System and method for intelligent optimization of multimedia content for internet advertising services

Abstract

The application discloses a multimedia content intelligent optimization system and method for internet advertisement service, relates to the field of artificial intelligence, and solves the problem of existing advertisement optimization difficulty, and comprises the following modules: an initialization module: obtaining historical advertisement information, extracting advertisement components, and counting position information of the advertisement components; a data integration module: obtaining content materials, comparing the content materials with the historical advertisement information, obtaining advertisement components corresponding to the content materials, integrating the content materials in combination with position value intervals; a promotion test module: manually checking target advertisements, promoting the target advertisements according to checking results, monitoring promotion data, and analyzing promotion results; and an optimization feedback module: adjusting the advertisements, repeatedly promoting and testing the adjusted advertisements, and optimizing and controlling the target advertisements according to promotion data of the adjusted advertisements. The application can reduce the optimization difficulty of the advertisements and timely feedback optimization conditions.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence and involves big data technology. Specifically, it is a multimedia content intelligent optimization system and method for internet advertising services. Background Technology

[0002] Existing methods for generating and optimizing internet advertising content have the following specific shortcomings: 1. Current internet advertising mainly involves collaborative creation by staff from various fields such as art and design, making ad generation difficult. At the same time, ad design is greatly influenced by personnel, and when disagreements arise among design-related personnel, it is difficult to effectively judge and control the generated ad results.

[0003] 2. Current content analysis of internet advertising is mainly based on specific page displays, including page text, page images, and page layout. This method is too disorganized in its analysis of advertising, and it is difficult to unify different analytical contents, resulting in high analysis complexity.

[0004] 3. Current content optimization for internet advertising mainly relies on manual optimization, with existing staff modifying the content. This method of modification is highly subjective and cannot form a standardized optimization process; at the same time, the results of optimization through this method cannot be fed back in a timely and effective manner.

[0005] To this end, we propose a multimedia content intelligent optimization system and method for internet advertising services. Summary of the Invention

[0006] In view of the shortcomings of existing technologies, the purpose of this invention is to provide a multimedia content intelligent optimization system and method for Internet advertising services. This invention aims to improve advertising content and enhance the promotion effect of advertising.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a multimedia content intelligent optimization system for internet advertising services, the specific working process of each module is as follows: Initialization module: Obtain historical advertising information, decompose the advertising layout based on the historical advertising information, and extract advertising components; perform retrieval based on advertising components and historical advertising information, count the position information of advertising components, and obtain the position value range of advertising components; Data integration module: Acquires content materials, compares them with historical advertising information to obtain the corresponding advertising components, integrates the content materials based on the position value range of the advertising components, and generates target advertisements; Promotion Testing Module: Manually review target ads, promote target ads across multiple channels based on the review results, monitor the promotion data of target ads, and analyze the promotion results of target ads based on the promotion data; Optimization Feedback Module: Based on the promotion results of the target advertisement, adjust the advertisements under different promotion channels to obtain the adjusted advertisements, and repeatedly conduct promotion tests on the adjusted advertisements; use the promotion data of the adjusted advertisements to provide optimization feedback to the target advertisements and control the optimization.

[0008] Furthermore, the location information of the ad components is analyzed, as follows: Statistical analysis of historical advertising information is performed, and advertising pages of different historical advertising information are recorded. The advertising pages are decomposed to obtain multi-dimensional information units. The advertising layout of historical advertising information is obtained, and the advertising information is modularized by combining the multi-dimensional information units to obtain advertising components. The system collects data on ad components from multiple different historical ad messages, calculates the usage frequency of these ad components, and constructs a component frequency list based on their usage frequency. It then iterates through the component frequency list, retrieves ad components using historical ad information, records the position information of the corresponding ad components within the historical ad messages, calculates the position information of ad components of the same category, analyzes the position values ​​of the ad components, and obtains the position value range of the ad components.

[0009] Furthermore, the advertising information is modularized, as follows: The source code of the tags on the advertising page is collected. Based on the source code, the tags on the advertising page are traversed, and the scope of each tag is recorded. According to the scope of the tag, the child tags inside the tag are collected iteratively. The non-tag format content of the child tags in each iteration is recorded to obtain an information list. The information list is traversed. If the non-tag format content has an address jump, the jump content is used to replace the non-tag format content, and the information list is updated to obtain multi-dimensional information units. Based on multi-dimensional information units, each data format is traversed, and the data content under the same source code tag is extracted by combining the source code of historical advertisements. The historical advertisement information is divided by data content and source code tags to obtain advertisement components.

