Intelligent advertisement putting system and method based on user behavior portrait

Abstract

The invention relates to the technical field of advertisement putting, in particular to an intelligent advertisement putting system and method based on a user behavior portrait, and the system comprises a user portrait construction module, an advertisement weight distribution module, a user portrait adjustment module and an advertisement putting optimization module. The user portrait construction module divides the browsing data into current browsing data and historical browsing data, rejects commodity prices which do not conform to the tendency when the user portrait is constructed currently based on the commodity prices which the user tends to in the historical browsing data, and constructs the user portrait; the advertisement weight distribution module distributes different advertisement putting weights according to the number of times of browsing the commodities in the current browsing data for putting the advertisements, so that the advertisements can accurately reach the users who are really interested in the related commodities and accord with the price expectation of the related commodities. The number of advertisements which are not interested in or do not conform to the consumption price level to the user is reduced, and waste of advertisement resources is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of advertisement delivery, and in particular to an intelligent advertisement delivery system and method based on user behavior profiling. Background Art

[0002] Advertising has always been an effective means for businesses to promote their brands. With the increasing popularity of the Internet, online advertising has gradually replaced traditional advertising methods such as print, television and radio, and has become one of the most effective advertising methods. E-commerce platforms face challenges in advertising, as they have a vast amount of data resources. Users’ operations such as browsing product details, adding to shopping carts, completing purchases, and evaluating products all leave detailed data traces. However, since the platform stores a variety of browsing records, the data is intertwined and mixed, and users’ diverse browsing behavior data is aggregated together to form a complex data pool. Problems arise when advertising is delivered to users based on such large and mixed browsing data. If it is difficult to extract the current user's real shopping intention and core demand tendency from the browsing data, the ads delivered at this time often do not match the products that the user actually wants to buy at the moment, which ultimately leads to a significant reduction in the accuracy of advertising delivery. In view of this, we propose an intelligent advertising delivery system and method based on user behavior portraits. Summary of the invention

[0003] The purpose of the present invention is to solve the problem that it is not possible to accurately provide users with advertisements that meet their current shopping needs.

[0004] To achieve the above-mentioned purpose, the present invention provides an intelligent advertising delivery system based on user behavior portraits, including a user portrait construction module, an advertising weight allocation module, a user portrait adjustment module and an advertising delivery optimization module; The user portrait construction module is used to divide the user browsing data into current browsing data and historical browsing data, select the browsed products in the current browsing data and the product price range that the user prefers in the historical browsing data, and remove the products that do not meet the preferred product price range among the products browsed by the user. The pictures and prices of the products browsed by the user are the user portraits; The advertisement weight allocation module is used to calculate the first cosine similarity between the user portrait and the audience characteristics in the advertisement library, and set the cosine similarity threshold, filter out advertisements within the cosine similarity threshold, and allocate different weights according to the number of times the current user browses the product, and deliver the filtered advertisements to the user according to different weights. The audience characteristics in the advertisement library refer to the product images in the above-mentioned user portrait construction module; The user portrait adjustment module is used to sense the price and number of times the user clicks on the product in the delivered advertisement, record the price and number of times the product price in the clicked advertisement is greater than the price and number of times the user prefers the product price range in the user portrait construction module, and when the number of times the product in the clicked advertisement is greater than the normal number of times the product is clicked, calculate the average of the price of the product in the clicked advertisement, and if the average is greater than the price range of the user prefers the product, select the adjustment coefficient from small to large to increase the upper limit of the price range of the user prefers the product, and the lower limit of the price range of the user prefers the product is adjusted in the same way; The advertising delivery optimization module is used to sense the product image clicked by the user in the delivered advertisement, set an adjustment threshold, and calculate the second cosine similarity between the clicked product image and the product image in the historical browsing data through the advertising weight allocation module. If the second cosine similarity is greater than the adjustment threshold, it is determined that the clicked product is the same as the product image in the historical browsing data, and the number of product views in the historical browsing data is transferred to the current user browsing number in the advertising weight allocation module, thereby increasing the weight of the corresponding product in the advertising weight allocation module.

