Advertisement recommendation method, system and device and storage medium

Abstract

The invention discloses an advertisement recommendation method, system and device and a storage medium, and the method comprises the steps: obtaining user portrait data of a current access user, the user portrait data comprising real-time behavior data and historical behavior data of the access user, and the historical behavior data comprising advertisement data of an advertisement displayed to the access user; determining at least one candidate advertisement data matched with the access user based on the user portrait data; inputting the candidate advertisement data and the user portrait data into a pre-trained advertisement putting pre-estimation model, and obtaining putting effect pre-estimation data corresponding to the candidate advertisement data output by the advertisement putting pre-estimation model; determining an optimal bid corresponding to each piece of candidate advertisement data based on the delivery effect estimation data corresponding to each piece of candidate advertisement data; and determining the optimal advertisement data according to the optimal bid corresponding to each candidate advertisement data, and displaying the optimal advertisement data to the access user, thereby improving the flexibility and accuracy of advertisement putting.

Description

Technical Field

[0001] The present invention relates to the technical field of advertisement delivery, and in particular to an advertisement recommendation method, system, device and storage medium. Background Art

[0002] With the rapid development of the Internet and mobile Internet, the network has become a part of people's lives. Through various types of clients, you can easily access the Internet at any time and obtain various information through the Internet. As an important information medium, the Internet has a large amount of advertising information online every day. Various types of advertisements are delivered through various Internet advertising platforms.

[0003] In the related art, advertisers purchase advertisements for fixed location display, and the advertisements will be displayed regardless of whether the user is interested or not. The advertisement display mechanism is simple and lacks flexibility and targeting, resulting in low advertisement click-through rate and conversion rate. Summary of the invention

[0004] In view of this, the present invention provides an advertisement recommendation method, system, device and storage medium to improve the pertinence and accuracy of advertisement recommendation, thereby improving the advertisement recommendation effect.

[0005] A first aspect of the present invention provides an advertisement recommendation method, comprising:

[0006] Obtaining user portrait data of the current visiting user, wherein the user portrait data includes real-time behavior data and historical behavior data of the visiting user, wherein the historical behavior data includes advertisement data of advertisements displayed to the visiting user;

[0007] Based on the user portrait data, determining at least one candidate advertisement data matching the visiting user;

[0008] Inputting each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model respectively, and obtaining delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model;

[0009] Determine the optimal bid corresponding to each of the candidate advertisement data based on the estimated data of delivery effect corresponding to each of the candidate advertisement data;

[0010] The optimal advertisement data is determined according to the optimal bids corresponding to the candidate advertisement data, and the optimal advertisement data is displayed to the visiting user.

[0011] A second aspect of the present invention provides an advertisement recommendation system, comprising:

[0012] A user portrait acquisition module, used to acquire user portrait data of the current visiting user, wherein the user portrait data includes real-time behavior data and historical behavior data of the visiting user, wherein the historical behavior data includes advertisement data of advertisements displayed to the visiting user;

[0013] A candidate advertisement determination module, configured to determine at least one candidate advertisement data matching the visiting user based on the user portrait data;

[0014] The delivery effect estimation module is used to input each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtain delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model;

[0015] An optimal bid determination module, used to determine the optimal bid corresponding to each candidate advertisement data based on the estimated delivery effect data corresponding to each candidate advertisement data;

[0016] The optimal advertisement determination module is used to determine the optimal advertisement data according to the optimal bid corresponding to each of the candidate advertisement data, and display the optimal advertisement data to the visiting user.

[0017] A third aspect of the present invention provides an electronic device, the electronic device comprising:

[0018] at least one processor; and

[0019] a memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the advertisement recommendation method as described in the first aspect above.

[0021] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the advertisement recommendation method as described in the first aspect is implemented.

[0022] A fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program, and the computer program, when executed by a processor, implements the advertisement recommendation method as described in the first aspect above.

[0023] In this embodiment, by obtaining the user portrait data such as the real-time behavior data, historical behavior data, and advertisement data of historically clicked advertisements of the current visiting user, at least one candidate advertisement data is matched for the current visiting user. And through the pre-trained advertisement delivery estimation model, the delivery effect estimation data corresponding to each candidate advertisement data of the current visiting user is estimated, and then based on the delivery effect estimation data corresponding to each candidate advertisement data, the optimal bid corresponding to each candidate advertisement data is determined, so as to determine the optimal advertisement data through bidding processing according to the optimal bid, and display the optimal advertisement data to the current visiting user. The advertisement displayed to the current user is matched with the real-time behavior and historical behavior of the current user. When the same user visits the same platform twice, the advertisement recommended to him may be different due to the real-time changes in the user portrait data, which improves the flexibility and accuracy of advertisement delivery, and improves the effect of advertisement delivery and the return on investment of advertisers.

