An interactive and precise advertising content push system and method

Abstract

The present invention discloses an interactive and precise advertising content push system and method in the field of advertising media, including generating non-directional scene fragments, obtaining currently browsed short video information, extracting characteristic elements from scenes in video information to generate non-directional scene fragments; extracting non-directional protagonist fragments, generating new simulation scenes, the new simulation scenes are composed of hybrid non-directional protagonist fragments and non-directional scene fragments, interactively implanting advertising products, leaving a joint gap at the joint of the simulation scene spliced ​​by the non-directional scene fragments, the shape of the gap is the same as the shape of the advertising product, and the user moves the advertising product into the gap to complete the puzzle. This technical solution promotes advertising in short videos or video websites, combines user interest settings during the promotion process to reduce user resistance to advertising promotion, and uses the interactive mode in the advertising animation to improve the audience's memory of the advertisement.

Description

Technical Field

[0001] The present invention belongs to the field of advertising media, and specifically relates to an interactive and precise advertising content push system and method. Background Art

[0002] With the continuous increase in advertising, how to accurately target advertisers and thus expand advertising benefits is the focus of current thinking and research. In order to improve the accuracy of advertising, with the development of modern intelligent technology, mining algorithms, recommendation algorithms, etc. are widely used in advertising to improve the accuracy of advertising. Currently, advertising includes user-based targeted delivery and behavior-based delivery. The former is mainly based on the content browsed by the user, and the latter is based on the user's interests.

[0003] Taking actual application cases as an example, WeChat merchants mainly use targeted delivery based on users, while live broadcast platforms use user interests as the delivery criteria. However, in current precision advertising, most advertisements only consider the attractiveness of the product itself and ignore user interests. In other words, while combining user targeted delivery, advertising should also be more integrated with user behavior. Summary of the invention

[0004] In order to solve the problem of ignoring customer interests in product delivery, the purpose of the present invention is to provide an interactive and precise method for pushing advertising content in combination with customer interests.

[0005] In order to achieve the above-mentioned purpose, the technical scheme of the present invention is as follows: A method for interactive and accurate push of advertising content comprises:

[0006] Generate undirected scene fragments, obtain the currently browsed short video information, extract feature elements of the scene in the video information to generate undirected scene fragments;

[0007] Extract non-directional protagonist fragments. When the protagonist is a person, extract the protagonist's clothes in the scene. When the protagonist is an animal, extract the animal fragments.

[0008] Action information collection: classify and collect human action information and animal action information;

[0009] Advertisement product fit evaluation: select the ad product with the highest fit from the user's current browsing history, and then evaluate the adjacent non-directional protagonist fragments and non-directional scene fragments based on the attributes of the ad product;

[0010] Generate a new simulation scene. The new simulation scene is composed of a combination of hybrid non-directional protagonist fragments and non-directional scene fragments. At this time, the new simulation scene is judged. When the protagonist fragment is a person, an AI face or spokesperson face is generated, and then the action information of the current protagonist fragment is matched. If the protagonist fragment is an animal, the action information is directly matched;

[0011] The interactive placement of advertising products leaves gaps at the joints of the simulated scene made up of non-directional scene fragments. The shape of the gap is the same as the shape of the advertising product. At this time, the user moves the advertising product into the gap to complete the puzzle.

[0012] The following beneficial effects are achieved after adopting the above scheme: This technical scheme promotes advertisements in short videos or video websites, combines user interest settings during the promotion process to reduce user resistance to advertising promotion, and uses the interactive mode in advertising animations to enhance the audience's memory of the advertisements.

[0013] This technical solution also uses non-directional scene fragments and non-directional protagonist fragments to utilize the user's interest points to generate a sense of familiarity and interest in the user, thereby generating a sense of familiarity and recollection, and increasing the degree of attention during the recollection process. When the advertisement is produced, the user who swipes up the short video will often think that it is a newly pushed interest video, thereby extending the viewing time, which is convenient for the implantation and active absorption of advertising concepts.