[0005] As a further improvement of the technical solution, the user portrait construction module perceives the user's purchase behavior, the time node when the user's purchase behavior occurs is the division node, the browsing data corresponding to the division node is the current browsing data, and the browsing data corresponding to the division node is the historical browsing data; As the user makes purchases again in the future, the new purchase behavior time node replaces the previous purchase behavior time node and becomes the new division node.

[0006] As a further improvement of the present technical solution, the commodity price range preferred by the user in the user portrait construction module is calculated by the quantile method, the corresponding price of each commodity purchased in the historical browsing data is retrieved, and the purchased commodity prices are sorted from large to small, and the purchased commodity price data set is divided into five numerical points of equal frequency parts, the second and fourth numerical points are selected from large to small, and the commodity prices corresponding to the second and fourth numerical points are calculated, and the amount between the commodity prices is the commodity price range preferred by the user.

[0007] As a further improvement of this technical solution, the advertisement weight allocation module allocates advertisement weights according to the feature vector of the user portrait. and the ad demographic vector , Pointing to the dimension of the quantity, calculate the first cosine similarity between two feature vectors; First cosine similarity ; in and The vectors are and The elements between Represents the first locations.

[0008] As a further improvement of the present technical solution, the advertisement weight allocation module sets the cosine similarity threshold based on industry experience. Different industries have accumulated certain experience data in long-term advertisement delivery practice.

[0009] As a further improvement of the technical solution, the advertisement weight allocation module allocates advertisements according to the number of times the current user browses the product. Since different browsing times reflect the user's interest in different products, advertisements related to products with high user interest are allocated higher weights. The specific calculation formula for different weights is as follows: ; in, The number of times users browse products. Display products Number of views, To distinguish different products. is the index variable used in summation The formula traverses all commodities.

[0010] As a further improvement of this technical solution, the user portrait adjustment module senses the price of the product in the advertisement clicked by the user. ,in is the number of clicks, and the formula for calculating the average price of the product in the clicked advertisement is: ,in, Indicates The price of the product per click; Perceive the price range of products preferred by users in the user portrait construction module ,in is the lower limit, is the lower limit, and the adjustment coefficient is ,and , When selecting, it is an equidistant selection, and the upper limit of the enlarged price range is: .

[0011] As a further improvement of the technical solution, the advertising delivery optimization module sets an adjustment threshold based on a large number of known identical product images, specifically through a deep learning model: the deep learning model divides the preprocessed image data into a training set, a validation set, and a test set. During the training process, the same images in the training set are input into the model. The deep learning model extracts the features of the images through a convolutional layer, gradually compresses the feature dimensions through a pooling layer, and outputs the prediction results for the same images through a fully connected layer; After the deep learning model training is completed, the same image pairs in the validation set are input into the deep learning model to obtain the similarity scores output by the deep learning model, and the mean of the similarity scores is calculated as the adjustment threshold.

[0012] As a further improvement of the technical solution, the advertising delivery optimization module uses image recognition technology to extract feature vectors in the product images before calculating the second cosine similarity of the product images; Image recognition technology maps the pixel values ​​of product images to the interval [0,1], resizes the product images to the same size, extracts feature vectors through convolutional neural networks, and slides the convolution kernel over the product images to perform convolution operations on local areas of the product images. The activation function is used to introduce nonlinear factors. The activation function enables the neural network to learn complex nonlinear relationships. The pooling layer downsamples the feature map to reduce the amount of data while retaining the main feature information. Through the combination of multiple convolutional layers, activation functions, and pooling layers, various features of product images from low-level to high-level are extracted.

[0013] The intelligent advertising delivery method based on user behavior profiling includes the following steps: Step 1: Divide current browsing data and historical browsing data to create user portraits; Step 2: Place ads that match the user profile. According to the number of times the user browses the product in the profile, different ad placement weights are assigned to place ads. Step 3: Adjust the user profile based on the price of the product the user clicks on the ad; Step 4: Determine whether the product image of the ad clicked by the user is the same as the product image in the historical browsing data. If they are the same, adjust the ad delivery weight.

