Advertisement recommendation method and device, computer device and storage medium

Abstract

The application relates to an advertisement recommendation method and device, computer equipment and a storage medium, the method comprising the following steps: obtaining a user identifier and a triggered carrier content identifier; obtaining a previously established corresponding relationship among the user identifier, the carrier content identifier and an advertisement identifier according to the user identifier and the carrier content identifier, the corresponding relationship being previously established according to the recognition degree of the advertisement displayed by each carrier content to the user identifier; determining the corresponding advertisement identifier according to the corresponding relationship; and sending the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier to a terminal corresponding to the user identifier after preset processing. The method makes the recommended advertisement meet the user interest, realizes accurate advertisement recommendation, and on this basis, the theme of the recommended advertisement is also matched with the theme of the carrier content, the sense of strangeness of advertisement insertion is reduced, and the advertisement delivery effect is improved.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to an advertising recommendation method, apparatus, computer device, and storage medium. Background Technology

[0002] Traditional advertising media are fixed, such as billboards. However, with the development of internet technology and the widespread use of mobile devices, advertising media have become more diverse, with more and more advertisements appearing in videos and e-books.

[0003] For example, video streaming platforms insert ads at the beginning, middle, or end of videos for advertising purposes. However, if the ads inserted in the same video are fixed for a period of time, this method of inserting fixed ads cannot accurately target users who are interested in the ads, resulting in poor advertising effectiveness. Summary of the Invention

[0004] Therefore, it is necessary to provide an advertising recommendation method, device, computer equipment, and storage medium to address the technical problem of poor advertising performance.

[0005] An advertising recommendation method includes:

[0006] Obtain the user identifier and the identifier of the triggered carrier content;

[0007] Based on the user identifier and carrier content identifier, obtain the pre-established correspondence between the user identifier, carrier content identifier and advertisement identifier. The correspondence is pre-established based on the user identifier's approval of the advertisements displayed on each carrier content.

[0008] The corresponding advertising identifier is determined based on the aforementioned correspondence.

[0009] After the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier are pre-processed, they are sent to the terminal corresponding to the user identifier.

[0010] An advertising recommendation device includes: an identifier acquisition module, a search module, an advertisement determination module, and a processing module;

[0011] The identifier acquisition module is used to acquire the user identifier and the triggered carrier content identifier;

[0012] The search module is used to obtain a pre-established correspondence between user identifier, carrier content identifier and advertisement identifier based on the user identifier and carrier content identifier. The correspondence is pre-established based on the user identifier's approval of the advertisements displayed on each carrier content.

[0013] The advertisement determination module is used to determine the corresponding advertisement identifier based on the correspondence relationship;

[0014] The processing module is used to perform preset processing on the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier, and then send them to the terminal corresponding to the user identifier.

[0015] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method described above.

[0016] A storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the above-described method.

[0017] The aforementioned advertising recommendation method, apparatus, computer equipment, and storage medium, after obtaining the user identifier and carrier content identifier, determine the corresponding advertising identifier based on the pre-established correspondence between the user identifier, carrier content identifier, and advertising identifier. The corresponding advertising and carrier content are then processed and sent to the terminal. Since the correspondence between the user identifier, carrier content identifier, and advertising identifier is pre-established based on the user's acceptance of the advertisements displayed on each carrier content, it reflects the degree of matching between user interests and carrier content, carrier content and advertisements, and user interests and advertisements. Therefore, the pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers reflects the relationship between user interests, carrier content themes, and advertising themes, ensuring that recommended advertisements match user interests and achieving accurate advertising recommendations. Furthermore, the theme of the recommended advertisements also aligns with the carrier content theme, reducing the abruptness of ad insertion and improving advertising effectiveness. Attached Figure Description

[0018] Figure 1 This is a diagram illustrating the application environment of an advertising recommendation method in one embodiment;

[0019] Figure 2 This is a flowchart illustrating an advertising recommendation method in one embodiment;

[0020] Figure 3 This is a flowchart illustrating the steps for obtaining the user's approval of the advertisements displayed on each carrier content in one embodiment.

[0021] Figure 4 A flowchart illustrating the steps of determining a topic model and determining a user interest topic matrix, an advertising topic matrix, and a carrier content matrix based on the topic model, as one embodiment;

[0022] Figure 5 This is a schematic diagram illustrating the relationship between the user's approval matrix C, the user interest topic model U, and the carrier content topic model V in one embodiment.

[0023] Figure 6 This is a schematic diagram illustrating the relationship between the user's approval matrix Y, the user interest topic model, and the advertising topic model A in one embodiment.

[0024] Figure 7 This is a schematic diagram illustrating the relationship between the similarity matrix Z of carrier content and advertisement, the carrier content theme model V, and the advertisement theme model A in one embodiment.

[0025] Figure 8 This is a matrix factorization model of the topic model in one embodiment;

[0026] Figure 9 A sequence diagram of an ad recommendation method in one embodiment.

[0027] Figure 10 This is a structural block diagram of an advertising recommendation device in one embodiment;

[0028] Figure 11 This is a structural block diagram of an advertising recommendation device in another embodiment;

[0029] Figure 12 This is a structural block diagram of a computing device in one embodiment. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0031] Figure 1 This is a diagram illustrating the application environment of an advertising recommendation method in one embodiment. (Refer to...) Figure 1 This advertising recommendation method is applied to an advertising recommendation system. The advertising recommendation system includes a terminal 110 and a server 120. The terminal 110 and server 120 are connected via a network. The terminal 110 can be a desktop terminal or a mobile terminal; a mobile terminal can be at least one of a mobile phone, tablet computer, or laptop computer. The server 120 can be a standalone server or a server cluster consisting of multiple servers.

