advertising can be improved. The above formula can achieve more accurate advertising, improve advertising exposure and conversion effects, while reducing budget waste and achieving advertisers' marketing goals. At the same time, by predicting and statistics based on the data of the most recent V cycles, the changing trends of user behavior and the effectiveness of advertising can be more accurately reflected, thereby adjusting the allocation of budgets. This enables advertisers to make decisions in a timely manner and improve the flexibility and efficiency of advertising; the allocation mechanism in the formula allocates the budget in a weighted manner according to the number of ad display opportunities, so that the budget is more inclined to time periods with more display opportunities. This can maximize the exposure of ads, increase the click-through rate and conversion rate of ads, and achieve better advertising results; the formula It reflects the number of ad display opportunities that meet the delivery rules. By counting and predicting the opportunities that meet the delivery rules, you can better control the quality and accuracy of ad delivery and avoid inefficient delivery and wasted budget.

[0119] In one embodiment of the present invention, the real-time monitoring of the advertising delivery effect and the continuous optimization of the recommendation model according to the monitoring effect include:

[0120] Determine key indicators for evaluating the effectiveness of advertising; the key indicators include click-through rate, conversion rate and return on investment;

[0121] The data monitoring module collects relevant data in the process of advertising delivery in real time, including advertising exposure rate, click volume and conversion events;

[0122] Based on the collected data, the data is analyzed through deep learning algorithms to obtain information on advertising delivery and generate monitoring reports;

[0123] Based on the monitoring results, adjust and optimize the selection and processing of advertising product features, adjust the parameters of the recommendation model, and evaluate and try different recommendation algorithms or improve existing algorithms;

[0124] Compare different optimization strategies through A / B testing methods; randomly divide a part of users into an experimental group and a control group, apply different models or strategies to advertise respectively, and evaluate the effect of the optimization strategy by comparing the changes in indicators of the two groups of data;

[0125] Establish a continuous monitoring mechanism, regularly check and evaluate the effectiveness of advertising, and perform iterative optimization based on feedback data; timely adjust the recommendation model and advertising strategy according to changes in different periods and needs.

[0126] The working principle of the above technical solution is as follows: according to the goals and needs of advertising delivery, key indicators for evaluating the effectiveness of advertising delivery, such as click-through rate (CTR), conversion rate (CVR) and return on investment (ROI) are determined; these indicators can quantify the effectiveness and commercial value of advertising, and are used to evaluate and compare the effects of different optimization strategies; in the process of advertising delivery, relevant data, including advertising exposure rate, click volume and conversion events, are collected in real time through the data monitoring module. These data can provide real-time information on the situation and effectiveness of advertising delivery, and provide a basis for subsequent analysis and optimization; by applying deep learning algorithms to analyze and process the collected data, detailed information on the advertising delivery situation is obtained; by analyzing the relationship between user behavior and advertising characteristics, potential laws and trends are discovered, and monitoring reports are generated for reference; according to the monitoring results, the parameters of the recommendation model are adjusted for the selection and processing of advertising product characteristics, and different recommendation algorithms are evaluated and tried or existing algorithms are improved. The model is trained and optimized using historical data and real-time data to improve the accuracy and effectiveness of advertising recommendations; through the A / B testing method, a part of users are randomly divided into an experimental group and a control group, and different models or strategies are applied to advertising delivery respectively. By comparing the changes in indicators of the two groups of data, the effectiveness of the optimization strategy is evaluated. Select the best solution based on the test results, and make subsequent optimizations and adjustments; establish a continuous monitoring mechanism, regularly check and evaluate the effectiveness of advertising, and iterate and optimize based on feedback data. As time and demand change, continue to improve the recommendation model and advertising strategy to maintain the effectiveness and competitiveness of advertising.

[0127] The effects of the above technical solutions are as follows: by collecting and analyzing data in the process of advertising in real time, it is possible to timely understand indicators such as advertising exposure rate, click volume and conversion events, so as to take optimization measures in time; by determining key indicators such as click-through rate, conversion rate and return on investment, it is possible to objectively evaluate the effect of advertising and use it as a basis for optimization; by using deep learning algorithms to analyze advertising data, it is possible to identify user behavior patterns and preferences, and then accurately carry out targeted advertising to improve conversion rate and return on investment; based on the collected data, a monitoring report is generated to provide advertisers with detailed advertising delivery and effect analysis to help them understand the advertising effect and optimize advertising strategies; according to the monitoring results, the selection and processing methods of advertising product features can be adjusted and optimized to improve the attractiveness and click-through rate of the advertisement; according to the monitoring results, different recommendation algorithms can be evaluated and tried or existing algorithms can be improved, and the parameters of the recommendation model can be adjusted to improve the effect of advertising; through the A / B testing method, the effects of different optimization strategies can be compared to help determine the best advertising delivery plan, and experimental data can be provided to support decision-making; a continuous monitoring mechanism is established to regularly check and evaluate the effect of advertising delivery, and iterative optimization is carried out based on feedback data to maintain the competitiveness and effect of advertising delivery.

