Omnibearing intelligent advertisement management system

Abstract

The invention relates to an omnibearing intelligent advertisement management system, which belongs to the technical field of advertisement management, and comprises an advertisement input module, a preprocessing module, an advertisement classification decision module, a label module, a comparison module and a delivery module, the output end of the preprocessing module is in signal connection with the input end of the advertisement classification decision module, the output end of the advertisement classification decision module is in signal connection with the input end of the label module, and the output end of the label module is in signal connection with the input end of the comparison module. The output end of the comparison module is in signal connection with the input end of the putting module. According to the omni-directional intelligent advertisement management system, through cooperation of the preprocessing module and the advertisement classification decision-making module, advertisements are divided into image-text advertisements and video advertisements, the advertisements are further classified in detail according to life scenes, and classification accuracy can be improved through cooperation of the comparison module and manual review.

Description

Technical Field

[0001] The present invention relates to the technical field of advertising management, and in particular to an all-round intelligent advertising management system. Background Art

[0002] The advertising management system is an automated management tool that emerged with the development of Internet advertising. It aims to help website owners and advertisers efficiently complete advertising delivery, resource allocation and data statistics, greatly improve delivery efficiency and accuracy, and reduce labor costs.

[0003] For example, a Chinese patent (publication number: CN110443651B) discloses a media advertising classification management system, comprising a client, wherein the output end of the client is unidirectionally electrically connected to a login module and a registration module, the output end of the registration module is unidirectionally electrically connected to the input end of the login module, the output end of the login module is unidirectionally electrically connected to an information entry unit, and the output end of the information entry unit is bidirectionally electrically connected to a detection unit. By configuring an information entry unit, a detection unit, a customer information database, a payment unit, an advertisement processing unit, a status temporary storage unit, an information monitoring unit, and an information retrieval unit, the present invention provides a management system that allows payment before advertisements are uploaded and can record the personal information of some customers with undesirable behavior. This solves the problem that traditional advertisement classification systems use a post-payment method, which is prone to vulnerabilities such as customers being able to upload advertisements without paying, and the system is unable to record the personal information of undesirable uploaders.

[0004] However, the advertising classification effect of this system is not good enough. The system only classifies advertisements through the advertising content classification module in the advertising processing unit. However, advertisements are diverse in form and rich in content. It is difficult to classify advertisements in detail using only one advertising content splitting module. Therefore, a comprehensive advertising management system is proposed to solve the above problems. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the present invention provides an all-round intelligent advertising management system with advantages such as good advertising classification effect, which solves the problem that the existing advertising management system has poor classification effect.

[0006] To achieve the above-mentioned objectives, the present invention provides the following technical solutions: an all-round intelligent advertising management system, comprising an advertising input module, a preprocessing module, an advertising classification decision module, a labeling module, a comparison module and a delivery module, wherein the advertising input module is connected to the preprocessing module via a two-way network communication signal connection, the output end of the preprocessing module is connected to the input end of the advertising classification decision module via a signal connection, the output end of the advertising classification decision module is connected to the input end of the labeling module via a signal connection, the output end of the labeling module is connected to the input end of the comparison module via a signal connection, the output end of the comparison module is connected to the input end of the delivery module via a signal connection, the delivery module is connected to the advertising input module via a two-way network communication signal connection, the comparison module is connected to the advertising input module via a two-way network communication signal connection, the output end of the delivery module is connected to the input end of the labeling module via a signal connection, and the output end of the preprocessing module is connected to the input end of the comparison module via a signal connection; The pre-processing module includes a preliminary classification unit, a metadata parsing unit and a dynamic adjustment threshold; The advertisement classification decision module includes a video advertisement processing unit and a graphic and text advertisement processing unit.

[0007] Furthermore, the advertisement input module includes a unified import unit and an autonomous upload unit, and the unified import unit includes a data interface and an operation console.