[0032] Figure 2 This is a flowchart illustrating an advertising recommendation method in one embodiment. This embodiment primarily applies this method to the above-mentioned... Figure 1 Let's take server 120 as an example. (Refer to...) Figure 2 The advertising recommendation method specifically includes the following steps:

[0033] S202, Obtain the user identifier and the triggered carrier content identifier.

[0034] The carrier content identifier is a unique identifier used to distinguish the content of an advertisement from its carrier. The carrier of an advertisement refers to the form in which the advertisement is published. For example, if the carrier of an advertisement is video, then the carrier content identifier is the identifier for the video content, used to distinguish different video content. Similarly, if the carrier of an advertisement is an e-book, then the carrier content identifier is the identifier for the e-book content, used to distinguish different e-book content.

[0035] A trigger refers to a pre-defined action performed by a user to load content. For example, opening an e-book or video within a corresponding application.

[0036] S204: Based on the user identifier and the carrier content identifier, obtain the pre-established correspondence between the user identifier, the carrier content identifier and the advertisement identifier. The correspondence is pre-established based on the user's recognition of the advertisement displayed on each carrier content.

[0037] User approval rating for advertisements displayed on various content platforms refers to the user's level of acceptance of the advertisements displayed on the currently viewed content platform. This approval rating is determined by the matching degree between the user's interest topics, the content platform's theme, and the advertisement's theme. Specifically, the user's approval rating for the advertisements displayed on the currently viewed content platform is positively correlated with the matching degree among these three factors.

[0038] For an advertising medium, the number of ads placed by an advertiser is fixed within a certain period. However, for the platform to which the advertising medium belongs (such as a video streaming platform or an e-book reading platform), there are multiple content formats. It is necessary to establish a pre-defined correspondence between user identifiers, different content format identifiers, and ad identifiers; that is, to determine which ads are displayed when a user is viewing different content formats.

[0039] Specifically, for a user identifier i, calculate the display of different advertisements f for a carrier content e. (0,1,...d) The degree of recognition is denoted by d, where d represents the number of advertisements. A correspondence is established between the user identifier, the carrier content identifier, and the advertisement identifier based on the advertisement identifier with the highest recognition. This process is repeated to determine the advertisement with the highest recognition for a given user identifier when viewing all carrier content, and to establish a correspondence between the user identifier, each carrier content identifier, and the advertisement identifier, as shown in Table 1.

[0040] Table 1. Correspondence between User Identifier, Content Identifier, and Advertising Identifier

[0041]

[0042]

[0043] In practical applications, the correspondence between user identifiers, carrier content identifiers, and advertising identifiers is found based on user identifiers and carrier content identifiers.

[0044] The correspondence between user identifiers, content identifiers, and ad identifiers is pre-established based on the user's level of approval for the ads displayed on each content platform. When a user approves of an ad displayed on a particular content platform, it indicates that the user agrees with the ad playing alongside that content; that is, the ad satisfies the user's interests, matches the content (they don't appear jarring together, but rather harmoniously), and the content also meets the user's needs. In other words, the degree of approval a user gives to ads displayed on each content platform reflects the degree of matching between user interests and content, content and ads, and user interests and ads.

[0045] S206: Determine the corresponding advertising identifier based on the correspondence.

[0046] Specifically, the advertising identifier is determined based on the correspondence between the found user identifier, carrier content identifier, and advertising identifier.

[0047] S208: After pre-processing the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier, send them to the terminal corresponding to the user identifier.

[0048] Preset processing refers to the process of associating advertising content with the advertising platform. The methods of preset processing are related to the operational strategy of the advertising platform and are not limited here. Taking video playback platforms as an example, advertisements can be inserted at the beginning, middle, or end of the video and sent to the user's terminal in the form of streaming media. Taking e-book reading platforms as an example, advertisements can be inserted at the beginning of each chapter of the e-book and sent to the user's terminal.

[0049] The aforementioned advertising recommendation method, after obtaining user identifiers and carrier content identifiers, determines the corresponding advertising identifier based on a pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers. The corresponding advertising and carrier content are then pre-processed and sent to the terminal. Since the correspondence between user identifiers, carrier content identifiers, and advertising identifiers is pre-established based on the user's level of acceptance of the advertisements displayed on each carrier content, it reflects the degree of matching between user interests and carrier content, carrier content and advertisements, and user interests and advertisements. Therefore, the pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers reflects the relationship between user interests, carrier content themes, and advertising themes, ensuring that recommended advertisements match user interests and achieving accurate advertising recommendations. Furthermore, the theme of the recommended advertisements also aligns with the carrier content theme, reducing the abruptness of ad insertion and improving advertising effectiveness.

[0050] Specifically, the steps for pre-establishing the correspondence between user identifiers, carrier content identifiers, and advertising identifiers include the following steps S1 to S2:

[0051] S1: Obtain the level of approval of each user for the advertisements displayed on each carrier.

[0052] User approval rating for advertisements displayed on various content platforms refers to the user's level of acceptance of the advertisements displayed on the currently viewed content platform. This approval rating is determined by the matching degree between the user's interest topics, the content platform's theme, and the advertisement's theme. Specifically, the user's approval rating for the advertisements displayed on the currently viewed content platform is positively correlated with the matching degree among these three factors.

[0053] User interest theme refers to the theme to which user interests belong, advertising theme refers to the theme to which the advertisement belongs, and carrier content theme refers to the theme to which the carrier content belongs. A theme refers to the category to which the content pertains. In one embodiment, a theme may include sports, health, automobiles, fashion, parenting, real estate, etc.