[0014] Compared with the prior art, the present invention has the following beneficial effects: In the intelligent advertising delivery system and method based on user behavior portrait, the user portrait construction module divides browsing data into current browsing data and historical browsing data, constructs a user portrait based on the commodity prices preferred by the user in the historical browsing data and excluding the commodity prices that do not conform to the preference when constructing the current user portrait, and the advertising weight allocation module allocates different advertising delivery weights according to the number of times the commodity is browsed in the current browsing data for advertising delivery. After the advertisement is delivered, the user portrait adjustment module adjusts the commodity price preferred by the user in the user portrait construction module according to the price and number of times the user clicks on the corresponding commodity of the advertisement, and the advertising delivery optimization module analyzes whether the picture of the corresponding commodity clicked by the user is the same as the commodity picture in the historical browsing data, and if the same, adjusts the weight allocated in the advertising weight allocation module, so that the advertisement can accurately reach those users who are truly interested in the relevant commodities and meet their price expectations, reduce the number of advertisements pushed to users that they are not interested in or do not meet their consumption price level, and avoid waste of advertising resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is the overall module principle diagram of the present invention; Figure 2 It is a working step diagram of the present invention.

[0016] The meaning of each number in the figure is: 100. User portrait building module; 200. Advertising weight allocation module; 300. User portrait adjustment module; 400. Advertising delivery optimization module. DETAILED DESCRIPTION

[0017] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.

[0018] like Figure 1-Figure 2 , an intelligent advertising delivery system based on user behavior portrait, including a user portrait construction module 100, an advertising weight allocation module 200, a user portrait adjustment module 300 and an advertising delivery optimization module 400; In order to filter out browsing data that matches the current user from the user browsing data, the user portrait construction module 100 is used to divide the user browsing data into current browsing data and historical browsing data, select the browsed products in the current browsing data and the product price range that the user prefers in the historical browsing data, and remove the products that do not match the preferred product price range among the products browsed by the user. At this time, the pictures and prices of the products browsed by the user are the user portrait; Specifically, the user portrait construction module 100 senses the user's purchase behavior, the time node when the user's purchase behavior occurs is the division node, the browsing data corresponding to the division node is the current browsing data, and the browsing data corresponding to the division node is the historical browsing data; As the user makes purchases again in the future, the new purchase behavior time node replaces the previous purchase behavior time node and becomes the new division node.

[0019] Furthermore, the commodity price range that the user prefers in the user portrait construction module 100 is calculated by the quantile method, the corresponding price of each commodity purchased in the historical browsing data is retrieved, and the purchased commodity prices are sorted from large to small, and the purchased commodity price data set is divided into five equal frequency parts of numerical points, and the second and fourth numerical points are selected from large to small, and the commodity prices corresponding to the second and fourth numerical points are calculated, and the amount between the commodity prices is the commodity price range that the user prefers; For example, the purchase price data set is divided into For the five numerical points of equal frequency, the corresponding purchase commodity prices are selected from 25% to 75% respectively. The commodity purchase price range corresponding to 25% to 75% is the commodity price range preferred by users.

[0020] In order to accurately deliver advertisements to users based on the current user portrait, the advertisement weight allocation module 200 is used to calculate the first cosine similarity between the user portrait and the audience characteristics in the advertisement library, and set the cosine similarity threshold, filter out advertisements within the cosine similarity threshold, and assign different weights according to the number of times the current user browses the product, and deliver the filtered advertisements to the user according to different weights. The audience characteristics in the advertisement library refer to the product images in the above-mentioned user portrait construction module 100; Specifically, the advertisement weight allocation module 200 allocates advertisement weights based on the feature vector of the user portrait. and the ad demographic vector , The dimension of the pointing quantity, that is, the number of different feature dimensions contained in the user portrait or the advertising audience characteristics, calculates the first cosine similarity between the two feature vectors. The first cosine similarity value is between -1 and 1. The closer the value is to 1, the more similar the user portrait is to the advertising audience characteristics. First cosine similarity ; in and The vectors are and The elements between Represents the first locations.

[0021] Furthermore, the advertisement weight allocation module 200 sets the cosine similarity threshold based on industry experience. Different industries have accumulated certain experience data in long-term advertisement delivery practices. In the fashion industry, for the advertisement delivery of clothing brands, if you want to accurately reach the target users, you will set the cosine similarity threshold at between; Because the consumption of fashion products is closely related to factors such as the user's age, gender, and style preferences, a high matching degree is required to arouse the user's interest. In some daily necessities industries, due to the high versatility of products, the cosine similarity threshold is set relatively low, at Therefore, you can refer to the experience of advertising similar products in your industry to set the cosine similarity threshold.