[0010] Furthermore, the position values ​​of the advertising components are analyzed, as follows: Collect data on the ad components in historical ad information, record the data format of each ad component's data content and the tag type of the source code tags to obtain the component parameters; based on the component parameters, iterate through the ad components, record the number of times ad components with the same component parameters appear in different historical ad information to obtain the ad component usage frequency; sort the ad components in descending order according to the usage frequency to obtain the component frequency list. The process involves iterating through the ad components to obtain their parameters and extracting their source code tags. The source code of historical ads is then retrieved. Format tags of equal size are created based on the source code tags, and the source code is iterated through to extract comparison tags. These comparison tags are compared with the source code tags, and the tag with the most identical tags is recorded. The page source code is compiled, and the position information of the comparison tags on the compiled page is extracted. This position information is used to record the position parameters of the ad components. The position parameters of the ad components across different historical ads are statistically analyzed to obtain a set of positions. Finally, the extreme values ​​of the position parameters are extracted from the set of positions to obtain the range of position values.

[0011] Further, generate targeted ads, as follows: The content materials are statistically analyzed to obtain a list of content materials; the component frequency list is obtained, and the ad components are traversed based on the component frequency list to extract the data content of the corresponding ad components; The content materials are extracted, and the data content of the content materials and advertising components is transmitted to the large model for content recognition. The data content with the highest similarity to the content materials is output, and the advertising components corresponding to the data content are recorded to obtain the preset components. The content material list is traversed, and the preset components of each content material are obtained. Obtain the location value range of the ad component, use the preset component as the search condition to extract the location value range of the corresponding ad component, set the location limit value based on the location value range of the preset component, construct component generation information based on the content material, preset component and location limit value, and generate and integrate the content through the component generation information to obtain the target ad.

[0012] Furthermore, the component generation information is as follows: Extract content materials; traverse the ad components according to the component frequency list and extract the data content of each ad component; compare the content materials and data content as comparison items and output the similarity between the content materials and data content; iterate through the data content to obtain the similarity xsd(i) between the content materials and different data content; count the similarity xsd(i), extract the content materials and data content corresponding to the maximum similarity, and associate the ad components corresponding to the data content with the content materials; The content materials are traversed according to the content material list. Each content material in the content material list is compared with the data content using a model to count the association between the content materials and the ad components. Based on the association between the content materials and the ad components, the ad components associated with the content materials are selected as preset components. The content materials and their corresponding preset components are acquired. The advertising components are retrieved based on the preset components. The position values ​​of the preset components are assigned based on the position value range of the advertising components to obtain the position limit values ​​of the preset components. The content materials, preset components and position limit values ​​are statistically analyzed to construct the component generation information.

[0013] Furthermore, the targeted advertisements will be promoted through multiple channels, as detailed below: The process involves: acquiring the target advertisement and generating a visual advertisement page; transmitting the visual advertisement page to the testing personnel, who then manually review the target advertisement based on the visual advertisement page; and controlling the target advertisement based on the results of the manual review. The target advertisement is tested based on the control results, promoted through multiple channels, and the promotion data of the target advertisement is monitored on different channels. The promotion results of the target advertisement are analyzed and recorded based on the promotion data.

[0014] Furthermore, the promotion results of the target advertisements are analyzed, as follows: A visual ad page is generated based on the target ad. Testers use the visual ad page to test the target ad and conduct promotional tests on the target ad. Acquire multiple advertising channels, conduct promotional tests on the target ad on each channel, monitor the promotional data of the target ad on different channels, and statistically analyze the promotional data types to obtain the data type gs; The promotional data is categorized and statistically analyzed based on the data type gs to obtain the promotional data list tgl; the number of promotional channels is counted and denoted as qs; and the promotional data for different channels is obtained based on the number of promotional channels and denoted as tgs(g, q). The mean of data for each data type is calculated based on the promotion data tgs(g, q) to obtain the baseline data jzs(g); Statistical analysis of the benchmark data was performed to construct a benchmark data list (jzl). Based on the promotion data and benchmark data, the data fluctuation of each data type under different promotion channels is calculated, and the weights of different data types are assigned based on the data fluctuation, resulting in the data weight sqz(g). Calculate the data weights and construct a data weight list qzl.