[0014] Furthermore, when generating a new simulation scenario, the undirected protagonist fragments are merged into the data set, the code of the data set forms a first Möbius ring set, and the undirected scene fragments form a second Möbius set, wherein the corresponding key for selecting the first Möbius set and the second Möbius set is product fit.

[0015] Beneficial effects: When this technical solution is used, the fragments are collected into two Möbius sets, thereby achieving linear operation and reducing the pressure of calculation. At the same time, since the Möbius set has no specific surface, it can combine the non-directional protagonist fragments and the non-directional scene fragments as much as possible.

[0016] Furthermore, it also includes the selection of preference settings. The rules for capturing undirected protagonist fragments and undirected scene fragments are to form weights according to browsing time, where the browsing time is T=(t1, t2, ..., tn). Due to the different browsing times, the preference weight is W=(w1, w2, wn). The size of Wi is positively correlated with the importance of Ti, and advertising products are pushed according to this preference weight.

[0017] Beneficial effects: Undirected protagonist fragments and undirected scene fragments capture user video records and encode and store the captured undirected fragments, deleting redundant tags in the encoding, such as references to <linlk>, (<!DOCTYPE>), css / js, etc., and finally retaining important tags of encoding features, such as <head>, <title>, etc.

[0018] Execute the division of the code, that is, divide the code according to the time weight T, so that it can be transformed into a tag combined with the user's browsing time, and then collide with the preference weight and the characteristics of the embedded advertisement. If a preference scenario fits the usage scenario of the advertising product, it means that the generated simulation scenario is valid, and then continue to pursue other code strings. Then use statistical methods to solve the weight values ​​corresponding to different labels. Theoretically, the weight values ​​of different labels are relative concepts and are only comparable to each other.

[0019] Furthermore, the collaborative recommendation algorithm of RBF is used in the process of pushing advertising products. This not only solves the problem of sparsity of preference setting data in traditional collaborative promotion, but also improves the efficiency of advertising push through user interest models.

[0020] Beneficial effects: For the problem of sparse data in preference settings, the RBF neural network algorithm is used to optimize it, and a new type of user interest point can be established. Then the Pearson similarity algorithm is used to solve the similarity between the user rating vector and the target user to perform rating prediction, and then several characteristic words of the advertisements are extracted.

[0021] Furthermore, it also includes sending bounties, where a single bounty is issued each time a user completes a puzzle, and the bounty includes currency or virtual gifts.

[0022] Furthermore, an interactive and accurate advertising content push system is provided, the system comprising

[0023] The non-directional scene extraction module is used to obtain environmental information and generate scene fragments according to the environmental information;

[0024] The module for extracting non-directional protagonists performs two-step judgments: judging whether the current collected protagonist is a person or an animal. If the protagonist is a person, the protagonist's clothes in the scene are extracted; if the protagonist is an animal, the animal fragments are extracted.

[0025] The action collection module collects the human action information and the animal action information by classification and matches the corresponding protagonist in the non-directional protagonist extraction module;

[0026] The advertising product evaluation module selects the advertising product with the highest user fit from the advertising products to be promoted, and then evaluates the adjacent non-directional protagonist fragments and non-directional scene fragments based on the attributes of the advertising product;

[0027] The simulation scene construction module combines the hybrid non-directional protagonist fragments and non-directional scene fragments into a new scene according to the directional characteristics of the product;

[0028] The interactive implantation module leaves a gap at the joint of the simulated scene, and the shape of the gap is the same as the shape of the advertised product. DETAILED DESCRIPTION

[0029] The following is further described in detail through specific implementation methods:

[0030] Embodiment 1: This embodiment includes a method for interactive and accurate advertising content push: comprising:

[0031] Generate undirected scene fragments, obtain the currently browsed short video information, extract feature elements of the scene in the video information to generate undirected scene fragments;

[0032] Extract non-directional protagonist fragments. When the protagonist is a person, extract the protagonist's clothes in the scene. When the protagonist is an animal, extract the animal fragments.