[0128] One embodiment of the present invention, as Figure 2 As shown, an advertisement delivery system based on user big data analysis includes:

[0129] Data acquisition module: acquires user information and advertisement product information from the backend database in real time; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information;

[0130] Area information determination module: obtains the advertisement product information in the area where the user is located according to the user IP information and geographic location information, and obtains the user identity information corresponding to the advertisement product according to the user behavior data;

[0131] User portrait building module: analyzes the advertising product information and the user identity information through a machine learning algorithm combined with a deep learning algorithm, and builds user portraits of users of different genders and ages in the area based on the analysis results;

[0132] Advertisement product recommendation module: establish a product recommendation model based on the user portrait, and place advertisement products based on the product recommendation model;

[0133] Model optimization module: monitor the advertising effect in real time and continuously optimize the recommendation model based on the monitoring results.

[0134] The working principle of the above technical solution is as follows: through the background database, the user's IP information, geographic location information and user identity information, as well as the type information, sales information, browsing information and search information of the advertising products are obtained in real time; according to the user's IP information and geographic location information, the user's area is determined, and the advertising product information in the area is obtained from the database; according to the user's behavior data, such as browsing history, search history, etc., the association information between the user and the advertising product, that is, the user identity information corresponding to the advertising product, is obtained; the advertising product information and user identity information are analyzed using machine learning algorithms and deep learning algorithms to understand the interests, preferences and consumption behaviors of users of different genders and different age groups, and establish user portraits; according to the user portraits, a product recommendation model is established, which can recommend advertising products according to the user's characteristics and historical behaviors; according to the product recommendation model, the advertising products are delivered. Through precise positioning and personalized recommendations, advertisements are delivered to the target user group to improve the conversion rate and user participation of advertisements; the delivery effect of advertisements is monitored in real time, including indicators such as click-through rate and conversion rate. Data analysis is performed based on the monitoring results to evaluate the effect of advertisements and continuously optimize the recommendation model.

[0135] The effects of the above technical solutions are as follows: by obtaining user information and advertising product information in real time through the background database, the timeliness and accuracy of the data can be ensured, providing a basis for subsequent analysis and recommendation; through user IP information and geographic location information, advertising products can be accurately delivered to users in the corresponding area, improving the coverage and accuracy of advertising; by combining deep learning algorithms to analyze advertising product information and user identity information, it is possible to explore user behavior patterns and preferences, and further understand user needs and interests; through the analysis results, user portraits of users of different genders and ages can be established to help advertisers better understand the target user groups and thus optimize advertising delivery strategies; based on user portraits, a product recommendation model is established, which can accurately recommend advertising products suitable for users based on their interests and preferences, thereby improving the click-through rate and conversion rate of advertisements; by real-time monitoring of advertising delivery effects, problems and optimization space can be discovered in a timely manner, thereby improving the effectiveness and ROI (return on investment) of advertising; based on the monitoring results, the recommendation model is continuously optimized, and the recommendation algorithm and parameters are adjusted to improve the effectiveness and accuracy of advertising delivery.

[0136] One embodiment of the present invention is an electronic device, comprising a memory, a processor, and a computer program stored in and executable on the memory, wherein the processor executes the program to implement any of the above-mentioned advertising delivery methods based on user big data analysis.

[0137] One embodiment of the present invention is a non-temporary computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement any of the above-described advertising delivery methods based on user big data analysis.