[0054] Suppose there exists a potential subject set S = {S1, S2, S3, ..., s} m Let m be the number of topics. User interests, videos, and ads all have some potential correlation with these topics; for example, users might be interested in certain topics, and videos and ads belong to certain topics. For instance, topics obtained through training could include: sports, health, cars, fashion, parenting, real estate, etc.

[0055] S2: Based on the most recognized advertisements, establish the correspondence between user identifiers, carrier content identifiers, and advertisement identifiers.

[0056] For an advertising medium, the number of ads placed by an advertiser is fixed within a certain period. However, for the platform to which the advertising medium belongs (such as a video streaming platform or an e-book reading platform), there are multiple content formats. It is necessary to establish a pre-defined correspondence between user identifiers, different content format identifiers, and ad identifiers; that is, to determine which ads are displayed when a user is viewing different content formats.

[0057] Specifically, for a user identifier i, calculate the display of different advertisements f for a carrier content e. (0,1,...d)The degree of user approval is denoted by d, where d represents the number of advertisements. A correspondence is established between the user identifier, the carrier content identifier, and the advertisement identifier based on the advertisement identifier with the highest degree of user approval. It should be understood that the advertisement with the highest degree of user approval for a given carrier content should be the advertisement corresponding to the highest match between the user's interest topic, the carrier content topic, and the advertisement topic. This process is repeated to determine the advertisement with the highest approval for a given user identifier when viewing all carrier content, establishing a correspondence between the user identifier, each carrier content identifier, and the advertisement identifier, as shown in Table 1.

[0058] Figure 3 A flowchart illustrating the steps for obtaining user IDs' approval of advertisements displayed on various carrier content, as described in one embodiment. Figure 3 As shown, this step includes the following steps: S302 to S308:

[0059] S302, obtain the user interest theme matrix, advertising theme matrix, and carrier content theme matrix.

[0060] The user interest topic matrix U indicates the degree to which each user's interest belongs to each topic. i,j U represents the degree of preference of user i for topic j. ij ∈[0,1],U m×n Let n represent the number of users and m represent the number of topics. A column vector in matrix U represents a user's preference for each topic, i.e., the degree to which a user's interest belongs to a particular topic. A user can like multiple topics, with varying degrees of preference for different topics. A user's preference for a topic reflects their interest in that topic, and there is a positive correlation between the degree of preference and the user's interest in that topic. The higher the user's preference for a topic, the more interested they are in that topic.

[0061] The content matrix V indicates the degree to which a content belongs to each topic. A piece of content can belong to multiple topics, with varying degrees of matching to different topics. V e,j V represents the degree of matching between the content e and the theme j. e,j ∈[0,1],V m ×b b represents the quantity of carrier content, and a column vector of matrix V identifies the degree to which a carrier content belongs to each topic.

[0062] The advertising theme matrix A indicates the degree to which each ad content belongs to a particular theme. An ad can belong to multiple themes, with varying degrees of relevance. f,j A represents the degree of match between advertisement f and topic j. f,j ∈[0,1],A m×d d represents the number of ads, and a column vector of matrix A indicates the degree to which an ad belongs to each topic.

[0063] S304. Calculate the user's recognition of the advertisements displayed on each carrier content using the user interest topic matrix, the advertisement topic matrix, and the carrier content topic matrix.

[0064] Specifically, based on the advertising theme matrix A, the carrier content theme matrix V, and the user interest theme matrix U, the degree of user acceptance of the advertisements displayed on each carrier content is calculated. That is, when user i is viewing a carrier content e, the degree of user acceptance of advertisement f can be expressed as:

[0065] L(u,v,a)=(1-λ)u T a+(1-β)v T a

[0066] u is the theme that user i likes, i.e., u = U. ·i v is the theme to which the content e belongs, i.e., v = V ·j 'a' is the topic distribution to which advertisement 'f' belongs, a = A ·k .

[0067] Specifically, before step S302, the method further includes the steps of determining a topic model and determining a user interest topic matrix, an advertising topic matrix, and a carrier content matrix based on the topic model.

[0068] Figure 4 A flowchart illustrating the steps of determining a topic model and, based on that model, a user interest topic matrix, an advertising topic matrix, and a carrier content matrix, as shown in one embodiment, is as follows: Figure 4 As shown, this step includes:

[0069] S402, acquire historical behavioral data, and determine the user's approval of the carrier content, the user's approval of the advertisement, and the similarity between the carrier content and the advertisement based on the historical behavioral data.

[0070] Historical behavioral data refers to the behavioral data generated by users during the use of an application. For example, in a video application, this includes users' video viewing behavior and ad clicking behavior. Specifically, historical behavioral data includes user viewing behavior data, user ad clicking behavior data, and ad clicking behavior data when playing content.

[0071] In one embodiment, viewing behavior data of each user identifier on the carrier content is obtained, and the user's approval of the carrier content is determined based on the viewing behavior data.

[0072] Among these, user approval of the content refers to the degree to which users like the content. User viewing behavior data refers to a user's viewing history for each piece of content, including viewing quality information, and the user's approval of the content is determined based on the viewing quality information. Viewing quality information refers to the degree to which a user completes watching a piece of content. Specifically, by obtaining the completion rate of each piece of content watched by each user, a user approval matrix C is determined. Taking video as an example, C... i,e The duration of user i watching video e divided by the duration of video e itself is equivalent to the completeness of user i's viewing of video e, which is a relatively sparse matrix.

[0073] In one embodiment, user click behavior data corresponding to a user identifier is obtained, and the user's approval of the advertisement is determined based on the user click behavior data.