[0024] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 The present invention is a flowchart of an advertisement recommendation method provided by an embodiment of the present invention.

[0027] Figure 2 It is a flowchart of a training method for an advertisement delivery prediction model according to an embodiment of the present invention.

[0028] Figure 3 It is a flow chart of an advertisement recommendation method provided by another embodiment of the present invention.

[0029] Figure 4 This is a flow chart of an advertisement recommendation method provided by yet another embodiment of the present invention.

[0030] Figure 5 It is a structural diagram of an advertisement recommendation system provided by an embodiment of the present invention.

[0031] Figure 6 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can cover sequential implementations other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] The traditional Internet advertising ecosystem generally includes three parties, namely advertisers, advertising agencies (i.e. advertising companies) and Internet media. In the RTB (Real Time Bidding) advertising transaction model, the original advertising ecosystem has changed. The entire ecosystem includes six entities: advertisers, demand-side platforms (DSP), advertising exchanges (Ad Exchange), data management platforms (DMP), supply-side platforms (SSP) and consumers. Advertisers put their advertising needs on the DSP platform, SSP puts its advertising traffic resources on the advertising exchange platform, and DSP completes bidding purchases through technical docking with the advertising exchange platform.

[0035] Therefore, when a user visits a website or application (Application, APP), SSP sends a user access signal to the AdExchange advertising trading platform, and Ad Exchange sends the specific information of the advertising space to DSP. DSP bids for it, and the highest bidder will get the opportunity to display the advertisement. After the bidding is successful, DSP uploads the advertising code to the AdExchange advertising trading platform, and the Ad Exchange advertising trading platform will deliver the advertising code to the web page or APP browsed by the user to realize advertising delivery.

[0036] This technical solution can be applied to the interaction scenario between SSP and Ad Exchange. Media parties (for example, website owners, game platforms, etc.) can manage their own advertising spaces on SSP and control the display of advertisements. SSP enables media parties to participate in advertising transactions, thereby making the media party's inventory ads available. Media parties hope that through SSP, inventory ads can obtain the highest effective cost per thousand impressions without having to sell them at a low price. Terminals connected to SSP can be used as terminals for consumers.

[0037] See also Figure 1 , shows a flow chart of an advertisement recommendation method provided by an embodiment of the present invention, the method may include the following steps:

[0038] Step 101, obtaining user portrait data of the current access user.

[0039] The current access user refers to the consumer on the SSP platform. In the SSP platform, a portrait database can be created to maintain the user portrait data of each consumer, and the user portrait data can be updated in real time by monitoring the real-time behavior of each consumer. Therefore, the user portrait data is dynamic portrait data. When a user accesses the SSP platform, the SSP platform can search for the corresponding user portrait data in the portrait database based on the user ID of the current access user.

[0040] User portrait data (Persona) refers to the construction of representative and typical user roles based on an in-depth understanding of the user group through research, interviews, data analysis and other means. User portrait data usually includes multiple dimensions such as the user's basic attributes, behavioral habits, psychological characteristics, and usage scenarios, aiming to help companies and product designers more accurately grasp user needs, optimize product design and improve user experience. In an example of this embodiment, the user portrait data at least includes real-time behavior data (such as context information, etc.) and historical behavior data (such as historical browsing records, historical purchase records, etc.) of the accessing user.

[0041] The historical behavior data may include advertisement data of advertisements displayed to visiting users, such as advertisement type, advertisement content, advertisement click-through rate, advertisement conversion rate, etc.

[0042] This embodiment does not limit the method of constructing user portrait data, and any one or more existing methods can be used to construct user portrait data for each consumer. However, it should be noted that in the process of constructing user portrait data, the collected user personal information is obtained after the user's authorization, and the collected information meets the requirements of legality and compliance.

[0043] Step 102: Based on the user portrait data, determine at least one candidate advertisement data that matches the visiting user.