[0015] Furthermore, feedback on the target ads will be optimized, as follows: Adjust the target ad to obtain the adjusted ad, and manually verify the adjusted ad; based on the verification results, conduct promotion tests on the adjusted ad and record the promotion data of the adjusted ad; Based on the promotion results of the target advertisement, obtain the baseline data list jzl, jzl = [jzs(1) to jzs(gs)] and the data weight list qzl, qzl = [sqz(1) to sqz(gs)] of the target advertisement; By combining the promotion data of the adjusted advertisement with the benchmark data list of the target advertisement, the average data value of each data type of the adjusted advertisement is obtained, and the data is aligned with the benchmark data list to obtain the adjusted average value list tjl, tjl = [tjz (1) to tjz (gs)]; The adjusted optimization value tyh is calculated based on the baseline data list and the adjusted mean list, combined with the data weight list. The adjustment status of the target advertisement is judged based on the adjusted optimization value: If tyh > 0, it indicates that the adjusted ad is better than the target ad, and the adjusted ad should be replaced with the target ad. If tyh≤0, it indicates that the adjusted ad has not been optimized compared to the target ad; therefore, the target ad should be left unchanged. Adjust and iterate on the target ads, and optimize and control the target ads.

[0016] A smart optimization method for multimedia content in internet advertising services, including the following optimization methods: Step S1: Obtain historical advertising information, decompose the advertising layout based on the historical advertising information, and extract advertising components; perform retrieval based on advertising components and historical advertising information, statistically analyze the position information of advertising components, and obtain the position value range of advertising components. Step S2: Obtain content materials, compare the content materials with historical advertising information to obtain the advertising components corresponding to the content materials, and integrate the content materials based on the position value range of the advertising components to generate the target advertisement; Step S3: Manually verify the target advertisement, promote the target advertisement through multiple channels based on the verification results, monitor the promotion data of the target advertisement, and analyze the promotion results of the target advertisement based on the promotion data; Step S4: Based on the promotion results of the target advertisement, adjust the advertisements under different promotion channels to obtain the adjusted advertisements, and repeat the promotion test of the adjusted advertisements; use the promotion data of the adjusted advertisements to optimize the target advertisements and control the optimization.

[0017] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention statistically analyzes the layout effect of different historical advertising information by statistically analyzing historical advertising information. Based on the advertising layout, it integrates multi-dimensional information with the source code tags and advertising content to construct an overall component. The advertising is analyzed through the component, which ensures the consistency of advertising content analysis and improves the accuracy of analysis.

[0018] 2. This invention uses artificial intelligence to fit users' advertising needs with advertising components, generating advertising components that meet those needs; it integrates these advertising components to generate target advertisements, thereby reducing the proportion of human involvement in advertisement generation and minimizing discrepancies and errors caused by human factors; at the same time, it uses modular generation of advertising components to reserve space for changes in user needs and to adjust target advertisements in a timely manner.

[0019] 3. This invention ensures the optimization efficiency and promotion effect of target ads by conducting multi-channel promotion tests on target ads, providing feedback on the promotion status of target ads based on promotion data, adjusting target ads based on a large model, repeating promotion tests, and controlling the final ad used based on the promotion status of different target ads. Attached Figure Description

[0020] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 This is a functional block diagram of the present invention; Figure 2 This is a schematic diagram of the advertising processing in this invention; Figure 3 This is a schematic diagram of the process of the present invention; Detailed Implementation It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0022] This application provides a multimedia content intelligent optimization system for internet advertising services. The executing entity of this system includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the system provided in this application: a server, a terminal, etc. In other words, the multimedia content intelligent optimization system for internet advertising services can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0023] Reference Figure 1 The diagram shown is a functional block diagram of a multimedia content intelligent optimization system for internet advertising services provided in an embodiment of the present invention. In this embodiment, the multimedia content intelligent optimization system for internet advertising services includes: an initialization module, a data integration module, a promotion testing module, and an optimization feedback module. The module mentioned in the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0024] In this embodiment of the invention, the functions of each module / unit are as follows: Initialization module: Obtain historical advertising information, decompose the advertising layout based on the historical advertising information, and extract advertising components; perform retrieval based on advertising components and historical advertising information, count the position information of advertising components, and obtain the position value range of advertising components; The specific workflow of the initialization module is as follows: Statistical analysis of historical advertising information is performed, and advertising pages of different historical advertising information are recorded. The advertising pages are decomposed to obtain multi-dimensional information units. The advertising layout of historical advertising information is obtained, and the advertising information is modularized by combining the multi-dimensional information units to obtain advertising components. It should be noted that: historical advertising information refers to advertising information that has been generated and used; advertising page refers to the advertising information display page, which includes the advertising background, advertising text, and advertising layout; advertising component refers to a part of the overall advertising, such as a small piece of text or an inserted image on the advertising page; it is mainly extracted based on the source code of the advertising page (such as text or images within the same container).