[0074] User approval rating refers to the degree to which a user likes a particular advertisement. User click behavior data is generated from user click behavior during historical ad servings. Specifically, user approval rating can be determined through user click behavior data, that is, by judging whether users clicked on ads during historical ad servings, thus obtaining the user approval rating matrix Y. For example, if user i clicked on ad f, then Y... i,f =1, otherwise Y i,f =0, Y is an extremely sparse matrix.

[0075] In one embodiment, ad click behavior data during playback of carrier content is used to determine the similarity between the carrier content and the ad.

[0076] The similarity between carrier content and advertisement refers to the degree of similarity between the theme of the carrier content and the theme of the advertisement. This similarity can be determined by pre-classifying the carrier content and then manually assigning similarity levels to the carrier content categories and advertisements based on the advertisement content. Alternatively, it can be confirmed based on ad click behavior data during carrier content playback. This data refers to user clicks on advertisements associated with the carrier content while it is playing. For example, data generated when a user clicks on an inserted advertisement while watching a video. Specifically, by determining whether a user clicks on the advertisement during video playback, the similarity information between the carrier content and the advertisement is obtained, resulting in the similarity score Z. For example, Z... e,f= The number of times ad f was clicked while playing video e / the total number of clicks on ad f. First, Z is regularized, and Z is also a sparse matrix. S404, based on the relationship between user approval of the carrier content, user interest topics, and carrier content topics, the relationship between user approval of the ad, ad topics, and user interest topics, and the relationship between the similarity between carrier content and ad, carrier content topics, and ad topics, the topic model is determined.

[0077] User interest topics refer to the topics to which users belong, advertising topics refer to the topics to which advertisements belong, and carrier content topics refer to the topics to which carrier content belongs. Assuming there exists a potential topic library, the user interest topics, advertising topics, and carrier content topics constitute the topics in this library. Solving the topic model yields the topic library. The principle of advertising recommendation is to strive for a match between user interest topics, advertising topics, and carrier content topics to gain user approval. The matrix C representing user approval of carrier content is as follows: Figure 5 As shown, through a series of training iterations, the user interest topic matrix U (representing the user's preference for each topic) and the carrier content topic matrix V (representing the degree of matching between the carrier content and the topic) are inferred on the right. Specifically, the user's approval of the carrier content can be represented as the similarity between the user's preferred topic (interest topic) and the topic to which the carrier content being viewed belongs. For example, if user i is interested in carrier content e, then C... i,e If the value is relatively large, then user i is interested in topic U. i The theme V to which the content e belongs. j They are quite similar, so It's also relatively large, meaning user i is interested in the content of the carrier, compared to C. i,e The expression is consistent. That is, the relationship between user approval of carrier content, user interest topics, and carrier content topics includes: user approval of carrier content is positively correlated with the similarity between user interest topics and carrier content topics.

[0078] User approval matrix Y, such as Figure 6 As shown, through a series of training iterations, the user interest topic matrix U (representing the degree of user preference for each topic) and the advertisement topic matrix A (representing the degree of matching between advertisements and topics) are inferred on the right. Specifically, advertisements that users approve of can be characterized as those whose user interest topics are similar to the topics to which the advertisements belong. For example, if user i is interested in advertisement f, then Y... i,f If the value is relatively large, then user i is interested in topic U. i The theme A to which advertisement f belongs f They are quite similar, so It's also relatively large, meaning user i is interested in the ad, compared to Y. i,fThe expression is consistent. That is, the relationship between user acceptance of an advertisement, the advertisement theme, and user interest themes includes: user acceptance of an advertisement is positively correlated with the similarity between the advertisement theme and the user interest themes.

[0079] The similarity matrix Z between the carrier content and the advertisement, such as Figure 7 As shown, through a series of training iterations, the carrier content-topic matrix V, representing the degree of matching between the carrier content and the theme, and the advertising-topic matrix A, representing the degree of similarity between the advertisement and the theme, are inferred on the right. Specifically, the similarity between the carrier content and the advertisement can be represented as the similarity between the theme to which the carrier content belongs and the subject to which the advertisement belongs. For example, if the carrier content e is similar to the advertisement, then Z... e,f If the value is relatively large, then the theme V of the carrier content e is... e The theme A to which advertisement f belongs f If they are quite similar, then V e t ×A f It's also quite large, meaning the content and advertising are similar to Z. e,f The expressions are consistent. That is, the relationship between the similarity between the carrier content and the advertisement, the theme of the carrier content and the theme of the advertisement includes: the degree of similarity between the carrier content and the advertisement, and the degree of similarity between the theme of the carrier content and the theme of the advertisement. Figure 8 This is a matrix factorization model of a topic model for one embodiment. Based on the topic matrix factorization model, the optimization equation can be obtained:

[0080] min: J=λ||CU T V||2+α||YU T A||2+β||ZV T A||2+δ(||U||2+||V||2+||A||2)

[0081] Among them, U T V represents the user's level of liking and approval of the content, U T A represents the user's level of liking for the advertisement, V T A represents the similarity between the carrier content and the advertisement, ||·||² squared loss function, λ, α, β, δ are the corresponding weight coefficients, ||U||² + ||V||² + ||A||² is the regularization term, C is the user's approval of the carrier content, Y is the user's approval of the advertisement, and Z is the similarity between the carrier content and the advertisement.

[0082] S406. Based on the topic model, determine the user interest topic matrix, advertising topic matrix, and carrier content matrix.

[0083] Specifically, step S406 includes the following steps:

[0084] S1: Determine the number of topics.

