Advertisement recommendation method and system

Abstract

The invention discloses an advertisement recommendation method and system, and relates to the field of advertisement recommendation, and the method comprises the steps: obtaining user behavior characteristics and advertisement characteristic label information of a user on a web service page and an app service page; the advertisement feature label information comprises the type of the advertisement and the feature of the advertisement; the features of the advertisement comprise the click rate and the exposure rate of the advertisement; constructing an advertisement system strategy model; the advertisement system strategy model is a neural network model; inputting the user behavior characteristics and the advertisement characteristic label information into the advertisement system strategy model, and outputting scoring results of various types of advertisements; and recommending the advertisement to the user according to the scoring result of the advertisement. Through the user behavior data and the configured corresponding strategy, the advertisement with the highest score for the user is screened out and then recommended to the user, so that the user experience is improved.

Description

Technical Field

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

[0002] At present, with the continuous development of the company's business, the number of users is gradually increasing. Different users are interested in very different content, and the fixed advertising space in the APP has gradually shown many disadvantages.

[0003] A single ad space is likely to waste traffic or lead to loss of potential users. After introducing users, it is impossible to determine the user's intention to download the APP, and it is impossible to carry out further targeted marketing activities. It is difficult to meet user needs, which reduces the user experience.

[0004] Therefore, among the numerous advertisements, how to accurately find the advertisements that users are most interested in and push them to users is exactly the effect that everyone wants to achieve at this stage. Summary of the invention

[0005] The purpose of the present invention is to provide an advertisement recommendation method and system, which can screen out advertisements with the highest scores for users through user behavior data and configured corresponding strategies, and then recommend them to users, thereby improving user experience.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] An advertisement recommendation method, the method comprising:

[0008] Obtaining user behavior characteristics and advertisement feature label information of users on web service pages and app service pages; the advertisement feature label information includes the type of advertisement and the characteristics of the advertisement; the characteristics of the advertisement include the click-through rate and exposure rate of the advertisement;

[0009] Constructing an advertising system strategy model; the advertising system strategy model is a neural network model;

[0010] Inputting the user behavior characteristics and the advertisement feature label information into the advertisement system strategy model, and outputting the scoring results of each type of advertisement;

[0011] Recommending advertisements to the user based on the rating results of the advertisements.

[0012] Optionally, obtaining the user behavior characteristics of the user on the web service page and the app service page and the characteristic tag information of the advertisement specifically includes:

[0013] Obtaining user historical behavior characteristics and historical advertising feature label information of the user on the web service page and the app service page;

[0014] Obtaining real-time user behavior characteristics and real-time advertising characteristic label information of the user on the web service page and the app service page;

[0015] The user's historical behavior characteristics, the historical advertisement feature label information, the user's real-time behavior characteristics and the real-time advertisement feature label information are aggregated and analyzed to obtain the user's aggregated behavior characteristics and aggregated advertisement feature label information.

[0016] Optionally, obtaining the user's historical behavior characteristics and historical advertisement characteristic label information on the web service page and the app service page specifically includes:

[0017] Based on all the user's behavior log information, the user's historical behavior characteristics and the historical advertising feature label information are cleaned out.

[0018] Optionally, the step of constructing the advertising system strategy model specifically includes:

[0019] Adjusting parameters of the advertising system strategy model;

[0020] Replacing the algorithm of the advertising system strategy model;

[0021] The effect of the advertising system strategy model after evaluation parameter adjustment and algorithm replacement.

[0022] Optionally, the adjusting the parameters of the advertising system strategy model specifically includes:

[0023] By modifying the click-through rate and exposure rate of the advertisement, the parameters of the advertising system strategy model are adjusted;

[0024] Optionally, the algorithm for replacing the advertising system strategy model specifically includes:

[0025] The algorithm of the advertising system strategy model is modified by switching the DeepFM model and the FNN model; the advertising system strategy model includes the DeepFM model and the FNN model.

[0026] The present invention also provides an advertisement recommendation system, the system comprising:

[0027] A data acquisition module, used to acquire user behavior characteristics and advertisement feature label information of users on web service pages and app service pages; the advertisement feature label information includes the type of advertisement and the characteristics of the advertisement; the characteristics of the advertisement include the click-through rate and exposure rate of the advertisement;

[0028] A strategy model building module, used to build an advertising system strategy model; the advertising system strategy model is a neural network model;

[0029] An advertisement scoring module, used to input the user behavior characteristics and the advertisement feature label information into the advertisement system strategy model, and output the scoring results of each type of advertisement;

[0030] The advertisement recommendation module is used to recommend advertisements to the user based on the rating results of the advertisements.