[0033] When generating a new simulation scene, the undirected protagonist fragments are imported into the data set, the code of the data set forms the first Möbius ring set, and the undirected scene fragments form the second Möbius set, where the key to the selection of the first Möbius set and the second Möbius set is the product fit. When in use, the fragments are collected in two Möbius sets to achieve linear operation and reduce the pressure of calculation. At the same time, because the Möbius set has no specific surface, the undirected protagonist fragments and the undirected scene fragments are combined as much as possible.

[0034] Action information collection: classify and collect human action information and animal action information;

[0035] Advertisement product fit evaluation: select the ad product with the highest fit from the user's current browsing history, and then evaluate the adjacent non-directional protagonist fragments and non-directional scene fragments based on the attributes of the ad product;

[0036] Generate a new simulation scene. The new simulation scene is composed of a combination of hybrid non-directional protagonist fragments and non-directional scene fragments. At this time, the new simulation scene is judged. When the protagonist fragment is a person, an AI face or spokesperson face is generated, and then the action information of the current protagonist fragment is matched. If the protagonist fragment is an animal, the action information is directly matched;

[0037] It also includes the selection of preference settings. The rules for capturing undirected protagonist fragments and undirected scene fragments are to form weights based on browsing time, where browsing time is T=(t1, t2, …, tn). Due to the different browsing times, the preference weight is W=(w1, w2, wn). The size of Wi is positively correlated with the importance of Ti, and advertising products are pushed according to this preference weight.

[0038] Undirected protagonist fragments and undirected scene fragments capture the user's video records and encode and store the captured undirected fragments, deleting redundant tags in the encoding, such as references to <linlk>, (<!DOCTYPE>), css / js, etc., and finally retaining important tags of encoding features, such as <head>, <title>, etc.

[0039] Execute the division of the code, that is, divide the code according to the time weight T, so that it can be transformed into a tag combined with the user's browsing time, and then collide with the preference weight and the characteristics of the embedded advertisement. If a preference scenario fits the usage scenario of the advertising product, it means that the generated simulation scenario is valid, and then continue to pursue other code strings. Then use statistical methods to solve the weight values ​​corresponding to different labels. Theoretically, the weight values ​​of different labels are relative concepts and are only comparable to each other.

[0040] For the problem of sparse data in preference settings, the RBF neural network algorithm is used to optimize it, and a new type of user interest point can be established. Then the Pearson similarity algorithm is used to solve the similarity between the user rating vector and the target user to perform rating prediction, and then several characteristic words of the advertisements are extracted.

[0041] The interactive placement of advertising products leaves gaps at the joints of the simulated scene made up of non-directional scene fragments. The shape of the gap is the same as the shape of the advertising product. At this time, the user moves the advertising product into the gap to complete the puzzle.

[0042] In the process of pushing advertising products, the collaborative recommendation algorithm of RBF is used for recommendation. This not only solves the problem of sparsity of preference setting data in traditional collaborative promotion, but also improves the efficiency of advertising push through user interest model.

[0043] It also includes bounty sending, where a single bounty is issued each time a user completes a puzzle. The bounty includes currency or virtual gifts.

[0044] This technical solution promotes advertisements on short videos or video websites, combines user interest settings during the promotion process to reduce user resistance to advertisement promotion, and uses the interactive mode in the advertisement animation to improve the audience's memory of the advertisement.

[0045] This technical solution also uses non-directional scene fragments and non-directional protagonist fragments to utilize the user's interest points to generate a sense of familiarity and interest in the user, thereby generating a sense of familiarity and recollection, and increasing the degree of attention during the recollection process. When the advertisement is produced, the user who swipes up the short video will often think that it is a newly pushed interest video, thereby extending the viewing time, which is convenient for the implantation and active absorption of advertising concepts.

[0046] Embodiment 2: This embodiment differs from the above embodiment in that it discloses an interactive and precise advertising content push system. The system includes:

[0047] The non-directional scene extraction module is used to obtain environmental information and generate scene fragments according to the environmental information;

[0048] The module for extracting non-directional protagonists performs two-step judgments: judging whether the current collected protagonist is a person or an animal. If the protagonist is a person, the protagonist's clothes in the scene are extracted; if the protagonist is an animal, the animal fragments are extracted.