[0008] Furthermore, the operating console includes a display screen, a keyboard, a mouse and a chassis, and the data interface is connected to the chassis via a two-way signal of network communication.

[0009] Furthermore, the video advertising processing unit includes a D-CNN model and a Transformer model, and the graphic advertising processing unit includes a ResNet model, a BERT model and a cross-modal attention mechanism.

[0010] Furthermore, the label module includes a brand label, a keyword label and a discount strength label, and the comparison module includes a comparison unit, a judgment unit and an alarm unit.

[0011] Furthermore, the output end of the preliminary classification unit is signal-connected to the input end of the alarm unit, and the output end of the alarm unit is signal-connected to the input end of the operating console.

[0012] Furthermore, the delivery module includes delivery suggestions, consent judgment and a storage unit, and the storage unit includes an internal database and a network database.

[0013] Furthermore, the output end of the network database is connected to the input end of the label module by signal, and the operating console is connected to the delivery module by two-way signal communication via network.

[0014] Furthermore, the output end of the comparison unit is signal-connected to the input end of the judgment unit, and the output end of the judgment unit is signal-connected to the input end of the alarm unit.

[0015] Furthermore, the output end of the metadata parsing unit is signal-connected to the input end of the video advertisement processing unit, and the output end of the metadata parsing unit is signal-connected to the input end of the graphic advertisement processing unit.

[0016] Compared with the existing technology, the present invention provides a comprehensive intelligent advertising management system with the following beneficial effects: 1. This all-round intelligent advertising management system, through the coordination of the pre-processing module and the advertising classification decision module, can carry out detailed classification of advertisements, dividing advertisements into graphic advertisements and video advertisements, and then further classify advertisements according to life scenarios, making it easier for subsequent users to retrieve advertisements. The accuracy of classification can be improved through the comparison module and manual review.

[0017] 2. This all-round intelligent advertising management system uses the label module to label advertisements with key feature labels such as brand labels, keyword labels and discount strength labels, which can facilitate the classification of advertisements. When combined with the delivery module and the advertisement input module, it can deliver advertisements more accurately. Moreover, through the coordination of delivery suggestions and consent judgments, it can play an auxiliary delivery role and provide auxiliary ideas to help users deliver advertisements.

[0018] 3. This all-round intelligent advertising management system uses the pre-processing module to preliminarily classify advertisements, and then uses the advertising classification decision module to classify advertisements more finely, which can effectively reduce the processing burden of each step of the program and improve classification efficiency. Compared with purely manual classification of advertisements, it can significantly reduce labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a system diagram of the present invention; Figure 2 A system diagram of the operating console of the present invention; Figure 3 This is a flowchart of the operation of the preliminary classification unit of the present invention; Figure 4 is a system diagram of the storage unit of the present invention; Figure 5 A flowchart of advertisement delivery according to the present invention; Figure 6 Flowchart of the system of the present invention.

[0020] In the figure: 1 advertising input module, 101 unified import unit, 1011 data interface, 1012 operation console, 102 autonomous upload unit, 2 preprocessing module, 201 preliminary classification unit, 202 metadata parsing unit, 203 dynamic adjustment threshold, 3 advertising classification decision module, 301 video advertising processing unit, 302 graphic advertising processing unit, 4 label module, 401 brand label, 402 keyword label, 403 discount intensity label, 5 comparison module, 501 comparison unit, 502 judgment unit, 503 alarm unit, 6 delivery module, 601 delivery suggestion, 602 consent judgment, 603 storage unit, 6031 internal database, 6032 network database. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0022] See also Figure 1 、 Figure 2 、 Figure 3 and Figure 6 In this embodiment, a comprehensive intelligent advertising management system includes an advertising input module 1, a preprocessing module 2, an advertising classification decision module 3, a label module 4, a comparison module 5 and a delivery module 6. The advertising input module 1 is connected to the preprocessing module 2 via a two-way network communication signal. The preprocessing module 2 is used to classify advertisements into video advertisements and graphic advertisements, which facilitates more accurate classification in the future. The output end of the preprocessing module 2 is signal-connected to the input end of the advertising classification decision module 3, and the output end of the advertising classification decision module 3 is signal-connected to the input end of the label module 4. The advertising classification decision module 3 can classify advertisements according to the actual usage scenarios of the advertisements, further improving the accuracy of classification. The output end of the label module 4 is signal-connected to the input end of the comparison module 5, and the output end of the comparison module 5 is signal-connected to the input end of the delivery module 6. The delivery module 6 is connected to the advertising input module 1 via a two-way network communication signal. The comparison module 5 is connected to the advertising input module 1 via a two-way network communication signal. The output end of the delivery module 6 is signal-connected to the input end of the label module 4, and the output end of the preprocessing module 2 is signal-connected to the input end of the comparison module 5.