[0085] Suppose there exists a potential subject set S = {s1, s2, s3, ..., s} m Let m be the number of topics. User interests, videos, and advertisements all have some potential connection to these topics; for example, users might be interested in certain topics, and videos and advertisements belong to certain topics. The number of topics m is set manually.

[0086] S2. Initialize the values ​​of the user interest topic matrix, carrier content topic matrix, and advertising topic matrix.

[0087] S3. Based on the values ​​of the carrier content theme matrix V and the advertising theme matrix A, calculate the derivative of the user interest theme matrix and solve to update the value of the current user interest theme matrix.

[0088] Specifically, by taking the derivative of the user interest topic matrix U, we can obtain the following equation:

[0089] tr[U T (λVV T +αAA T +δI)U]=2(λ+α)tr(U T A)

[0090] Where tr(*) is the trace of the matrix, the value of the user interest topic matrix U can be obtained by using the relationship between the eigenvalues ​​and eigenvectors of the matrix.

[0091] S4. Based on the values ​​of the current user interest topic matrix U and the advertising topic matrix A, calculate the derivative of the carrier content topic matrix V, and solve to update the value of the current carrier content topic matrix V.

[0092] S5. Based on the values ​​of the current carrier content theme matrix V and the current user interest theme matrix U, calculate the derivative of the advertising theme matrix A and solve to update the value of the current advertising theme matrix A.

[0093] The order of steps S3, S4 and S5 is not limited. The optimal values ​​of two topic matrices can be fixed arbitrarily, and the derivative of the other topic matrix can be obtained by solving for the value of the topic matrix.

[0094] S6. Calculate the error of the topic model based on the values ​​of the current advertisement-topic matrix A, the current carrier content topic matrix V, and the current user interest topic matrix U.

[0095] Specifically, the calculated values ​​of the current advertising theme matrix A, the current carrier content theme matrix V, and the user interest theme matrix U are substituted into the theme model formula to calculate the error value min:j.

[0096] S7. When the error exceeds the threshold, return to step S3 and continue iterating until the stopping condition for iteration is met. The iterative steps include steps S3 to S6.

[0097] S8. Stop iteration when the error is less than the threshold. The threshold is set manually. When the error is less than the threshold, the corresponding values ​​of the user interest topic matrix, the carrier content topic matrix, and the advertising topic matrix are optimal.

[0098] Because the correspondence between user identifiers, carrier content identifiers, and ad identifiers is established in advance using user interest topic matrices, carrier content topic matrices, and ad topic matrices, the degree of user recognition of the ads displayed on each carrier content is calculated. The user interest topic matrix reflects the user's preference (interest) for each topic, the carrier content topic matrix reflects the matching degree between carrier content and each topic, and the ad topic model reflects the similarity between ads and each topic. Therefore, the pre-established correspondence between user identifiers, carrier content identifiers, and ad identifiers reflects the relationship between user interests, carrier content topics, and ad topics, ensuring that recommended ads match user interests and achieving accurate ad recommendations. On this basis, the theme of the recommended ads also aligns with the carrier content theme, reducing the abruptness of ad insertion and improving the effectiveness of ad delivery.

[0099] The advertising recommendation method of this application will be explained below in the context of specific application scenarios.

[0100] Figure 9 This is a sequence diagram of an example of an advertising recommendation method. Figure 9 As shown, it includes the following steps:

[0101] S1: The server obtains the degree of approval of each user for the advertisements displayed on each carrier content.

[0102] Specifically, a user interest topic matrix, an advertising topic matrix, and a carrier content matrix are obtained, and the user interest topic matrix, advertising topic matrix, and carrier content matrix are used to calculate the degree of recognition of each user identifier for the advertisements displayed on each carrier content.

[0103] Specifically, historical behavioral data is acquired, and based on this data, user approval of the carrier content, user approval of the advertisement, and the similarity between the carrier content and the advertisement are determined. Based on the relationships among user approval of the carrier content, user interest topics, and carrier content topics; user approval of the advertisement, the advertisement topic, and user interest topics; and the similarity between carrier content and the advertisement, and the carrier content topic and the advertisement topic, a theme model is determined. Based on the theme model, a user interest topic matrix, an advertisement theme matrix, and a carrier content matrix are determined.

[0104] Among them, the relationship between user approval of the content, the theme of the content, and user interests is as follows: Figure 5 As shown, the relationship between user acceptance of advertisements, advertisement themes, and user interest themes is as follows: Figure 6 As shown, the similarity between the carrier content and the advertisement, and the relationship between the carrier content theme and the advertisement theme are as follows: Figure 7 As shown, Figure 8 This is a matrix decomposition model of a topic model for one embodiment.

[0105] Based on the topic matrix decomposition model, the optimization equation can be obtained:

[0106] min: J=λ||CU T V||2+α||YU T A||2+β||ZV T A||2+δ(||U||2+||V||2+||A||2)

[0107] Among them, U T V represents the user's level of liking for the content, U T A represents the user's level of liking for the advertisement, V T A represents the similarity between the carrier content and the advertisement, ||·||² squared loss function, λ, α, β, δ are the corresponding weight coefficients, ||U||² + ||V||² + ||A||² is the regularization term, C is the user's approval of the carrier content, Y is the user's approval of the advertisement, and Z is the similarity between the carrier content and the advertisement.

[0108] S2: Based on the most recognized advertisement, establish the correspondence between the user identifier, carrier content identifier, and advertisement identifier.

[0109] S3: The user terminal obtains the user's trigger operation on the carrier content, obtains the user identifier and the carrier content identifier, and sends them to the server.

[0110] S4: The server obtains the pre-established correspondence between user identifier, carrier content identifier, and advertisement identifier based on the user identifier and carrier content identifier.