[0025] The system collects data on ad components from multiple different historical ad messages, calculates the usage frequency of these ad components, and constructs a component frequency list based on their usage frequency. It then iterates through the component frequency list, retrieves ad components using historical ad information, records the position information of the corresponding ad components within the historical ad messages, calculates the position information of ad components of the same category, analyzes the position values ​​of the ad components, and obtains the position value range of the ad components.

[0026] The modular processing of advertising information is as follows: The source code of the tags on the advertising page is collected. Based on the source code, the tags on the advertising page are traversed, and the scope of each tag is recorded. According to the scope of the tag, the child tags inside the tag are collected iteratively. The non-tag format content of the child tags in each iteration is recorded to obtain an information list. The information list is traversed. If the non-tag format content has an address jump, the jump content is used to replace the non-tag format content, and the information list is updated to obtain multi-dimensional information units. It should be noted that: multi-dimensional information units refer to data content in different data formats; they reflect the information diversity of advertising information.

[0027] Based on multi-dimensional information units, each data format is traversed, and the data content under the same source code tag is extracted by combining the source code of historical advertisements. The historical advertisement information is divided by data content and source code tags to obtain advertisement components.

[0028] The specific analysis of the position values ​​is as follows: Collect data from multiple advertising components with different historical advertising information, record the data format of each advertising component's data content and the tag type of the source code tags to obtain component parameters; based on the component parameters, iterate through the advertising components, record the number of times advertising components with the same component parameters appear in different historical advertising information to obtain the usage frequency of the advertising components; sort the advertising components in descending order according to the usage frequency to obtain a component frequency list. The system iterates through the ad components based on a component frequency list, using the ad components as search criteria to retrieve historical ad information and obtain their component parameters. Based on these parameters, it extracts the source code tags of the ad components. The system then retrieves the page source code of historical ads from the historical ad information. Based on the source code tags, it establishes format tags of equal size and iterates through the page source code to extract comparison tags. These comparison tags are compared with the source code tags, and the comparison tag with the most common tags is recorded. The page source code is then compiled, and the position information of the comparison tags on the compiled page is extracted. This position information is used as the ad component's position information within the historical ad information, and the position parameters are recorded based on this position information. The position parameters of the ad components across different historical ad information are statistically analyzed to obtain a position set. Finally, the extreme values ​​of the position parameters are extracted from the position set to obtain the position value range.

[0029] Data integration module: Acquires content materials, compares them with historical advertising information to obtain the corresponding advertising components, integrates the content materials based on the position value range of the advertising components, and generates target advertisements; It should be noted that: content materials refer to the information that an advertisement needs to specifically display, including but not limited to text and images.

[0030] The specific workflow of the data integration module is as follows: Please see Figure 2 ; Analyze the content materials to obtain a list of content materials; Obtain the component frequency list based on historical advertising information, traverse the advertising components based on the component frequency list, and extract the data content of the corresponding advertising components; The content materials are extracted, and the data content of the content materials and advertising components is transmitted to the large model for content recognition. The data content with the highest similarity to the content materials is output, and the advertising components corresponding to the data content are recorded to obtain the preset components. The content material list is traversed, and the preset components of each content material are obtained.

[0031] It should be noted that the above-mentioned large models refer to large models capable of content understanding and parsing, such as Gemini and Claude 3. The system obtains the location value range of the ad component, uses the preset component as the search condition to extract the location value range of the corresponding ad component, and sets the location limit value based on the location value range of the preset component. The system constructs component generation information based on the content material, the preset component, and the location limit value. The system generates and integrates the content through AIGC and the component generation information to obtain the target ad. It should be noted that the position limit value is any value within the position value range. For example, if the position value range is [0, 2], then the position limit value can be 0, 1, or 2. It should be noted that AIGC (Artificial Intelligence Generated Content) refers to a technology system that uses artificial intelligence technologies (such as machine learning, deep learning, natural language processing, computer vision, etc.) to automatically generate various forms of content, including text, images, audio, video, and code.