[0023] In addition, the advertising input module 1 includes a unified import unit 101 and an autonomous upload unit 102. The unified import unit 101 includes a data interface 1011 and an operating console 1012. The staff can insert a data storage device such as a USB into the data interface 1011 to transmit data with the operating console 1012. The operating console 1012 includes a display screen, a keyboard, a mouse and a chassis. The data interface 1011 is connected to the chassis through a two-way signal network communication. The display screen, keyboard and mouse can improve the efficiency of the staff.

[0024] In addition, the pre-processing module 2 includes a preliminary classification unit 201, a metadata parsing unit 202 and a dynamic adjustment threshold 203. The preliminary classification unit 201 mainly classifies advertisements by judging the file format and file size, and divides advertisements into video advertisements and graphic advertisements. The metadata parsing unit 202 implements the classification of advertising materials. In essence, it makes collaborative decisions by deconstructing the structural fingerprint and dynamic characteristics of digital resources. The unit deeply scans the resource request pattern and content composition of the underlying advertisement. For graphic advertisements, the system will capture high-density static element features including a large number of discretely loaded image resources, densely distributed text modules and low Continuous resource request behavior, and for video ads, continuous streaming features, video duration markers, and extremely low text visual coupling are identified. This resource structure difference constitutes the skeleton of the preliminary classification. Through the rigid rules of file format and file size, the system can quickly complete the preliminary classification, greatly reducing the cost of manual review, and the metadata parsing unit 202 provides secondary verification, which can effectively improve the accuracy of the preliminary classification of ads. At the same time, when the metadata parsing unit 202 repeatedly judges that it is inconsistent with the preliminary classification unit 201, the user can also adjust the file size threshold by dynamically adjusting the threshold 203 to further improve the accuracy of the classification.

[0025] It can be understood that the advertising classification decision module 3 includes a video advertising processing unit 301 and a graphic advertising processing unit 302. The video advertising processing unit 301 includes a 3D-CNN model and a Transformer model. The 3D-CNN model simultaneously captures the spatial features of the video such as brand logo, product appearance and temporal dynamics such as action continuity through a three-dimensional convolution kernel. When extracting brand features, the 3D-CNN can identify the spatiotemporal distribution of brand logos that appear repeatedly in advertisements. For discount information, the model can combine text and superimpose animations such as "50% off" to generate discount information. OFF" barrage and sound prompts such as promotional narration, and comprehensively judge the discount strength. When analyzing advertising copy, the Transformer model can associate text in video frames such as subtitles and labels with voice content to extract core keywords. 3D-CNN is more efficient in action recognition of short advertising clips such as product display animations and local feature extraction. Transformer is better at semantic coherence analysis of long videos such as brand implantation in storylines. The combination of the two can significantly improve the classification effect of video advertisements. The graphic advertising processing unit 302 includes a ResNet-50 model, a BERT model, and a cross-modal attention mechanism. The shallow network of the ResNet-50 model can capture detailed textures such as brand logos and product appearances, while the deep network recognizes overall structure and high-level semantics such as the abstract expression of brand logos. The BERT model The model can perform deep semantic analysis of advertising texts, accurately identify brand names, product attribute keywords such as "natural" and "limited edition", and descriptions of discount levels such as "30% off" and "buy one get one free", and can understand the implicit meaning of promotional terms in context. The cross-modal attention mechanism can further break through the barriers between images and texts. The three work together to not only independently extract brand, keyword and discount information, but also enhance overall understanding through multimodal fusion. The output end of the metadata parsing unit 202 is connected to the input end signal of the video advertising processing unit 301, and the output end of the metadata parsing unit 202 is connected to the input end signal of the graphic advertising processing unit 302. After being processed by the metadata parsing unit 202 and confirmed to be consistent with the judgment result of the preliminary classification unit 201, the advertisements will be divided into two major categories: video advertisements and graphic advertising, and then sent to the video advertising processing unit 301 and the graphic advertising processing unit 302 respectively for more detailed classification.