[0044] In this step, after obtaining the user portrait data of the current visiting user, advertisement matching can be performed based on the user portrait data, thereby determining at least one candidate advertisement data that matches the current visiting user. In one implementation, the SSP platform can send the user portrait data to the Ad Exchange platform, and the Ad Exchange platform matches at least one candidate advertisement data corresponding to the current user portrait data from the advertisement pool. This embodiment does not limit the specific matching method. For example, an advertisement matching model or advertisement matching algorithm can be set in advance in the Ad Exchange platform. The Ad Exchange platform can extract corresponding input features from the user portrait data based on the input item requirements corresponding to the advertisement matching model or advertisement matching algorithm, and calculate the similarity between the input features and the advertisement features of each advertisement in the advertisement pool, and use several advertisement data with the highest similarity as candidate advertisement data.

[0045] Step 103, respectively input each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtain delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model.

[0046] After the Ad Exchange platform determines the candidate advertising data, it can use the pre-established advertising delivery estimation model to estimate the delivery effect estimation data corresponding to each candidate advertising data. As an example, the delivery effect estimation data may at least include CTR (Click-Through Rate) and CVR (Conversion Rate) data. CTR is an important indicator for measuring advertising effectiveness. It represents the ratio of the number of times an ad is clicked after being displayed to the total number of times the ad is displayed. In simple terms, CTR is the ratio of the number of times an ad is clicked to the number of times the ad is displayed. CVR is a key indicator for measuring advertising effectiveness. It represents the ratio of the number of conversions generated after an ad is clicked to the total number of times the ad is clicked. In simple terms, CVR is the ratio of conversions generated after an ad is clicked.

[0047] In one embodiment, step 103 may further include the following steps:

[0048] Step 103 - 1 , extracting designated user data from the user portrait data, and converting the designated user data into a user feature vector.

[0049] Step 103 - 2 , extracting designated advertisement data from the candidate advertisement data, and converting the designated advertisement data into an advertisement feature vector.

[0050] Step 103-3: for each candidate advertisement data, the corresponding advertisement feature vector and the user feature vector are input into the advertisement delivery estimation model, and the advertisement delivery estimation model estimates the delivery effect estimation data of the current visiting user on the current candidate advertisement data.

[0051] A feature vector is an important concept in the field of machine learning and data mining. It is a vector composed of a set of features, used to represent data instances or samples. These features can be different types of data such as numerical, categorical, and sequential data. In this embodiment, the feature vector includes a user feature vector for representing user features, and an advertisement feature vector for representing advertisement features.

[0052] The specified user data can be extracted from the user portrait data and the specified advertisement data can be extracted from the candidate advertisement data by using a pre-set extraction rule. The extraction rule is a custom rule that matches the recommendation purpose of this embodiment. For example, the specified user data may include user preferences, user location, user real-time behavior data, user historical behavior data, advertisement data of advertisements displayed to the current visiting user, etc. The specified advertisement data may include information such as the user's click-through rate, conversion rate, advertisement content, and advertisement type of the advertisement.

[0053] This embodiment does not limit the method for generating feature vectors and the number of dimensions of feature vectors. Manual encoding or autoencoder methods can be used to convert specified user data into user feature vectors, and to convert specified advertisement data into advertisement feature vectors.

[0054] For each candidate advertisement data, after obtaining its corresponding advertisement feature vector, the advertisement feature vector and the user feature vector can be input as input data into the pre-generated advertisement delivery estimation model. In the advertisement delivery estimation model, by processing the advertisement feature vector and the user feature vector, the delivery effect estimation data of the current visiting user on the current candidate advertisement data can be estimated.

[0055] In one example, the advertising delivery prediction model may include a linear regression model or other machine learning models.

[0056] Step 104: Determine the optimal bid corresponding to each of the candidate advertisement data based on the estimated data of delivery effect corresponding to each of the candidate advertisement data.

[0057] After obtaining the delivery effect estimation data corresponding to each candidate advertisement data, the Ad Exchange platform can further estimate the optimal bid of the candidate advertisement data according to the delivery effect estimation data.

[0058] In one implementation, step 104 may further include the following steps:

[0059] A preset bidding algorithm is used to determine the optimal bid corresponding to the estimated delivery effect data.

[0060] The present embodiment does not limit the bidding algorithm, and it may be any one or more existing bidding algorithms. In one implementation, if multiple bidding algorithms are used to calculate the bid, the average of the bids obtained by the multiple bidding algorithms is used as the optimal bid.