[0032] The specific process for obtaining the target advertisement is as follows: Content materials are extracted from the content material list; ad components are traversed from the component frequency list to extract the data content of each ad component; content materials and data content are used as comparison items and transmitted to the large model, which receives the comparison items, performs content comparison, and outputs the similarity between content materials and data content; the data content is iterated to obtain the similarity xsd(i) between content materials and different data content, where xsd(i) represents the similarity between content materials and data content i; the similarity xsd(i) is statistically analyzed, and the content materials and data content corresponding to the maximum similarity are extracted, and the ad components corresponding to the data content are associated with the content materials; The content materials are traversed according to the content material list. Each content material in the content material list is compared with the data content using a model to count the association between the content materials and the ad components. Based on the association between the content materials and the ad components, the ad components associated with the content materials are selected as preset components. The process involves acquiring content materials and their corresponding preset components, retrieving ad components based on the preset components, assigning position values ​​to the preset components based on the position value range of the ad components, obtaining position constraints for the preset components, and statistically analyzing the content materials, preset components, and position constraints to construct component generation information that corresponds one-to-one with the content materials. For all content materials, the process involves statistically analyzing the component generation information and generating components based on AIGC and the component generation information: extracting preset components from the component generation information, recording the tags and formats of the preset components, and obtaining a generation framework; identifying content materials through AIGC and generating material styles and attributes for the content materials; filling the generation framework with the material styles and attributes to obtain generated components, and then laying out the generated components based on the position constraints. The component generation information is traversed to obtain all generated components and their corresponding component layouts; the generated components and their corresponding component layouts are then integrated to obtain the target advertisement.

[0033] It should be noted that: Component generation via AIGC refers to replacing the data content of the ad component with content materials while maintaining the original component format. At the same time, the replacement style of the content materials is controlled through a large model, and the position of the generated component is determined according to the position constraints.

[0034] Promotion Testing Module: Manually review target ads, promote target ads across multiple channels based on the review results, monitor the promotion data of target ads, and analyze the promotion results of target ads based on the promotion data; The specific workflow for the promotion and testing module is as follows: The process involves: acquiring the target advertisement and generating a visual advertisement page; transmitting the visual advertisement page to the testing personnel, who then manually review the target advertisement based on the visual advertisement page; and controlling the target advertisement based on the results of the manual review. The target advertisement is tested based on the control results, promoted through multiple channels, and the promotion data of the target advertisement is monitored on different channels. The promotion results of the target advertisement are analyzed and recorded based on the promotion data.

[0035] The specific process for obtaining promotional data for the target advertisement is as follows: A visual ad page is generated based on the target ad. The testers use the visual ad page to test the target ad. If there are any abnormalities in the visual ad page (such as prohibited content or messy component layout), the abnormalities are recorded and used as a condition for avoiding ad generation. The target ad is then regenerated and manually tested again. If there are no abnormalities in the visual ad page, the target ad is promoted and tested. Acquire multiple advertising channels, conduct promotional tests on the target ad on each channel, monitor the promotional data of the target ad on different channels, and statistically analyze the promotional data types to obtain the data type gs; It should be noted that: promotional data types refer to the types of data that can demonstrate the effectiveness of advertising, such as the number of user clicks, the number of user searches, and the duration of user viewing.

[0036] The promotional data is classified and statistically analyzed according to the data type gs to obtain the promotional data list tgl, tgl = [tgs(1), tgs(2), ..., tgs(gs)]; the promotional data in the promotional data list is uniformly represented by tgs(g); The number of promotion channels is counted and denoted as qs; promotion data under different channels is obtained based on the number of promotion channels and denoted as tgs(g,q), where tgs(g,q) represents the promotion data of the g-th data type under the q-th promotion channel; The mean of data for each data type is calculated based on the promotion data tgs(g, q) to obtain the baseline data jzs(g); ; Statistical analysis is performed on the benchmark data to construct a benchmark data list jzl, jzl = [jzs(1) to jzs(gs)].

[0037] Based on the promotion data tgs(g, q) and the benchmark data jzs(g), the data fluctuation of each data type under different promotion channels is calculated, and the weights of different data types are assigned based on the data fluctuation to obtain the data weight sqz(g). ; Statistical analysis of data weights is performed to construct a data weight list qzl, qzl = [sqz(1) to sqz(gs)].

[0038] It should be noted that: if the promotional data of one data type is 1, 2, 3, 4, 5; and the promotional data of another data type is 10, 20, 30, 40, 50; then when the promotional data of the two data types changes, the advertising promotion effect reflected will be different. Therefore, weights are set according to the promotional data, and the data fluctuation ratio is used to uniformly quantify the data fluctuation status. By taking the difference with 1, the data type with large data changes is assigned a smaller weight value, and the data type with small data changes is assigned a larger weight value, which ensures the comprehensiveness of the judgment of different types of promotional data and improves the effectiveness of the judgment of advertising promotion effect.