[0026] In this embodiment, the metadata parsing unit 202 and the preliminary classification unit 201 are first used to determine the size and format of the advertising file, so as to classify the advertisements into video advertisements and graphic advertisements. The 3D-CNN model and the Transformer model are used to process the video advertisements and extract the brands, keywords and discount strength of the video advertisements. Then, the ResNet-50 model, the BERT model and the cross-modal attention mechanism are used to process the graphic advertisements and extract the brands, keywords and discount strength of the graphic advertisements, so as to better match the label module 4 to label the advertisements with corresponding labels.

[0027] Please refer again Figure 1 、 Figure 4 、 Figure 5 and Figure 6 In order to facilitate the accurate delivery of subsequent advertisements, in this embodiment, the label module 4 includes a brand label 401, a keyword label 402 and a discount strength label 403, and the comparison module 5 includes a comparison unit 501, a judgment unit 502 and an alarm unit 503. The output end of the preliminary classification unit 201 is connected to the input end signal of the alarm unit 503, and the output end of the alarm unit 503 is connected to the input end signal of the operation console 1012. When the preliminary classification unit 201 encounters any one of the two judgments of judging whether the advertisement file format and the advertisement file size exceed the threshold value, the alarm unit 503 will notify the operation console 1012, thereby manually distinguishing the advertisement, which can improve the effect of preliminary classification of the advertisement. By using the label module 4 to label the advertisement with labels with specific information such as brand labels 401, keyword labels 402 and discount strength labels 403, it can not only facilitate the subsequent classification and storage of the advertisements, but also improve the connection effect of the advertisements.

[0028] It can be known that the output end of the comparison unit 501 is connected to the input end signal of the judgment unit 502, and the output end of the judgment unit 502 is connected to the input end signal of the alarm unit 503. After the label module 4 labels the advertisement, the label and the original label are compared and judged through the comparison module 5. When it is judged that the two are the same or similar, they will be stored in the corresponding position of the internal database 6031.

[0029] Furthermore, the delivery module 6 includes a delivery suggestion 601, an agreement judgment 602 and a storage unit 603. The storage unit 603 includes an internal database 6031 and a network database 6032. The internal database 6031 is used to store the advertising data transmitted through the comparison module 5, and the network database 6032 is mainly used to collect common advertising tags on the Internet to provide a basis for labeling by the label module 4. The output end of the network database 6032 is connected to the input end signal of the label module 4. The network database 6032 can provide stable network data for the label module 4, thereby helping the label module 4 to label key tags for the advertisements. The operating console 1012 is connected to the delivery module 6 through a two-way signal of network communication.

[0030] In this embodiment, when it is necessary to place an advertisement, since the advertisements have been classified and labeled at the beginning, the placement module 6 can generate a placement suggestion 601 based on the label, such as placing beauty and cosmetics graphic advertisements on the Xiaohongshu platform, etc., which can provide certain assistance to users in placing advertisements and further improve the accuracy of advertisement placement. However, the placement suggestion 601 mainly plays an auxiliary role, and ultimately the user needs to make the decision, which can retain the user's sufficient self-sufficiency rights and more effectively assist the user in placing advertisements.