[0031] Optionally, the data acquisition module specifically includes:

[0032] A historical data acquisition unit, used to acquire the user's historical behavior characteristics and historical advertisement feature label information on the web service page and the app service page;

[0033] A real-time data acquisition unit, used to acquire the user's real-time behavior characteristics and real-time advertising feature label information on the web service page and the app service page;

[0034] The data aggregation and analysis unit is used to aggregate and analyze the user's historical behavior characteristics, the historical advertisement feature label information, the user's real-time behavior characteristics, and the real-time advertisement feature label information to obtain the user's aggregated behavior characteristics and aggregated advertisement feature label information.

[0035] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0036] The present invention provides an advertisement recommendation method and system, which obtains relevant behavioral characteristics of users when browsing pages and the click-through rate and exposure rate of relevant advertisements, and scores the advertisement data in combination with a neural network model, thereby determining the advertisements that the users are interested in based on the scores, and then pushing the advertisements of interest to the users. It is possible to push advertisements to users in a targeted manner, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. 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 paying creative labor.

[0038] Figure 1 A flow chart of an advertisement recommendation method provided in Example 1 of the present invention;

[0039] Figure 2 A flow chart of the method of step S1 provided in Example 1 of the present invention;

[0040] Figure 3 A flow chart of the method of step S2 provided in Example 1 of the present invention;

[0041] Figure 4 This is a block diagram of an advertisement recommendation system provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of 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.

[0043] The purpose of the present invention is to provide an advertisement recommendation method and system, which can screen out advertisements that users really like based on a large amount of user behavior data and recommend them to the users.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] Example 1

[0046] like Figure 1 As shown, the user calls the advertising interface instruction through the web service or App.

[0047] S0: When a user opens a web service page or a corresponding page in an app, the web service and the app will call the recommended advertising interface command to the server-side advertising service. At the same time, the user's username and other information will be passed to the advertising service in the request.

[0048] like Figure 1 As shown, this embodiment provides an advertisement recommendation method, the method comprising:

[0049] S1: Obtaining user behavior characteristics and advertisement feature label information of users on web service pages and app service pages; the advertisement feature label information includes the type of advertisement and the characteristics of the advertisement; the characteristics of the advertisement include the click-through rate and exposure rate of the advertisement.

[0050] The advertising service obtains the classification attributes of the current user and the classification attributes of the advertisement from the prepared data according to the user request information in requset. The data source is prepared by obtaining the user's behavioral attributes and the label data of the advertisement, and adding the real-time user's behavioral attributes and the characteristic label data of the advertisement to clean up the characteristic data such as the user type, the advertisement type, the click rate, and the exposure rate.

[0051] Among them, Figure 2 As shown, step S1 specifically includes the step of executing the pre-test case, which specifically includes the following contents:

[0052] S11: Obtain the user's historical behavior characteristics and historical advertisement feature label information on the web service page and the app service page.

[0053] All user behavior log information is cleaned up through a distributed computing platform to extract the user's historical behavior characteristics and the historical advertising feature label information.

[0054] Through batch processing operations on a distributed computing platform, user feature information of test case data and click exposure data of advertisements are pre-generated and stored in the corresponding search engine.

[0055] S12: Obtain the real-time behavior characteristics and real-time advertisement characteristic tag information of the user on the web service page and the app service page.

[0056] Real-time user behavior data is used to calculate the current user feature label data and real-time advertising click exposure feature data through the stream computing platform.

[0057] Through the stream computing platform, based on the user's behavior data, the user's real-time feature label data and the advertisement's real-time click exposure feature data are analyzed, and the data is stored in the corresponding search engine.

[0058] S13: performing data aggregation and analysis on the user's historical behavior characteristics, the historical advertisement feature label information, the user's real-time behavior characteristics, and the real-time advertisement feature label information to obtain user aggregated behavior characteristics and aggregated advertisement feature label information.

[0059] The data is aggregated and analyzed to obtain the user feature label data and ad click exposure feature data for final use.

[0060] The batch processed non-real-time data and the stream processed real-time data are thrown into the model for analysis and processing. The main purpose of the model here is to analyze the characteristic label data of users and advertisements based on the non-real-time data and real-time data, and calculate the final characteristic label of users and the final click exposure characteristic data of advertisements.

[0061] In this embodiment, when the precondition information of the associated advertising feature information and user feature information is queried, a precondition test case is generated according to the precondition information, the precondition interface is tested according to the generated precondition test case, and the parameter information of the case to be tested corresponding to the test case identifier returned by the precondition test case by the precondition interface is obtained, thereby ensuring the accuracy of the parameters of the case to be tested.

[0062] S2: Constructing an advertising system strategy model; the advertising system strategy model is a neural network model.