[0061] In other embodiments, the bidding algorithm may also be an algorithm provided by a bidding engine. Specifically, effect data such as CVR and CTR may be input into the bidding engine, and the bidding engine calculates the optimal bid corresponding to each candidate advertising data.

[0062] For example, the bidding algorithm can be expressed as follows:

[0063] Optimal bid = (CVR / CTR)*set unit price / (set return on investment + 1)

[0064] Among them, ROI (Return on Investment) refers to the value that should be returned through investment, that is, the economic return obtained from an investment activity. In layman's terms, it is the ratio of the income obtained to the investment cost.

[0065] Step 105: determine optimal advertisement data according to the optimal bids corresponding to the candidate advertisement data, and display the optimal advertisement data to the visiting user.

[0066] In one embodiment, after obtaining the optimal bids corresponding to the candidate advertisement data, the optimal bids may be subjected to bidding, and the optimal advertisement may be determined according to the bidding results. For example, the candidate advertisement data corresponding to the highest bid among the optimal bids may be used as the optimal advertisement data. The optimal advertisement data is then displayed to the current visiting user.

[0067] In this embodiment, by obtaining the user portrait data such as the real-time behavior data, historical behavior data, and advertisement data of historically clicked advertisements of the current visiting user, at least one candidate advertisement data is matched for the current visiting user. And through the pre-trained advertisement delivery estimation model, the delivery effect estimation data corresponding to each candidate advertisement data of the current visiting user is estimated, and then based on the delivery effect estimation data corresponding to each candidate advertisement data, the optimal bid corresponding to each candidate advertisement data is determined, so as to determine the optimal advertisement data through bidding processing according to the optimal bid, and display the optimal advertisement data to the current visiting user. The advertisement displayed to the current user is matched with the real-time behavior and historical behavior of the current user. When the same user visits the same platform twice, the advertisement recommended to him may be different due to the real-time changes in the user portrait data, which improves the flexibility and accuracy of advertisement delivery, and improves the effect of advertisement delivery and the return on investment of advertisers.

[0068] refer to Figure 2 , shows a flow chart of a method for training an advertisement delivery prediction model according to an embodiment of the present invention, the method may include the following steps:

[0069] Step S1: Obtain sample user portrait data of each sample user.

[0070] Among them, the sample user portrait data includes the real-time behavior data and historical behavior data of the sample users. The historical behavior data further includes advertising data of advertisements displayed to the sample users, such as advertising click behavior, conversion behavior, advertising type, advertising content, etc.

[0071] In one embodiment, each consumer user in the portrait database may be used as a sample user, and real-time updated sample user portrait data of each sample user is obtained from the portrait database.

[0072] Step S2: construct a sample feature vector based on the sample user portrait data.

[0073] The sample feature vector may include user attribute features of each sample user and advertisement attribute features of the advertisement displayed to the sample user, such as advertisement type, advertisement click-through rate, and advertisement conversion rate. The user attribute features may include, for example, the user's age range, gender, occupation, consumption habits, preferences, etc.

[0074] In one implementation, the specified user data can be extracted from the user portrait data as user attribute data, and the user attribute data can be converted into user attribute features. Then, the specified advertisement data can be extracted from the advertisement displayed for the user to construct the advertisement attribute features. Then, the user attribute features and the advertisement attribute features are combined into a sample feature vector.

[0075] Step S3: Use the sample feature vector to train the original logistic regression model to generate an advertising delivery prediction model.

[0076] In this embodiment, the advertisement delivery prediction model is a logistic regression model. When training the advertisement delivery prediction model, a feature matrix is ​​constructed using sample feature vectors, and the feature matrix is ​​used to train the original logistic regression model to generate the advertisement delivery prediction model.

[0077] In one implementation, step S3 may further include the following steps:

[0078] Step S31, performing feature engineering and feature preprocessing on the sample feature vector to generate a feature set, and dividing the feature set into a training set and a test set.

[0079] In one implementation, the feature preprocessing may include but is not limited to the following preprocessing operations: normalization, standardization, missing value processing, etc.

[0080] In one implementation, when the feature set is divided into a training set and a test set, the splitting may be performed at a ratio of 8:2. Furthermore, the feature set within a period of time (eg, the past 7 days) may be split into a training set and a test set at a ratio of 8:2.