[0031] It is understandable that this system performs preliminary classification of advertisements through the pre-processing module 2, and then classifies advertisements according to specific usage scenarios through the advertisement classification decision module 3. Video advertisements capture dynamic brand logos and action details through the spatiotemporal modeling capabilities of the 3D-CNN model, supplemented by the long sequence semantic analysis of the Transformer model to accurately extract keywords such as "limited time special offer" and discount strength such as "buy one get one free" in video voice or subtitles, avoiding information omissions of traditional single models, while graphic advertisements use the deep visual features of ResNet-50 to identify brands in complex backgrounds. Brand Logo, BERT analyzes the implicit promotional semantics in the advertising slogan, and then through the cross-modal attention mechanism, dynamically aligns the image and text semantics, such as associating the "half price" text with the price tag image, solving the problem of image and text separation, and improving the accuracy of extracting brand, keyword and discount information. In addition, by cooperating with the preprocessing module 2, the advertising classification decision module 3, the label module 4, the comparison module 5 and the delivery module 6, while processing the advertisements efficiently and automatically, the advertisements that are difficult to process automatically are uploaded to manual processing through the alarm unit 503 and the operation console 1012, which can further improve the accuracy of advertising classification and delivery.

[0032] The working principle of the above embodiment is: (1) Users can upload advertisements in two ways: by accessing the self-uploading unit 102 through an electronic device to upload advertisements, or by delivering the advertisement file to the corresponding staff, who then upload the advertisements through the unified import unit 101. Both uploading methods require uploading advertisement data and original tags related to the advertisements, and the original tags are filled in by the user.

[0033] (2) The successfully uploaded advertisement will be sent to the preliminary classification unit 201. The preliminary classification unit 201 will judge the file format and the file size at the same time. When the file format is one of the multiple video file formats and the file size is greater than the set threshold, the advertisement will be judged as a video advertisement. When the file format is not one of the multiple video file formats and the file size is not greater than the set threshold, the advertisement will be judged as a graphic advertisement. When one of the judgment of the file format and the judgment of the file size is judged as no, the alarm unit 503 will be notified and the operation console 1012 will be informed. At this time, the staff can manually judge the advertisement. The advertisement classified by the preliminary classification unit 201 will enter the metadata parsing unit 202. When the judgment result of the metadata parsing unit 202 is inconsistent with the judgment result of the preliminary classification unit 201, the alarm unit 503 will be used to inform the operation console 1012 for manual judgment.

[0034] (3) After the pre-processing of the advertisement is completed, the video advertisement will be sent to the video advertisement processing unit 301, and the graphic advertisement will be sent to the graphic advertisement processing unit 302. By cooperating with the 3D-CNN model and the Transformer model in the video advertisement processing unit 301, keywords such as brand, keyword and discount strength in the video advertisement can be extracted. By cooperating with the ResNet-50 model, the BERT model and the cross-modal attention mechanism in the graphic advertisement processing unit 302 to process the graphic advertisement, the brand, keyword and discount strength in the graphic advertisement can be smoothly extracted, so as to better cooperate with the label module 4 to label the advertisement accordingly.

[0035] (4) After the advertisement is labeled with the corresponding label, the label applied by the label module 4 is compared with the original label through the comparison unit 501, and the judgment unit 502 judges whether the two labels are the same or similar. If they are the same or similar, the classified advertisement will be stored in the corresponding position of the internal database 6031 according to the label. If the judgment unit 502 judges that it is not, the alarm unit 503 will be used to inform the operation console 1012 to modify the label. The modified label will be directly stored in the corresponding position of the internal database 6031 according to the label type.