[0022] The advertisement weight allocation module 200 allocates advertisements according to the number of times the current user browses a product. Different browsing times reflect the user's interest in different products. A higher weight is allocated to advertisements related to products with high user interest, so there are more opportunities to display advertisements to users. The specific calculation formula for different weights is as follows: ; in, The number of times users browse products. Display products Number of views, To distinguish different products. is the index variable used in summation The formula goes through all commodities; For example, if a user frequently browses a certain brand of mobile phone, it means that he or she has a strong interest in the phone; while for other electronic products that the user browses only occasionally, the user's interest is relatively low. By assigning weights based on the number of views, this interest difference can be more accurately reflected in advertising. By assigning higher weights to ads related to products that users are very interested in, there are more opportunities to show these ads to users, increasing the probability that users will be exposed to ads of interest.

[0023] The present invention takes into account the situation that the current user preferred commodity price will increase or decrease with the user's income. At this time, the user preferred commodity price range set in the user portrait construction module 100 will be inaccurate. Therefore, the user portrait adjustment module 300 senses the price and number of times the user clicks on the commodity in the delivered advertisement, and records the price of the commodity in the clicked advertisement > the price and number of times the user preferred commodity price range in the user portrait construction module 100. When the number of times the commodity in the advertisement is clicked > the normal number of times the commodity is clicked, the average of the price of the commodity in the clicked advertisement is calculated. If the average value > the user preferred commodity price range, the adjustment coefficient is selected from small to large to increase the upper limit of the user preferred commodity price range. After increasing the upper limit of the price range of the user's preferred product, if the number of times the advertised product is clicked is greater than the number of times the advertised product is clicked before the increase, the upper limit of the price range of the user's preferred product is increased again, otherwise the upper limit of the price range of the user's preferred product is stopped; Similarly, when the mean is less than the price range of the products preferred by the users, the lower limit of the price range of the products preferred by the users can be lowered by the above method; Specifically, the user portrait adjustment module 300 senses the price of the product in the advertisement clicked by the user. ,in is the number of clicks, and the formula for calculating the average price of the product in the clicked advertisement is: ,in, Indicates The price of the product per click; Perceive the user profile building module 100 to determine the price range of the product that the user prefers ,in is the lower limit, is the lower limit, and the adjustment coefficient is ,and , When selecting, it is an equidistant selection, and the upper limit of the enlarged price range is: .

[0024] If the prices of the products clicked by the user are mostly higher than the upper limit of the original price range, it means that the user may have a new interest in higher-priced products. The user portrait adjustment module 300 will accordingly appropriately expand the upper limit of the price range, or move the price range upward as a whole. Conversely, if the clicked prices are mostly lower than the lower limit of the original range, it may mean that the user's price preference has decreased, and the user portrait adjustment module 300 will lower the price range.

[0025] The advertisement delivery optimization module 400 is used to sense the product image clicked by the user in the delivered advertisement, set an adjustment threshold, and calculate the second cosine similarity between the clicked product image and the product image in the historical browsing data through the advertisement weight allocation module 200. If the second cosine similarity is greater than the adjustment threshold, it is determined that the clicked product is the same as the product image in the historical browsing data, and the product browsing times in the historical browsing data are transferred to the current user browsing times in the advertisement weight allocation module 200, and the weight of the corresponding product in the advertisement weight allocation module 200 is increased. From the perspective of user cognition and memory, people tend to pay more attention and attention to familiar and frequently contacted things. When a product frequently appears in the user's browsing history and is clicked again, it means that the product has left a deep impression in the user's memory, and its attractiveness has withstood the test of time and multiple browsing comparisons; The method of integrating the historical browsing times and adjusting the weights in the advertisement delivery optimization module 400 conforms to the cognitive and memory rules of users, strengthens the display of the products that are deeply remembered by these users in advertisement delivery, conforms to the user's subconscious tendency to pay attention to the things that interest them, is more likely to resonate with and actively respond to users, and improves the influence and acceptance of advertisements in the minds of users; Specifically, the advertisement delivery optimization module 400 sets an adjustment threshold based on a large number of known identical product images, specifically through a deep learning model, the deep learning model divides the pre-processed image data into a training set, a validation set, and a test set. During the training process, the same images in the training set are input into the model, and the deep learning model extracts the features of the images through a convolutional layer, gradually compresses the feature dimensions through a pooling layer, and outputs the prediction results for the same images through a fully connected layer; After the deep learning model training is completed, the same image pairs in the validation set are input into the deep learning model to obtain the similarity scores output by the deep learning model, and the mean of the similarity scores is calculated as the adjustment threshold.