[0081] Step S32, establishing a linear regression model as the original logistic regression model, and transferring the training set into the original logistic regression model for training.

[0082] In one implementation, a linear regression model can be built using python as the original logistic regression model.

[0083] Step S33, using the test set to predict the trained logistic regression model, and calculating the deviation between the predicted value and the true value.

[0084] In the implementation, when the test set is input into the logistic regression model for prediction, the mean square error can be calculated based on the predicted result (i.e., the predicted value), and a line chart can be used to compare the deviation between the true value and the predicted value.

[0085] Step S34, optimizing the training set based on the deviation until the deviation is within a preset range and the training is terminated to obtain an advertising delivery prediction model.

[0086] During implementation, the training features can be continuously fine-tuned and optimized through grid search to obtain a logistic regression model within a relatively ideal error range as an advertising delivery estimation model.

[0087] In this embodiment, the sample data used to train the advertising delivery prediction model is real-time user portrait data, so that the delivery effect prediction data output by the advertising delivery prediction model is closer to the user's real-time behavior, thereby improving the accuracy of the delivery effect prediction data.

[0088] See also Figure 3 , shows a flow chart of an advertisement recommendation method provided by another embodiment of the present invention, the method may include the following steps:

[0089] Step 201, obtaining user portrait data of the current access user.

[0090] Among them, the user portrait data includes the real-time behavior data and historical behavior data of the visiting user, and the historical behavior data includes the advertising data of the advertisements displayed to the visiting user.

[0091] Step 202: Based on the user portrait data, determine at least one candidate advertisement data that matches the visiting user.

[0092] Step 203, respectively input each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtain delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model.

[0093] Step 204: Determine the optimal bid corresponding to each of the candidate advertisement data based on the estimated data of delivery effect corresponding to each of the candidate advertisement data.

[0094] Step 205: determine the optimal advertisement data according to the optimal bids corresponding to the candidate advertisement data, and display the optimal advertisement data to the visiting user.

[0095] Step 206, monitoring the delivery effect of the optimal advertisement.

[0096] For example, the delivery effect may include whether the user clicks on the ad, whether the ad is converted, etc.

[0097] Step 207: Optimize the advertisement delivery estimation model based on the delivery effect.

[0098] In this embodiment, the advertisement delivery prediction model is optimized in real time through user feedback on advertisements, thereby further improving the accuracy of delivery effect prediction data.

[0099] See also Figure 4 , shows a flowchart of an advertisement recommendation method provided by another embodiment of the present invention, the method may include the following steps:

[0100] Step 301: Acquire historical behavior data of each user, wherein the historical behavior data includes advertisement data of advertisements displayed to the user.

[0101] Step 302: construct user portrait data based on the historical behavior data.

[0102] In one implementation, a portrait database may be constructed to maintain user portrait data of each user.

[0103] Step 303, obtain real-time behavior data by monitoring the real-time behavior of each user, and update the user portrait data with the real-time behavior data.

[0104] During implementation, the real-time computing technology of big data flink can be used to profile and mark the real-time data of user behavior and update it dynamically.

[0105] Step 304, obtaining user portrait data of the current accessing user.

[0106] Among them, the user portrait data includes the real-time behavior data and historical behavior data of the visiting user, and the historical behavior data includes the advertising data of the advertisements displayed to the visiting user.

[0107] Step 305: Based on the user portrait data, determine at least one candidate advertisement data that matches the visiting user.

[0108] Step 306, respectively input each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtain delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model.

[0109] Step 307: Determine the optimal bid corresponding to each of the candidate advertisement data based on the estimated data of delivery effect corresponding to each of the candidate advertisement data.

[0110] Step 308: determine the optimal advertisement data according to the optimal bids corresponding to the candidate advertisement data, and display the optimal advertisement data to the visiting user.

[0111] In this embodiment, the user portrait data can be dynamically updated through real-time user behavior, so that the candidate advertising data obtained based on the user portrait data is more in line with the user's latest behavior, further improving the pertinence and accuracy of advertising recommendations.