[0039] Optimization Feedback Module: Based on the promotion results of the target advertisement, adjust the advertisements under different promotion channels to obtain the adjusted advertisements, and repeatedly conduct promotion tests on the adjusted advertisements; use the promotion data of the adjusted advertisements to provide optimization feedback to the target advertisements and control the optimization. The specific workflow of the optimized feedback module is as follows: The target ad is regenerated using AIGC, and then adjusted to obtain an adjusted ad. The adjusted ad is then manually verified. Based on the verification results, the adjusted ad is tested for promotion, and the promotion data of the adjusted ad is recorded. Based on the promotion results of the target advertisement, obtain the baseline data list jzl, jzl = [jzs(1) to jzs(gs)] and the data weight list qzl, qzl = [sqz(1) to sqz(gs)] of the target advertisement; By combining the promotion data of the adjusted advertisement with the benchmark data list of the target advertisement, the average data value of each data type of the adjusted advertisement is obtained, and the data is aligned with the benchmark data list to obtain the adjusted average value list tjl, tjl = [tjz (1) to tjz (gs)]; The adjusted optimization value tyh is calculated based on the baseline data list and the adjusted mean list, combined with the data weight list. ; If the baseline data list is [2, 20], the adjusted mean list is [3, 10], and the data weight list is [0.9, 0.1], then substituting these values ​​into the formula, we get tyh = 1 × 0.9 + (-10) × 0.1 = -0.1.

[0040] The adjustment status of the target advertisement is judged based on the adjusted optimization value: If tyh > 0, it indicates that the adjusted ad is better than the target ad, and the adjusted ad should be replaced with the target ad. If tyh≤0, it indicates that the adjusted ad has not been optimized compared to the target ad; therefore, the target ad should be left unchanged. Adjust and iterate on the target ads, and optimize and control the target ads.

[0041] Compared to the problems described in the background technology, this invention analyzes the layout effects of different historical advertising information by statistically analyzing historical advertising information. Based on the advertising layout, it integrates multi-dimensional information such as advertising source code tags and advertising content to construct an overall component. This component-based approach ensures consistency in advertising content analysis and improves accuracy. Furthermore, this invention uses artificial intelligence to fit user advertising needs with advertising components, generating advertising components that meet those needs. These components are then integrated to generate target ads, reducing the proportion of human intervention in ad generation and minimizing discrepancies and errors caused by human factors. Modular generation of advertising components also allows for future adjustments to target ads based on changes in user needs. Finally, this invention conducts multi-channel promotional tests on target ads, provides feedback on their promotional status based on promotional data, adjusts target ads based on a large model, repeats promotional tests, and controls the final ad used based on the promotional status of different target ads, ensuring optimization efficiency and promotional effectiveness. Therefore, the multimedia content intelligent optimization system and method for internet advertising services provided by this invention can improve advertising content and enhance advertising promotion effectiveness.

[0042] This application provides a method for intelligent optimization of multimedia content for internet advertising services. The executing entity of this method includes, but is not limited to, at least one electronic device that can be configured to execute the method provided in this application, such as a server or a terminal. In other words, the method for intelligent optimization of multimedia content for internet advertising services can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0043] Reference Figure 3 The diagram shown is a flowchart illustrating a method for intelligent optimization of multimedia content for internet advertising services according to an embodiment of the present invention. In this embodiment, the method for intelligent optimization of multimedia content for internet advertising services includes: Step S1: Obtain historical advertising information, decompose the advertising layout based on the historical advertising information, and extract advertising components; perform retrieval based on advertising components and historical advertising information, statistically analyze the position information of advertising components, and obtain the position value range of advertising components. Step S2: Obtain content materials, compare the content materials with historical advertising information to obtain the advertising components corresponding to the content materials, and integrate the content materials based on the position value range of the advertising components to generate the target advertisement; Step S3: Manually verify the target advertisement, promote the target advertisement through multiple channels based on the verification results, monitor the promotion data of the target advertisement, and analyze the promotion results of the target advertisement based on the promotion data; Step S4: Based on the promotion results of the target advertisement, adjust the advertisements under different promotion channels to obtain the adjusted advertisements, and repeat the promotion test of the adjusted advertisements; use the promotion data of the adjusted advertisements to optimize the target advertisements and control the optimization.

[0044] In detail, the steps in the intelligent optimization method for multimedia content for internet advertising services described in this embodiment of the invention adopt the same approach as described above. Figure 1 The same technical means are used in the intelligent optimization system for multimedia content for internet advertising services described in the article, and it can produce the same technical effects, so it will not be elaborated here.