[0049] The action collection module collects the human action information and the animal action information by classification and matches the corresponding protagonist in the non-directional protagonist extraction module;

[0050] The advertising product evaluation module selects the advertising product with the highest user fit from the advertising products to be promoted, and then evaluates the adjacent non-directional protagonist fragments and non-directional scene fragments based on the attributes of the advertising product;

[0051] The simulation scene construction module combines the hybrid non-directional protagonist fragments and non-directional scene fragments into a new scene according to the directional characteristics of the product;

[0052] The interactive implantation module leaves a gap at the joint of the simulated scene, and the shape of the gap is the same as the shape of the advertised product.

[0053] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0054] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is not described in detail here. The ordinary technicians in the relevant field know all the common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent. The protection scope required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A method for interactive and accurate advertising content push, characterized by: include Generate undirected scene fragments, obtain the currently browsed short video information, extract feature elements of the scene in the video information to generate undirected scene fragments; Extract fragments of the non-directional protagonist. When the protagonist is a person, extract the protagonist's clothes in the scene. When the protagonist is an animal, extract the animal fragments. Action information collection: classify and collect human action information and animal action information; Advertisement product fit evaluation: select the ad product with the highest fit from the user's current browsing history, and then evaluate the adjacent non-directional protagonist fragments and non-directional scene fragments based on the attributes of the ad product; Generate a new simulation scene. The new simulation scene is composed of a combination of hybrid non-directional protagonist fragments and non-directional scene fragments. At this time, the new simulation scene is judged. When the protagonist fragment is a person, an AI face is generated, and then the action information of the current protagonist fragment is matched. If the protagonist fragment is an animal, the action information is directly matched; The interactive placement of advertising products leaves gaps at the joints of the simulated scene made up of non-directional scene fragments. The shape of the gap is the same as the shape of the advertising product. At this time, the user moves the advertising product into the gap to complete the puzzle.

2. The method for interactive and accurate advertising content push according to claim 1, characterized in that: When generating a new simulation scenario, the undirected protagonist fragments are merged into the data set, the code of the data set forms the first Möbius ring set, and the undirected scene fragments form the second Möbius set, wherein the key corresponding to the selection of the first Möbius set and the second Möbius set is the product fit.

3. The method for interactive and accurate advertising content push according to claim 2, characterized in that: It also includes the selection of preference settings. The rules for capturing undirected protagonist fragments and undirected scene fragments are to form weights based on browsing time, where browsing time is T=(t1, t2, …, tn). Due to the different browsing times, the preference weight is W=(w1, w2, wn). The size of Wi is positively correlated with the importance of Ti, and advertising products are pushed according to this preference weight.

4. The method for interactive and accurate advertising content push according to claim 3, characterized in that: In the process of pushing advertising products, the collaborative recommendation algorithm of RBF is used for recommendation. This not only solves the problem of sparsity of preference setting data in traditional collaborative promotion, but also improves the efficiency of advertising push through user interest model.

5. The method for interactive and accurate advertising content push according to claim 4, characterized in that: It also includes bounty sending, where a single bounty is issued each time a user completes a puzzle. The bounty includes currency or virtual gifts.

6. The method for interactive and accurate advertising content push according to claim 5, characterized in that: It is executed by the interactive and precise advertising content push system, which includes The non-directional scene extraction module is used to obtain environmental information and generate scene fragments according to the environmental information; The module for extracting non-directional protagonists performs two-step judgments: judging whether the current collected protagonist is a person or an animal. If the protagonist is a person, the protagonist's clothes in the scene are extracted; if the protagonist is an animal, the animal fragments are extracted. The action collection module collects the human action information and the animal action information by classification and matches the corresponding protagonist in the non-directional protagonist extraction module; The advertising product evaluation module selects the advertising product with the highest user fit from the advertising products to be promoted, and then evaluates the adjacent non-directional protagonist fragments and non-directional scene fragments based on the attributes of the advertising product; The simulation scene construction module combines the hybrid non-directional protagonist fragments and non-directional scene fragments into a new scene according to the directional characteristics of the product; The interactive implantation module leaves a gap at the joint of the simulated scene, and the shape of the gap is the same as the shape of the advertised product.