[0026] Furthermore, before calculating the second cosine similarity of the product images, the advertisement placement optimization module 400 uses image recognition technology to extract feature vectors from the product images; First, the pixel values ​​of the product images are mapped to the interval [0,1], and the product images are resized to the same size. The feature vectors are extracted through the convolutional neural network. The convolution layer slides the convolution kernel on the product image and performs convolution operations on the local area of ​​the product image. The activation function is used to introduce nonlinear factors. The role of the activation function is to enable the neural network to learn complex nonlinear relationships. The pooling layer downsamples the feature map to reduce the amount of data while retaining the main feature information. Through the combination of multiple convolutional layers, activation functions, and pooling layers, various features of product images from low-level to high-level are extracted.

[0027] The intelligent advertising delivery method based on user behavior profiling includes the following steps: Step 1: Divide current browsing data and historical browsing data to create user portraits; Step 2: Place ads that match the user profile. According to the number of times the user browses the product in the profile, different ad placement weights are assigned to place ads. Step 3: Adjust the user profile based on the price of the product the user clicks on the ad; Step 4: Determine whether the product image of the ad clicked by the user is the same as the product image in the historical browsing data. If they are the same, adjust the ad delivery weight.

[0028] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent advertising delivery system based on user behavior portraits, characterized in that: It includes a user portrait building module (100), an advertisement weight allocation module (200), a user portrait adjustment module (300) and an advertisement delivery optimization module (400); The user portrait construction module (100) is used to divide the user browsing data into current browsing data and historical browsing data, select the browsed products in the current browsing data and the product price range preferred by the user in the historical browsing data, and remove the products that do not meet the preferred product price range among the products browsed by the user. The pictures and prices of the products browsed by the user are the user portrait; The advertisement weight allocation module (200) is used to calculate the first cosine similarity between the user portrait and the audience characteristics in the advertisement library, and set the cosine similarity threshold, filter out advertisements within the cosine similarity threshold, and allocate different weights according to the number of times the current user browses the product, and deliver the filtered advertisements to the user according to the different weights. The audience characteristics in the advertisement library refer to the product images in the above-mentioned user portrait construction module (100); The user portrait adjustment module (300) is used to sense the price and number of times of the product clicked by the user in the placed advertisement, record the price and number of times when the price of the product in the clicked advertisement is greater than the price and number of times of the user's preferred product price range in the user portrait construction module (100), and when the number of times of clicking the product in the advertisement is greater than the normal number of times of clicking the product, calculate the average of the price of the product in the clicked advertisement, and if the average is greater than the user's preferred product price range, select an adjustment coefficient from small to large to increase the upper limit of the user's preferred product price range, and the lower limit of the user's preferred product price range is adjusted in the same way; The advertising delivery optimization module (400) is used to sense the product image clicked by the user in the delivered advertisement, set an adjustment threshold, and calculate the second cosine similarity between the clicked product image and the product image in the historical browsing data through the advertising weight allocation module (200). If the second cosine similarity is greater than the adjustment threshold, it is determined that the clicked product is the same as the product image in the historical browsing data, and the product browsing times in the historical browsing data are transferred to the current user browsing times in the advertising weight allocation module (200), thereby increasing the weight of the corresponding product in the advertising weight allocation module (200).

2. The intelligent advertising delivery system based on user behavior profiling according to claim 1 is characterized in that: The user portrait construction module (100) senses the user's purchase behavior, the time node when the user's purchase behavior occurs is the division node, the browsing data corresponding to the division node is the current browsing data, and the browsing data corresponding to the division node is the historical browsing data; As the user makes purchases again in the future, the new purchase behavior time node replaces the previous purchase behavior time node and becomes the new division node.