[0036] (5) When it is necessary to place an advertisement, the staff will retrieve the corresponding advertisement from the internal database 6031 through the operation console 1012. At this time, the placement module 6 will generate a placement suggestion 601 based on the type of advertisement. The user can directly judge the placement suggestion 601. If the user agrees with the content of the placement suggestion 601, the placement can be carried out. If the user disagrees, the user can ignore the placement suggestion 601 and enter a new placement suggestion 601 through the operation console 1012. After clicking on the consent judgment 602, the placement can be carried out.

[0037] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0038] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A comprehensive intelligent advertising management system, characterized by: The invention comprises an advertisement input module (1), a pre-processing module (2), an advertisement classification decision module (3), a label module (4), a comparison module (5) and a delivery module (6), wherein the advertisement input module (1) is connected to the pre-processing module (2) via a network communication bidirectional signal connection, the output end of the pre-processing module (2) is connected to the input end signal connection of the advertisement classification decision module (3), the output end of the advertisement classification decision module (3) is connected to the input end signal connection of the label module (4), the output end of the label module (4) is connected to the input end signal connection of the comparison module (5), the output end of the comparison module (5) is connected to the input end signal connection of the delivery module (6), the delivery module (6) is connected to the advertisement input module (1) via a network communication bidirectional signal connection, the comparison module (5) is connected to the advertisement input module (1) via a network communication bidirectional signal connection, the output end of the delivery module (6) is connected to the input end signal connection of the label module (4), and the output end of the pre-processing module (2) is connected to the input end signal connection of the comparison module (5); The pre-processing module (2) includes a preliminary classification unit (201), a metadata parsing unit (202) and a dynamic adjustment threshold (203); The advertisement classification decision module (3) includes a video advertisement processing unit (301) and a graphic advertisement processing unit (302).

2. The all-round intelligent advertising management system according to claim 1, characterized in that: The advertisement input module (1) comprises a unified import unit (101) and an autonomous upload unit (102); the unified import unit (101) comprises a data interface (1011) and an operation console (1012).

3. The all-round intelligent advertising management system according to claim 2, characterized in that: The operating console (1012) comprises a display screen, a keyboard, a mouse and a chassis, and the data interface (1011) is connected to the chassis via a network communication bidirectional signal.

4. The all-round intelligent advertising management system according to claim 1, characterized in that: The video advertisement processing unit (301) includes a 3D-CNN model and a Transformer model, and the graphic advertisement processing unit (302) includes a ResNet-50 model, a BERT model and a cross-modal attention mechanism.

5. The all-round intelligent advertising management system according to claim 2, characterized in that: The label module (4) includes a brand label (401), a keyword label (402) and a discount strength label (403), and the comparison module (5) includes a comparison unit (501), a judgment unit (502) and an alarm unit (503).

6. The all-round intelligent advertising management system according to claim 5, characterized in that: The output end of the preliminary classification unit (201) is signal-connected to the input end of the alarm unit (503), and the output end of the alarm unit (503) is signal-connected to the input end of the operating console (1012).

7. The all-round intelligent advertising management system according to claim 1, characterized in that: The delivery module (6) includes delivery suggestions (601), consent judgment (602) and a storage unit (603), and the storage unit (603) includes an internal database (6031) and a network database (6032).

8. The all-round intelligent advertising management system according to claim 2, characterized in that: The output end of the network database (6032) is connected to the input end signal of the label module (4), and the operating console (1012) is connected to the delivery module (6) via a network communication two-way signal.

9. The all-round intelligent advertising management system according to claim 5, characterized in that: The output end of the comparison unit (501) is signal-connected to the input end of the judgment unit (502), and the output end of the judgment unit (502) is signal-connected to the input end of the alarm unit (503).

10. The all-round intelligent advertising management system according to claim 1, characterized in that: The output end of the metadata parsing unit (202) is signal-connected to the input end of the video advertisement processing unit (301), and the output end of the metadata parsing unit (202) is signal-connected to the input end of the graphic advertisement processing unit (302).

Images