[0138] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An advertising delivery method based on user big data analysis, characterized in that: The method comprises: Acquire user information and advertisement product information from the backend database in real time; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information; Obtain the advertising product information in the area where the user is located based on the user's IP information and geographic location information, and obtain the user identity information corresponding to the advertising product based on the user's behavior data; Analyze the advertisement product information and the user identity information by using a machine learning algorithm combined with a deep learning algorithm, and establish user profiles of users of different genders and ages in the area based on the analysis results; Establishing a product recommendation model based on the user portrait, and placing advertisement products based on the product recommendation model; Monitor the advertising effect in real time and continuously optimize the recommendation model based on the monitoring results; The user information and advertisement product information are obtained in real time from the backend database; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information; including: Connect with the backend database through the API interface, and monitor the data changes of user information and advertising product information table through the monitoring mechanism provided by the backend database; When data changes are detected, corresponding events are triggered, and the data changes include insertion, update, modification and deletion; Through Kafka, the data change status information in the database is published to the message queue in the form of messages; By using machine learning algorithms, every message received is processed and analyzed in real time; According to the result of data processing, relevant user information and advertisement product information are updated in real time; and the result of real-time updating of relevant user information and advertisement product information is stored in the database; The system processes and analyzes each received message in real time by using machine learning algorithms, including: Divide the received message data set into small data blocks according to certain rules, and distribute the data blocks to available computing nodes; Each Map task gets a data block. For each message, the Map module converts it into a key-value pair. The Map module outputs the generated intermediate results to the distributed file system in the form of key-value pairs; The Shuffle module partitions and sorts the intermediate results output by the Map phase according to the key, and sends the data with the same key to the same Reduce task node; Each Reduce module receives partitioned intermediate result data from the Shuffle module; The Reduce module processes each received key-value pair, including aggregation, screening and statistics, to obtain the final output result; The Reduce module outputs the processed results to the database; The step of establishing a product recommendation model according to the user portrait and placing advertisement products according to the product recommendation model comprises: Through collaborative filtering algorithms, the correlation and similarity between users and advertised products are calculated, and user portrait data is matched with the features of advertised products. Generating a candidate advertising product list for each user according to the matching degree between the user portrait and the advertising product; Formulate advertising delivery mechanisms and strategies based on advertising budget, delivery channels and marketing goals; and allocate budget for the next advertising delivery cycle; The calculation formula for allocation is: ； in, Indicates that campaign x is The number of ad display opportunities that meet the delivery rules received in time period y of an ad delivery cycle, G represents the total budget of ad campaign a set by the advertiser in a complete ad delivery cycle, Y represents the number of time periods in the ad delivery cycle, V is a time window set by the demand-side platform, V ≥ 1, which is used to limit the data of the most recent V ad delivery cycles as the basis for prediction and statistics.

2. The method for placing advertisements based on user big data analysis according to claim 1, characterized in that: The user identity information includes user registration information, purchase preferences and user behavior data; the user registration information includes user gender, age, user name and mobile phone number; the purchase preferences include category preferences, brand preferences, price sensitivity, special demand media influence and regional cultural differences; the user behavior data includes user browsing behavior, click behavior, user purchase history and search behavior.

3. The method for placing advertisements based on user big data analysis according to claim 1, characterized in that: The method of analyzing the advertisement product information and the user identity information by combining a machine learning algorithm with a deep learning algorithm, and establishing user portraits of users of different genders and ages in the region according to the analysis results, includes: Acquire a data set of advertisement product information and user identity information through a database; and pre-process the data set; Extracting features from the user identity information, the features including key features and main features, the key features including user gender and age, and the main features including purchase preferences and user behavior data; Associate the advertising product information with the user identity information, and connect the two data sets through common attributes to form an integrated data set; Dividing the overall data set into a training set and a test set; Train the model and tune its parameters using the training set, and use the test set to evaluate and verify the trained model; According to the trained model, new user data is predicted and classified, and users are divided into different age groups of different genders through the above features, and user portraits are established; The user portrait results are visualized through visualization tools.

4. The method for placing advertisements based on user big data analysis according to claim 1, characterized in that: The real-time monitoring of the advertising effect and the continuous optimization of the recommendation model according to the monitoring effect include: Determine key indicators for evaluating the effectiveness of advertising; the key indicators include click-through rate, conversion rate and return on investment; The data monitoring module collects relevant data in the process of advertising delivery in real time, including advertising exposure rate, click volume and conversion events; Based on the collected data, the data is analyzed through deep learning algorithms to obtain information on advertising delivery and generate monitoring reports; Based on the monitoring results, adjust and optimize the selection and processing of advertising product features, adjust the parameters of the recommendation model, and evaluate and try different recommendation algorithms or improve existing algorithms; Compare different optimization strategies through A / B testing methods; randomly divide a part of users into an experimental group and a control group, apply different models or strategies to advertise, and evaluate the effect of the optimization strategy by comparing the changes in indicators of the two groups of data; Establish a continuous monitoring mechanism, regularly check and evaluate the effectiveness of advertising, and perform iterative optimization based on feedback data; timely adjust the recommendation model and advertising strategy according to changes in different periods and needs.

5. A system for implementing the method for placing advertisements based on user big data analysis as claimed in claim 1, characterized in that: The system comprises: Data acquisition module: acquires user information and advertisement product information from the backend database in real time; the user information includes user IP information, geographic location information and user identity information; the advertisement product information includes product type information, product sales information, product browsing information and product search information; Regional information determination module: obtains the advertising product information in the area where the user is located based on the user's IP information and geographic location information, and obtains the user identity information corresponding to the advertising product based on the user's behavior data; User portrait building module: analyzes the advertising product information and the user identity information through a machine learning algorithm combined with a deep learning algorithm, and builds user portraits of users of different genders and ages in the area based on the analysis results; Advertisement product recommendation module: establish a product recommendation model based on the user portrait, and place advertisement products based on the product recommendation model; Model optimization module: monitor the advertising effect in real time and continuously optimize the recommendation model based on the monitoring results.

6. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the memory, wherein the processor executes the program to implement the advertising delivery method based on user big data analysis as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement an advertising delivery method based on user big data analysis as described in any one of claims 1-4.

Images