3. The intelligent advertising delivery system based on user behavior profiling according to claim 2 is characterized in that: The user profile building module (100) calculates the commodity price range that the user prefers by using the quantile method, retrieves the corresponding price of each commodity purchased in the historical browsing data, and sorts the purchased commodity prices from large to small, divides the purchased commodity price data set into five numerical points of equal frequency, selects the second and fourth numerical points from large to small, calculates the commodity prices corresponding to the second and fourth numerical points, and the amount between the commodity prices is the commodity price range that the user prefers.

4. The intelligent advertising delivery system based on user behavior profiling according to claim 1 is characterized in that: The advertisement weight allocation module (200) allocates advertisement weights based on the feature vector of the user portrait. and the ad demographic vector , Pointing to the dimension of the quantity, calculate the first cosine similarity between two feature vectors; First cosine similarity ; in and The vectors are and The elements between Represents the first locations.

5. The intelligent advertising delivery system based on user behavior profiling according to claim 4 is characterized in that: The advertisement weight allocation module (200) sets the cosine similarity threshold based on industry experience. Different industries have accumulated certain experience data in long-term advertisement placement practice.

6. The intelligent advertising delivery system based on user behavior profiling according to claim 2 is characterized in that: The advertisement weight allocation module (200) allocates advertisements according to the number of times the current user browses a product. Since different numbers of browses reflect the user's interest in different products, advertisements related to products with high user interest are allocated higher weights. The specific calculation formula for different weights is as follows: ； in, The number of times users browse products. Display products Number of views, To distinguish different products. is the index variable used in summation The formula traverses all commodities.

7. The intelligent advertising delivery system based on user behavior profiling according to claim 1 is characterized by: The user portrait adjustment module (300) senses the price of the product in the advertisement clicked by the user ,in is the number of clicks, and the formula for calculating the average price of goods in click ads is: ,in, Indicates The price of the product per click; Perceiving the price range of commodities preferred by users in the user portrait building module (100) ,in is the lower limit, is the lower limit, and the adjustment coefficient is ,and , When selecting, it is an equidistant selection, and the upper limit of the enlarged price range is: .

8. The intelligent advertising delivery system based on user behavior profiling according to claim 7 is characterized in that: The advertising placement optimization module (400) sets an adjustment threshold based on a large number of known identical product images, specifically through a deep learning model: the deep learning model divides the pre-processed image data into a training set, a validation set, and a test set. During the training process, the same images in the training set are input into the model. The deep learning model extracts features of the images through a convolutional layer, gradually compresses the feature dimensions through a pooling layer, and outputs prediction results for the same images through a fully connected layer; After the deep learning model training is completed, the same image pairs in the validation set are input into the deep learning model to obtain the similarity scores output by the deep learning model, and the mean of the similarity scores is calculated as the adjustment threshold.

9. The intelligent advertising delivery system based on user behavior profiling according to claim 8 is characterized in that: The advertising placement optimization module (400) uses image recognition technology to extract feature vectors from the product images before calculating the second cosine similarity of the product images; Image recognition technology maps the pixel values ​​of product images to the interval [0,1], resizes the product images to the same size, extracts feature vectors through convolutional neural networks, and slides the convolution kernel over the product images to perform convolution operations on local areas of the product images. The activation function is used to introduce nonlinear factors. The activation function enables the neural network to learn complex nonlinear relationships. The pooling layer downsamples the feature map to reduce the amount of data while retaining the main feature information. Through the combination of multiple convolutional layers, activation functions, and pooling layers, various features of product images from low-level to high-level are extracted.

10. The method for intelligent advertising delivery based on user behavior profiling is applied to the intelligent advertising delivery system based on user behavior profiling as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Divide current browsing data and historical browsing data to create user portraits; Step 2: Place ads that match the user profile. According to the number of times the user browses the product in the profile, different ad placement weights are assigned to place ads. Step 3: Adjust the user profile based on the price of the product the user clicks on the ad; Step 4: Determine whether the product image of the ad clicked by the user is the same as the product image in the historical browsing data. If they are the same, adjust the ad delivery weight.

Images