[0045] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0046] Finally, it should be noted that deleting any one of the above embodiments does not affect the technical solutions of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multimedia content intelligent optimization system for internet advertising services, characterized in that: include: Initialization module: Obtains historical ad information, decomposes the ad layout based on the historical ad information, and extracts ad components; Based on the ad component and historical ad information, the location information of the ad component is statistically analyzed to obtain the location value range of the ad component. Data integration module: Acquires content materials, compares them with historical advertising information to obtain the corresponding advertising components, integrates the content materials based on the position value range of the advertising components, and generates target advertisements; Promotion Testing Module: Manually review target ads, promote target ads across multiple channels based on the review results, monitor the promotion data of target ads, and analyze the promotion results of target ads based on the promotion data; Optimization Feedback Module: Based on the promotion results of the target advertisement, adjust the advertisements under different promotion channels to obtain the adjusted advertisements, and repeatedly conduct promotion tests on the adjusted advertisements; use the promotion data of the adjusted advertisements to provide optimization feedback to the target advertisements and control the optimization.

2. The intelligent optimization system for multimedia content in internet advertising services according to claim 1, characterized in that, The location information of the ad components is analyzed as follows: Statistical analysis of historical advertising information is performed, and advertising pages of different historical advertising information are recorded. The advertising pages are decomposed to obtain multi-dimensional information units. The advertising layout of historical advertising information is obtained, and the advertising information is modularized by combining the multi-dimensional information units to obtain advertising components. The system collects data on ad components from multiple different historical ad messages, calculates the usage frequency of these ad components, and constructs a component frequency list based on their usage frequency. It then iterates through the component frequency list, retrieves ad components using historical ad information, records the position information of the corresponding ad components within the historical ad messages, calculates the position information of ad components of the same category, analyzes the position values ​​of the ad components, and obtains the position value range of the ad components.

3. The intelligent optimization system for multimedia content in internet advertising services according to claim 2, characterized in that, The advertising information is processed in a modular fashion, as follows: The source code of the tags on the advertising page is collected, and the tags on the advertising page are traversed based on the source code. The scope of each tag is recorded. According to the scope of the tag, the sub-tags inside the tag are collected iteratively. The non-tag format content of the sub-tags is recorded in each iteration to obtain an information list. The information list is traversed. If there is an address jump in the non-tag format content, the jump content is used to replace the non-tag format content, and the information list is updated to obtain multi-dimensional information units. Based on multi-dimensional information units, each data format is traversed, and the data content under the same source code tag is extracted by combining the source code of historical advertisements. The historical advertisement information is divided by data content and source code tags to obtain advertisement components.

4. The intelligent optimization system for multimedia content in internet advertising services according to claim 2, characterized in that, The position values ​​of the ad components are analyzed as follows: Collect data on the ad components in historical ad information, record the data format of each ad component's data content and the tag type of the source code tags to obtain the component parameters; based on the component parameters, iterate through the ad components, record the number of times ad components with the same component parameters appear in different historical ad information to obtain the ad component usage frequency; sort the ad components in descending order according to the usage frequency to obtain the component frequency list. The ad components are traversed to obtain their component parameters, and the source code tags of the ad components are extracted based on the component parameters; the page source code of the historical ads is obtained; format tags of the same size are set based on the source code tags, and the page source code is traversed to extract comparison tags; Compare the comparison tags with the source code tags, and record the comparison tag that has the most identical tags with the source code tags; The page source code is compiled, and the position information of the comparison tag on the compiled page is extracted as the position information of the ad component on the historical ad information. The position parameters are recorded by the position information. The position parameters of the ad component on different historical ad information are statistically analyzed to obtain the position set. The extreme values ​​of the position parameters are extracted based on the position set to obtain the position value range.

5. The intelligent optimization system for multimedia content in internet advertising services according to claim 1, characterized in that, Generate the target advertisement as follows: The content materials are statistically analyzed to obtain a list of content materials; the component frequency list is obtained, and the ad components are traversed based on the component frequency list to extract the data content of the corresponding ad components; Extract content materials, transmit the data content of content materials and advertising components to a large model for content recognition, output the data content with the highest similarity to the content materials, record the advertising components corresponding to the data content, and obtain the preset components; Iterate through the list of content materials and retrieve the preset components for each content material; Obtain the location value range of the ad component, use the preset component as the search condition to extract the location value range of the corresponding ad component, set the location limit value based on the location value range of the preset component, construct component generation information based on the content material, preset component and location limit value, and generate and integrate the content through the component generation information to obtain the target ad.