[0112] See also Figure 5 , shows a schematic diagram of the structure of an advertisement recommendation system provided by an embodiment of the present invention. Figure 5 As shown, the device comprises:

[0113] A user portrait acquisition module 401 is used to acquire user portrait data of the current visiting user, wherein the user portrait data includes real-time behavior data and historical behavior data of the visiting user, wherein the historical behavior data includes advertisement data of advertisements displayed to the visiting user;

[0114] A candidate advertisement determination module 402, configured to determine at least one candidate advertisement data matching the accessing user based on the user portrait data;

[0115] The delivery effect estimation module 403 is used to input each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtain delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model;

[0116] The optimal bid determination module 404 is used to determine the optimal bid corresponding to each candidate advertisement data based on the estimated delivery effect data corresponding to each candidate advertisement data;

[0117] The optimal advertisement determination module 405 is used to determine the optimal advertisement data according to the optimal bids corresponding to the candidate advertisement data, and display the optimal advertisement data to the visiting user.

[0118] In one embodiment, the delivery effect estimation module 403 is specifically used to:

[0119] Extracting designated user data from the user portrait data, and converting the designated user data into a user feature vector;

[0120] Extracting designated advertisement data from the candidate advertisement data, and converting the designated advertisement data into an advertisement feature vector;

[0121] For each candidate advertisement data, the corresponding advertisement feature vector and the user feature vector are input into the advertisement delivery estimation model, and the advertisement delivery estimation model estimates the delivery effect estimation data of the current visiting user on the current candidate advertisement data.

[0122] In one embodiment, the optimal bid determination module 404 is specifically configured to:

[0123] A preset bidding algorithm is used to determine the optimal bid corresponding to the estimated delivery effect data.

[0124] In one embodiment, the system further includes a model training module for generating an advertisement delivery estimation model, and the model training module further includes the following modules:

[0125] A sample portrait acquisition module is used to acquire sample user portrait data of each sample user, wherein the sample user portrait data includes real-time behavior data and historical behavior data of the sample user, wherein the historical behavior data includes advertisement data of advertisements displayed to the sample user;

[0126] A sample feature vector construction module is used to construct a sample feature vector according to the sample user portrait data, wherein the sample feature vector includes user attribute characteristics of each sample user and advertisement attribute characteristics of the advertisement displayed for the sample user, wherein the advertisement attribute characteristics include advertisement type, advertisement click-through rate and advertisement conversion rate;

[0127] The training module is used to train the original logistic regression model using the sample feature vector to generate an advertising delivery prediction model.

[0128] In one embodiment, the training module is specifically used to:

[0129] Performing feature engineering and feature preprocessing on the sample feature vector to generate a feature set, and dividing the feature set into a training set and a test set;

[0130] Establishing a linear regression model as an original logistic regression model, and transferring the training set into the original logistic regression model for training;

[0131] The test set is used to predict the trained logistic regression model, and the deviation between the predicted value and the true value is calculated;

[0132] The training set is optimized based on the deviation, and the training is terminated until the deviation is within a preset range, thereby obtaining an advertisement delivery prediction model.

[0133] In one embodiment, the system further comprises:

[0134] A delivery effect monitoring module, used for monitoring the delivery effect of the optimal advertisement after displaying the optimal advertisement data to the visiting user;

[0135] A model optimization module is used to optimize the advertisement delivery estimation model based on the delivery effect.

[0136] In one embodiment, the system further comprises:

[0137] A historical behavior data acquisition module, used to acquire historical behavior data of each user before acquiring the user portrait data of the current visiting user, wherein the historical behavior data includes advertisement data of advertisements displayed to the user;

[0138] A portrait construction module, used to construct user portrait data based on the historical behavior data;

[0139] A real-time behavior data acquisition module is used to obtain real-time behavior data by monitoring the real-time behavior of each user;

[0140] A portrait updating module is used to update the user portrait data with the real-time behavior data.

[0141] The advertisement recommendation provided by the embodiment of the present invention can execute the advertisement recommendation method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the advertisement recommendation method.

[0142] See also Figure 6 , shows a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, blade servers, mainframe computers, and other suitable computers. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0143] like Figure 6 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0144] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0145] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the advertisement recommendation method.

[0146] In some embodiments, the advertising recommendation method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the advertising recommendation method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the advertising recommendation method in any other appropriate manner (e.g., by means of firmware).

[0147] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0148] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0149] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0150] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0151] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0152] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0153] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the advertisement recommendation method provided by any embodiment of the present invention is implemented.