[0111] S5: Determine the corresponding advertising identifier based on the correspondence.

[0112] S6: After pre-processing the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier, send them to the terminal corresponding to the user identifier.

[0113] In this embodiment, the default process is to insert advertisements at the beginning, middle, or end of the video and send them to the user terminal in the form of streaming media.

[0114] S7: The user terminal displays the carrier content and advertisements according to the preset processing method.

[0115] Correspondingly, advertisements are displayed when the video starts playing, during the middle of the video, or when the video ends.

[0116] The aforementioned advertising recommendation method, after obtaining user identifiers and carrier content identifiers, determines the corresponding advertising identifier based on a pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers. The corresponding advertising and carrier content are then pre-processed and sent to the terminal. Since the correspondence between user identifiers, carrier content identifiers, and advertising identifiers is established by pre-calculating the user's acceptance level of the advertisements displayed on each carrier content using a user interest theme model, a carrier content theme model, and an advertising theme model, and reflecting the user's preference (interest) for each theme, the carrier content theme model reflects the matching degree between the carrier content and each theme, and the advertising theme model reflects the matching degree between the advertisement and each theme, the pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers reflects the relationship between user interests, carrier content themes, and advertising themes. This ensures that the recommended advertisements match user interests, achieving accurate advertising recommendations. Furthermore, the theme of the recommended advertisements also aligns with the carrier content theme, reducing the abruptness of ad insertion and improving advertising effectiveness.

[0117] Figure 10 This is a structural block diagram of an advertising recommendation device in one embodiment. Figure 10 As shown, the advertising recommendation device includes: an identifier acquisition module 902, a search module 904, an advertising determination module 906, and a processing module 908.

[0118] The identifier acquisition module 902 is used to acquire the user identifier and the triggered carrier content identifier.

[0119] The lookup module 904 is used to obtain the pre-established correspondence between user identifier, carrier content identifier and advertisement identifier based on user identifier and carrier content identifier. The correspondence is pre-established based on the user's recognition of the advertisement displayed on each carrier content.

[0120] The advertisement identification module 906 is used to determine the corresponding advertisement identifier based on the correspondence relationship.

[0121] The processing module 908 is used to pre-process the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier and then send them to the terminal corresponding to the user identifier.

[0122] The aforementioned advertising recommendation device, after acquiring user identifiers and carrier content identifiers, determines the corresponding advertising identifier based on a pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers. It then processes the corresponding advertising and carrier content and sends them to the terminal. Since the correspondence between user identifiers, carrier content identifiers, and advertising identifiers is pre-established based on the user's level of acceptance of the advertisements displayed on each carrier content, it reflects the degree of matching between user interests and carrier content, carrier content and advertisements, and user interests and advertisements. Therefore, the pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers reflects the relationship between user interests, carrier content themes, and advertising themes, ensuring that recommended advertisements match user interests and achieving accurate advertising recommendations. Furthermore, the theme of the recommended advertisements also aligns with the carrier content theme, reducing the abruptness of advertisement insertion and improving advertising effectiveness.

[0123] Figure 11 This is a structural block diagram of an advertising recommendation device in another embodiment. Figure 11 As shown, the device also includes a calculation module 910 and a relationship establishment module 912.

[0124] The calculation module 910 is used to obtain the degree of approval of each user for the advertisements displayed on each carrier content.

[0125] The relationship establishment module 912 is used to establish a correspondence between user identifiers, carrier content identifiers, and advertisement identifiers based on the advertisements with the highest recognition.

[0126] Specifically, the calculation module 910 is used to obtain the user interest topic matrix, the advertising topic matrix, and the carrier content matrix, and to calculate the user's recognition of the advertisements displayed on each carrier content using the user interest topic matrix, the advertising topic matrix, and the carrier content matrix.

[0127] Specifically, the advertising recommendation device also includes a degree calculation module 914, a topic model determination module 916, and a decomposition module 918.

[0128] The degree calculation module 914 is used to acquire historical behavior data and determine the user's approval of the carrier content, the user's approval of the advertisement, and the similarity between the carrier content and the advertisement based on the historical behavior data.

[0129] The topic model determination module 916 is used to determine the topic model based on the relationship between user approval of carrier content, user interest topics and carrier content topics, user approval of advertisement, advertisement topics and user interest topics, and the similarity between carrier content and advertisement, carrier content topics and advertisement topics.

[0130] The decomposition module 918 is used to determine the user interest topic matrix, advertising topic matrix, and carrier content matrix based on the topic model.

[0131] Specifically, the relationship between user approval of the content, user interest topics, and the content topic is as follows: user approval of the content is positively correlated with the similarity between user interest topics and the content topic.

[0132] The relationship between user acceptance of an advertisement, the advertisement theme, and user interest themes includes: user acceptance of an advertisement is positively correlated with the similarity between the advertisement theme and the user interest themes.

[0133] The relationship between the similarity between the carrier content and the advertisement, the theme of the carrier content, and the theme of the advertisement includes: the degree of similarity between the carrier content and the advertisement, and the degree of similarity between the theme of the carrier content and the theme of the advertisement.

[0134] Specifically, historical behavioral data includes user viewing behavior data, user ad-clicking behavior data, and ad-clicking behavior data during the playback of carrier content; the degree calculation module is used to obtain viewing behavior data of each user identifier on the carrier content, and determine the user's acceptance of the carrier content based on the viewing behavior data; obtain user ad-clicking behavior data corresponding to the user identifier, and determine the user's actual acceptance of the ad based on the user ad-clicking behavior data; obtain ad-clicking behavior data during the playback of carrier content, and determine the similarity between the carrier content and the ad based on the ad-clicking behavior data during the playback of carrier content.