[0063] When the advertising service receives a user request, it will first confirm whether the current system's policy model status is enabled. If enabled, the latest policy model will be obtained. If not enabled, the default policy model will be obtained. The model has a switch flag, which can be judged by the corresponding flag. This operation can change the advertising policy model through configuration, which is convenient for learning and testing new models later. The default strategy is to feel the user's tags and ad click exposure. The new strategy may be a new model later, or it may add new feature identifiers for calculation, etc. The purpose of this switch is mainly for grayscale testing and switching.

[0064] Among them, Figure 3 As shown, step S2 specifically includes the step of executing the pre-test case, which specifically includes the following contents:

[0065] S21: Adjusting parameters of the advertising system strategy model.

[0066] By modifying the percentage scores of parameters such as the click-through rate and exposure rate of the advertisement, the parameters of the advertisement system strategy model are adjusted. The adjusted parameters of the model are used to analyze whether the model meets the needs of the system.

[0067] S22: Replace the algorithm of the advertising system strategy model.

[0068] The algorithm of the advertising system strategy model is modified by switching the DeepFM model and the FNN model; the algorithm of the adjusted model is used to analyze whether the model meets the needs of the system.

[0069] The advertising system strategy model includes the DeepFM model and the FNN model.

[0070] S23: evaluating the effect of the advertising system strategy model after parameter adjustment and algorithm replacement.

[0071] Analyze the model effect through the data collection center. Release multiple models to the production environment in grayscale, and then analyze the user behavior logs of the data collection platform to confirm the final effect of the model.

[0072] In this embodiment, when the model scores the precondition information, the model is adjusted through parameters and algorithms to obtain the model data that best matches the user behavior, so as to improve the final test effect.

[0073] S3: Input the user behavior characteristics and the advertisement feature label information into the advertisement system strategy model, and output the scoring results of each type of advertisement.

[0074] The data used in the above acquisition is thrown into the acquired strategy model for running. The strategy model is based on the feature label of the current user through the DeepFM strategy model. It is just the calculation method of the final output of the last layer of the model, y FM ,y DNN They are the output results of the previous network models FM and DNN, which score the known feature labels such as the click exposure of the advertisement.

[0075] S4: Recommending advertisements to the user according to the rating results of the advertisements.

[0076] Sort the ads by score and return them to the user for recommendation. Sort the ads by score and return the ads with the highest score to the user.

[0077] This method provides an advertising delivery method based on user characteristics, takes historical data and real-time data as input, and combines big data algorithms and deep learning of models to more effectively select suitable advertisements.

[0078] In this embodiment, the most critical technical points are the three modules of model analysis of user behavior characteristics, data analysis of ad click exposure characteristics, and strategy model scoring. These three modules are mainly used to analyze user portraits, obtain user feature labels, obtain ad click exposure feature information, and finally calculate user preference scores based on the model. By analyzing user preferences, it is possible to filter out ads that users really like and recommend them to users.

[0079] Example 2

[0080] This embodiment provides an advertisement recommendation system, the system comprising:

[0081] The data acquisition module is used to obtain the user behavior characteristics and advertising feature label information of users on the web service page and the app service page; the advertising feature label information includes the type of advertisement and the characteristics of the advertisement; the characteristics of the advertisement include the click-through rate and exposure rate of the advertisement.

[0082] Wherein, the data acquisition module specifically includes:

[0083] The historical data acquisition unit is used to acquire the user's historical behavior characteristics and historical advertising feature label information on the web service page and the app service page.

[0084] The real-time data acquisition unit is used to acquire the real-time behavior characteristics and real-time advertisement feature label information of the user on the web service page and the app service page.

[0085] The data aggregation and analysis unit is used to aggregate and analyze the user's historical behavior characteristics, the historical advertisement feature label information, the user's real-time behavior characteristics, and the real-time advertisement feature label information to obtain the user's aggregated behavior characteristics and aggregated advertisement feature label information.

[0086] The strategy model building module is used to build an advertising system strategy model; the advertising system strategy model is a neural network model.

[0087] The advertisement scoring module is used to input the user behavior characteristics and the advertisement feature label information into the advertisement system strategy model and output the scoring results of each type of advertisement.

[0088] The advertisement recommendation module is used to recommend advertisements to the user based on the rating results of the advertisements.

[0089] Figure 4 It shows a more comprehensive overall framework of this system, including:

[0090] User behavior collection platform, distributed computing platform, stream computing platform, advertising system, web server, and user terminal.