[0154] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0155] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0156] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An advertisement recommendation method, characterized in that: include: Obtaining user portrait data of the current visiting user, wherein the user portrait data includes real-time behavior data and historical behavior data of the visiting user, wherein the historical behavior data includes advertisement data of advertisements displayed to the visiting user; Based on the user portrait data, determining at least one candidate advertisement data matching the visiting user; Inputting each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model respectively, and obtaining delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model; Determine the optimal bid corresponding to each of the candidate advertisement data based on the estimated data of delivery effect corresponding to each of the candidate advertisement data; The optimal advertisement data is determined according to the optimal bids corresponding to the candidate advertisement data, and the optimal advertisement data is displayed to the visiting user.

2. The method according to claim 1, characterized in that: The step of inputting each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtaining delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model, comprises: Extracting designated user data from the user portrait data, and converting the designated user data into a user feature vector; Extracting designated advertisement data from the candidate advertisement data, and converting the designated advertisement data into an advertisement feature vector; For each candidate advertisement data, the corresponding advertisement feature vector and the user feature vector are input into the advertisement delivery estimation model, and the advertisement delivery estimation model estimates the delivery effect estimation data of the current visiting user on the current candidate advertisement data.

3. The method according to claim 1 or 2, characterized in that: The determining the optimal bid corresponding to each candidate advertisement data based on the estimated delivery effect data corresponding to each candidate advertisement data includes: A preset bidding algorithm is used to determine the optimal bid corresponding to the estimated delivery effect data.

4. The method according to claim 1, characterized in that: The advertising delivery estimation model is generated in the following way: Obtaining sample user portrait data of each sample user, wherein the sample user portrait data includes real-time behavior data and historical behavior data of the sample user, and the historical behavior data includes advertisement data of advertisements displayed to the sample user; Constructing a sample feature vector according to the sample user portrait data, wherein the sample feature vector includes user attribute characteristics of each sample user and advertisement attribute characteristics of the advertisement displayed for the sample user, wherein the advertisement attribute characteristics include advertisement type, advertisement click-through rate, and advertisement conversion rate; The sample feature vector is used to train the original logistic regression model to generate an advertising delivery prediction model.

5. The method according to claim 4, characterized in that The method of using the sample feature vector to train the original logistic regression model to generate an advertisement delivery prediction model includes: Performing feature engineering and feature preprocessing on the sample feature vector to generate a feature set, and dividing the feature set into a training set and a test set; Establishing a linear regression model as an original logistic regression model, and transferring the training set into the original logistic regression model for training; The test set is used to predict the trained logistic regression model, and the deviation between the predicted value and the true value is calculated; The training set is optimized based on the deviation, and the training is terminated until the deviation is within a preset range, thereby obtaining an advertisement delivery prediction model.

6. The method according to claim 1, 4 or 5, characterized in that: After presenting the optimal advertisement data to the visiting user, the method further includes: Monitoring the delivery effect of the optimal advertisement; The advertisement delivery prediction model is optimized based on the delivery effect.

7. The method according to claim 1, characterized in that Before obtaining the user portrait data of the current access user, the method further includes: Acquire historical behavior data of each user, wherein the historical behavior data includes advertisement data of advertisements displayed to the user; Constructing user portrait data based on the historical behavior data; By monitoring the real-time behavior of each user, real-time behavior data is obtained, and the real-time behavior data is used to update the user portrait data.

8. An advertisement recommendation system, characterized in that: include: A user portrait acquisition module, used to acquire user portrait data of the current visiting user, wherein the user portrait data includes real-time behavior data and historical behavior data of the visiting user, wherein the historical behavior data includes advertisement data of advertisements displayed to the visiting user; A candidate advertisement determination module, configured to determine at least one candidate advertisement data matching the visiting user based on the user portrait data; The delivery effect estimation module is used to input each candidate advertisement data and the user portrait data into a pre-trained advertisement delivery estimation model, and obtain delivery effect estimation data corresponding to each candidate advertisement data output by the advertisement delivery estimation model; An optimal bid determination module, used to determine the optimal bid corresponding to each candidate advertisement data based on the estimated delivery effect data corresponding to each candidate advertisement data; The optimal advertisement determination module is used to determine the optimal advertisement data according to the optimal bid corresponding to each of the candidate advertisement data, and display the optimal advertisement data to the visiting user.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the advertisement recommendation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the advertisement recommendation method according to any one of claims 1 to 7 is implemented.

Images