[0135] The aforementioned advertising recommendation device, through the pre-established correspondence between user identifiers, carrier content identifiers, and advertising identifiers, reflects the relationship between user interests, carrier content themes, and advertising themes. This ensures that recommended advertisements match user interests, achieving precise advertising recommendations and improving the accuracy of advertising delivery. Furthermore, the themes of the recommended advertisements also align with the carrier content themes, reducing the abruptness of ad insertion and enhancing user viewing satisfaction.

[0136] Figure 12 An internal structural diagram of a computer device in one embodiment is shown. Specifically, this computer device may be... Figure 1 Server 120 in the middle. For example... Figure 12 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program that, when executed by the processor, enables the processor to implement an advertising recommendation method. The internal memory may also store a computer program that, when executed by the processor, enables the processor to implement the advertising recommendation method.

[0137] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0138] In one embodiment, the advertising recommendation device provided in this application can be implemented as a computer program, which can be implemented in the form of, for example, Figure 12 It runs on the computer device shown. The computer device's memory can store the various program modules that make up the advertising recommendation device, for example, Figure 10 The diagram shows an identifier acquisition module, a search module, and a processing module. The computer program comprised of these modules causes the processor to execute the steps of the advertising recommendation methods described in the various embodiments of this application.

[0139] For example, Figure 12 The computer device shown can be used as follows Figure 10 The identifier acquisition module in the illustrated advertising recommendation device performs the steps of acquiring the user identifier and the triggered carrier content identifier. The computer device can perform the step of retrieving the pre-established correspondence between the user identifier, carrier content identifier, and advertisement identifier based on the user identifier and carrier content identifier through the lookup module. The computer device can perform the step of sending the advertisement corresponding to the advertisement identifier and the carrier content corresponding to the carrier content identifier, after pre-processing them, to the terminal corresponding to the user identifier through the processing module.

[0140] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the methods described in the above embodiments.

[0141] A storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the methods described in the above embodiments. Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0142] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0143] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An advertising recommendation method, comprising: Obtain the user identifier and the identifier of the triggered carrier content; Based on the user identifier and carrier content identifier, a pre-established correspondence between user identifiers, carrier content identifiers, and advertisement identifiers is obtained. This correspondence is pre-established based on the user identifier's approval rating of the advertisements displayed on each carrier content. The user identifier's approval rating of the advertisements displayed on each carrier content reflects the degree of matching between user interests and carrier content, carrier content and advertisements, and user interests and advertisements. Specifically, the user identifier's approval rating of the advertisements displayed on each carrier content is calculated using a user interest topic matrix, an advertisement topic matrix, and a carrier content topic matrix. The user interest topic matrix indicates the degree to which each user identifier's interests belong to each topic; the advertisement topic matrix indicates the degree to which each advertisement content belongs to each topic; and the carrier content topic matrix indicates the degree to which each carrier content belongs to each topic. The corresponding advertising identifier is determined based on the aforementioned correspondence. The advertisement corresponding to the advertisement identifier is interspersed at the beginning, middle or end of the carrier content corresponding to the carrier content identifier, and sent to the terminal corresponding to the user identifier in the form of streaming media for display. The methods for determining the user interest topic matrix, the advertising topic matrix, and the carrier content matrix include: Historical behavior data is acquired, and user approval of carrier content, user approval of advertisement, and similarity between carrier content and advertisement are determined based on the historical behavior data. The similarity between carrier content and advertisement is determined based on advertisement click behavior data when playing carrier content in the historical behavior data. Based on the relationship between the user's acceptance of the carrier content, the user's interest topics, and the carrier content topics, the relationship between the user's acceptance of the advertisement, the advertisement topics, and the user's interest topics, as well as the relationship between the similarity between the carrier content and the advertisement, the carrier content topics, and the advertisement topics, a topic model is determined. Determine the number of topics in the potential topic set, and initialize the values ​​of the user interest topic matrix, carrier content topic matrix, and advertising topic matrix; Based on the values ​​of the carrier content theme matrix and the advertising theme matrix, the derivative of the user interest theme matrix is ​​calculated to solve for and update the current value of the user interest theme matrix. Based on the values ​​of the current user interest topic matrix and the advertising topic matrix, the derivative of the carrier content topic matrix is ​​calculated to solve for and update the value of the current carrier content topic matrix. Based on the values ​​of the current carrier content topic matrix and the current user interest topic matrix, the derivative of the advertising topic matrix is ​​calculated to solve for and update the value of the current advertising topic matrix. Substitute the values ​​of the current advertising theme matrix, the current carrier content theme matrix, and the current user interest theme matrix into the theme model formula to calculate the error value of the theme model; When the error is greater than the threshold, return to the step of taking the derivative of the user interest topic matrix based on the value of the carrier content topic matrix and the value of the advertising topic matrix, and solving to update the value of the current user interest topic matrix, until the error is less than the threshold; The values ​​of the current user interest topic matrix, the current carrier content topic matrix, and the current advertising topic matrix corresponding to the error being less than the threshold are determined as the user interest topic matrix, the advertising topic matrix, and the carrier content matrix.

2. The method according to claim 1, characterized in that, Also includes: To obtain the level of user approval for the advertisements displayed on each platform; Based on the most recognized advertisements, establish the correspondence between the user identifier, the carrier content identifier, and the advertisement identifier.