6. The intelligent optimization system for multimedia content in internet advertising services according to claim 5, characterized in that, The component generation information is as follows: Extract content materials; traverse the ad components according to the component frequency list and extract the data content of each ad component; compare the content materials and data content as comparison items and output the similarity between the content materials and data content; iterate through the data content to obtain the similarity xsd(i) between the content materials and different data content; count the similarity xsd(i), extract the content materials and data content corresponding to the maximum similarity, and associate the ad components corresponding to the data content with the content materials; The content materials are traversed according to the content material list. Each content material in the content material list is compared with the data content using a model to count the association between the content materials and the ad components. Based on the association between the content materials and the ad components, the ad components associated with the content materials are selected as preset components. The content materials and their corresponding preset components are acquired. The advertising components are retrieved based on the preset components. The position values ​​of the preset components are assigned based on the position value range of the advertising components to obtain the position limit values ​​of the preset components. The content materials, preset components and position limit values ​​are statistically analyzed to construct the component generation information.

7. The intelligent optimization system for multimedia content in internet advertising services according to claim 1, characterized in that, Promote the target ads across multiple channels, as follows: The target advertisement is obtained and a visual advertisement page is generated; the visual advertisement page is transmitted to the inspection personnel, who then manually verify the target advertisement based on the visual advertisement page. Control the target advertisements based on the results of manual verification; The target advertisement is tested based on the control results of the target advertisement, and the advertisement is promoted through multiple channels. The promotion data of the target advertisement on different channels is monitored, the promotion results of the target advertisement are analyzed based on the promotion data, and the promotion results are recorded.

8. The intelligent optimization system for multimedia content in internet advertising services according to claim 7, characterized in that, The results of the target advertisement promotion were analyzed as follows: A visual ad page is generated based on the target ad. Testers use the visual ad page to test the target ad and conduct promotional tests on the target ad. Acquire multiple advertising channels, conduct promotional tests on the target ad on each channel, monitor the promotional data of the target ad on different channels, and statistically analyze the promotional data types to obtain the data type gs; The promotional data is categorized and statistically analyzed based on the data type gs to obtain the promotional data list tgl; the number of promotional channels is counted and denoted as qs; and the promotional data for different channels is obtained based on the number of promotional channels and denoted as tgs(g, q). The mean of data for each data type is calculated based on the promotion data tgs(g, q) to obtain the baseline data jzs(g); Statistical analysis of the benchmark data was performed to construct a benchmark data list (jzl). Based on the promotion data and benchmark data, the data fluctuation of each data type under different promotion channels is calculated, and the weights of different data types are assigned based on the data fluctuation, resulting in the data weight sqz(g). Calculate the data weights and construct a data weight list qzl.

9. The intelligent optimization system for multimedia content in internet advertising services according to claim 8, characterized in that, Optimize the target ad as follows: Adjust the target ad to obtain the adjusted ad, and manually verify the adjusted ad; based on the verification results, conduct promotion tests on the adjusted ad and record the promotion data of the adjusted ad; Based on the promotion results of the target advertisement, obtain the baseline data list jzl, jzl = [jzs(1) to jzs(gs)] and the data weight list qzl, qzl = [sqz(1) to sqz(gs)] of the target advertisement; By combining the promotion data of the adjusted advertisement with the benchmark data list of the target advertisement, the average data value of each data type of the adjusted advertisement is obtained, and the data is aligned with the benchmark data list to obtain the adjusted average value list tjl, tjl = [tjz (1) to tjz (gs)]; The adjusted optimization value tyh is calculated based on the baseline data list and the adjusted mean list, combined with the data weight list. The adjustment status of the target advertisement is judged based on the adjusted optimization value: If tyh > 0, it indicates that the adjusted ad is better than the target ad, and the adjusted ad should be replaced with the target ad. If tyh≤0, it indicates that the adjusted ad has not been optimized compared to the target ad; therefore, the target ad should be left unchanged. Adjust and iterate on the target ads, and optimize and control the target ads.

10. A method for intelligent optimization of multimedia content for internet advertising services, applicable to the intelligent optimization system for multimedia content for internet advertising services as described in any one of claims 1-9, characterized in that, The optimization method includes: Step S1: Obtain historical advertising information, decompose the advertising layout based on the historical advertising information, and extract advertising components; perform retrieval based on advertising components and historical advertising information, statistically analyze the position information of advertising components, and obtain the position value range of advertising components. Step S2: Obtain content materials, compare the content materials with historical advertising information to obtain the advertising components corresponding to the content materials, and integrate the content materials based on the position value range of the advertising components to generate the target advertisement; Step S3: Manually verify the target advertisement, promote the target advertisement through multiple channels based on the verification results, monitor the promotion data of the target advertisement, and analyze the promotion results of the target advertisement based on the promotion data; Step S4: Based on the promotion results of the target advertisement, adjust the advertisements under different promotion channels to obtain the adjusted advertisements, and repeat the promotion test of the adjusted advertisements; use the promotion data of the adjusted advertisements to optimize the target advertisements and control the optimization.

Images