[0091] The user behavior collection platform is responsible for the behavior data of mobile phone users and the click-through rate and exposure rate of advertisements;

[0092] The distributed computing platform is responsible for cleaning the user behavior data, ad click rate, and exposure rate into the data needed for the model;

[0093] The stream computing platform is responsible for processing real-time user behavior data and ad click-through rates and exposure rates to clean out the data required for the model;

[0094] The advertisement system is responsible for advertisement retrieval and advertisement recommendation; wherein the advertisement recommendation method described in the embodiment is executed in the advertisement system.

[0095] The Web server is used to realize data interaction between the user end and the system;

[0096] The user end is a mobile device such as a computer or a smartphone, which is the vehicle through which customers browse advertisements.

[0097] Among them, the user behavior collection platform (i.e., the data collection platform in the figure) is unidirectionally connected to the advertising system, distributed computing platform, and stream computing platform to provide data; and unidirectionally receives the return data from the web server;

[0098] The distributed computing platform and the stream computing platform are unidirectionally connected to the advertising system respectively;

[0099] The advertising system is connected to the web server in a one-way manner;

[0100] The Web server has a bidirectional connection with the client and a unidirectional connection with the data collection platform.

[0101] Through the data collection platform, we obtain user behavior information and information such as ad clicks and exposures. Through summary analysis, we analyze the user's characteristic tags and the ad's characteristic tags, and then synchronize them with ad retrieval to optimize the accuracy of ad retrieval.

[0102] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0103] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. An advertisement recommendation method, characterized in that: The method comprises: Obtaining user behavior characteristics and advertisement feature label information of users on web service pages and app service pages; the advertisement feature label information includes the type of advertisement and the characteristics of the advertisement; the characteristics of the advertisement include the click-through rate and exposure rate of the advertisement; Constructing an advertising system strategy model; the advertising system strategy model is a neural network model; Inputting the user behavior characteristics and the advertisement feature label information into the advertisement system strategy model, and outputting the scoring results of each type of advertisement; Recommending advertisements to the user based on the rating results of the advertisements.

2. The method according to claim 1, characterized in that The obtaining of user behavior characteristics and characteristic label information of advertisements on the web service page and the app service page specifically includes: Obtaining user historical behavior characteristics and historical advertising feature label information of the user on the web service page and the app service page; Obtaining real-time user behavior characteristics and real-time advertising characteristic label information of the user on the web service page and the app service page; The user's historical behavior characteristics, the historical advertisement feature label information, the user's real-time behavior characteristics and the real-time advertisement feature label information are aggregated and analyzed to obtain the user's aggregated behavior characteristics and aggregated advertisement feature label information.

3. The method according to claim 2, characterized in that The obtaining of the user's historical behavior characteristics and historical advertisement characteristic label information on the web service page and the app service page specifically includes: Based on all the user's behavior log information, the user's historical behavior characteristics and the historical advertising feature label information are cleaned out.

4. The method according to claim 1, characterized in that: The construction of the advertising system strategy model specifically includes: Adjusting parameters of the advertising system strategy model; Replacing the algorithm of the advertising system strategy model; The effect of the advertising system strategy model after evaluation parameter adjustment and algorithm replacement.

5. The method according to claim 4, characterized in that The adjusting the parameters of the advertising system strategy model specifically includes: By modifying the click-through rate and exposure rate of the advertisement, the parameters of the advertisement system strategy model are adjusted.

6. The method according to claim 4, characterized in that The algorithm for replacing the advertising system strategy model specifically includes: The algorithm of the advertising system strategy model is modified by switching the DeepFM model and the FNN model; the advertising system strategy model includes the DeepFM model and the FNN model.

7. An advertisement recommendation system based on the method according to any one of claims 1 to 6, characterized in that: The system comprises: A data acquisition module, used to acquire user behavior characteristics and advertisement feature label information of users on web service pages and app service pages; the advertisement feature label information includes the type of advertisement and the characteristics of the advertisement; the characteristics of the advertisement include the click-through rate and exposure rate of the advertisement; A strategy model building module, used to build an advertising system strategy model; the advertising system strategy model is a neural network model; An advertisement scoring module, used to input the user behavior characteristics and the advertisement feature label information into the advertisement system strategy model, and output the scoring results of each type of advertisement; The advertisement recommendation module is used to recommend advertisements to the user based on the rating results of the advertisements.

8. The system according to claim 7, characterized in that The data acquisition module specifically includes: A historical data acquisition unit, used to acquire the user's historical behavior characteristics and historical advertisement feature label information on the web service page and the app service page; A real-time data acquisition unit, used to acquire the user's real-time behavior characteristics and real-time advertising feature label information on the web service page and the app service page; The data aggregation and analysis unit is used to aggregate and analyze the user's historical behavior characteristics, the historical advertisement feature label information, the user's real-time behavior characteristics, and the real-time advertisement feature label information to obtain the user's aggregated behavior characteristics and aggregated advertisement feature label information.

Images