3. The method according to claim 1, characterized in that, The relationship between user approval of the content, user interest topics, and the content topic includes: user approval of the content is positively correlated with the similarity between user interest topics and the content topic. The relationship between user acceptance of an advertisement, the advertisement theme, and user interest themes includes: user acceptance of an advertisement is positively correlated with the similarity between the advertisement theme and the user interest themes; The relationship between the similarity between the carrier content and the advertisement, the theme of the carrier content, and the theme of the advertisement includes: the similarity between the carrier content and the advertisement, and the similarity between the theme of the carrier content and the theme of the advertisement.

4. The method according to claim 1, characterized in that, The historical behavioral data includes user viewing behavior data and user ad clicking behavior data; the steps of acquiring user historical behavioral data and determining user approval of the carrier content and user approval of the ad based on the historical behavioral data include... Acquire viewing behavior data of each user identifier on the carrier content, and determine the user's approval of the carrier content based on the viewing behavior data; Obtain user click behavior data corresponding to user identifiers, and determine the user's approval of the advertisement based on the user click behavior data.

5. An advertising recommendation device, comprising: The module includes: identifier acquisition module, search module, advertisement identification module, processing module, degree calculation module, topic model identification module, and decomposition module. The identifier acquisition module is used to acquire the user identifier and the triggered carrier content identifier; The search module is used to obtain a pre-established correspondence between user identifiers, carrier content identifiers, and advertisement identifiers based on the user identifiers and carrier content identifiers. This correspondence is pre-established based on the user identifiers' approval rating of the advertisements displayed on each carrier content. The user identifiers' approval rating of the advertisements displayed on each carrier content reflects the degree of matching between user interests and carrier content, carrier content and advertisements, and user interests and advertisements. Specifically, the user identifiers' approval rating of the advertisements displayed on each carrier content is calculated using a user interest topic matrix, an advertisement topic matrix, and a carrier content topic matrix. The user interest topic matrix indicates the degree to which each user identifier's interests belong to each topic; the advertisement topic matrix indicates the degree to which each advertisement content belongs to each topic; and the carrier content topic matrix indicates the degree to which each carrier content belongs to each topic. The advertisement determination module is used to determine the corresponding advertisement identifier based on the correspondence relationship; The processing module is used to insert the advertisement corresponding to the advertisement identifier into the beginning, middle or end of the carrier content corresponding to the carrier content identifier, and send it to the terminal corresponding to the user identifier in the form of streaming media for display. The degree calculation module is used to acquire historical behavior data and determine the user's approval of the carrier content, the user's approval of the advertisement, and the similarity between the carrier content and the advertisement based on the historical behavior data. The similarity between the carrier content and the advertisement is determined based on the advertisement click behavior data when playing the carrier content in the historical behavior data. The topic model determination module is used to determine the topic model based on the relationship between the user's acceptance of the carrier content, the user's interest topic and the carrier content topic, the relationship between the user's acceptance of the advertisement, the advertisement topic and the user's interest topic, and the relationship between the similarity between the carrier content and the advertisement, the carrier content topic and the advertisement topic. The decomposition module is used to determine the number of topics in the potential topic set, initialize the values ​​of the user interest topic matrix, the carrier content topic matrix, and the advertising topic matrix; calculate the derivative of the user interest topic matrix with respect to the values ​​of the carrier content topic matrix and the advertising topic matrix, and solve to update the value of the current user interest topic matrix; calculate the derivative of the carrier content topic matrix with respect to the values ​​of the current user interest topic matrix and the advertising topic matrix, and solve to update the value of the current carrier content topic matrix; calculate the derivative of the advertising topic matrix with respect to the values ​​of the current carrier content topic matrix and the current user interest topic matrix, and solve to update the value of the current advertising topic matrix; substitute the values ​​of the current advertising topic matrix, the current carrier content topic matrix, and the current user interest topic matrix into the topic model formula to calculate the error value of the topic model; when the error is greater than a threshold, return to the step of calculating the derivative of the user interest topic matrix with respect to the values ​​of the carrier content topic matrix and the advertising topic matrix, and solving to update the value of the current user interest topic matrix, until the error is less than the threshold; The values ​​of the current user interest topic matrix, the current carrier content topic matrix, and the current advertising topic matrix corresponding to the error being less than the threshold are determined as the user interest topic matrix, the advertising topic matrix, and the carrier content matrix.

6. The apparatus according to claim 5, characterized in that, Also includes: The calculation module and the relationship establishment module; The calculation module is used to obtain the degree of approval of each user identifier for the advertisements displayed on each carrier content; The relationship establishment module is used to establish the correspondence between the user identifier, the carrier content identifier, and the advertisement identifier based on the advertisement with the highest recognition.

7. The apparatus according to claim 5, characterized in that, The relationship between user approval of the content, user interest topics, and the content topic includes: user approval of the content is positively correlated with the similarity between user interest topics and the content topic. The relationship between user acceptance of an advertisement, the advertisement theme, and user interest themes includes: user acceptance of an advertisement is positively correlated with the similarity between the advertisement theme and the user interest themes; The relationship between the similarity between the carrier content and the advertisement, the theme of the carrier content, and the theme of the advertisement includes: the similarity between the carrier content and the advertisement, and the similarity between the theme of the carrier content and the theme of the advertisement.

8. The apparatus according to claim 5, characterized in that, The historical behavior data includes user viewing behavior data and user clicking on advertisement behavior data; the degree calculation module is used to obtain the viewing behavior data of each user identifier on the carrier content, and determine the user's approval of the carrier content based on the viewing behavior data. Obtain user click behavior data corresponding to user identifiers, and determine the user's actual acceptance of the advertisement based on the user click behavior data.

9. A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as claimed in any one of claims 1 to 4.

10. A storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method as claimed in any one of claims 1 to 4.

Images