Advertisement publishing information management system and method

Abstract

The present application relates to the field of advertising management technology, and discloses an advertising publishing information management system and method, which first obtains the advertising content uploaded by the user and the list of its target distribution platforms, and extracts the content specification requirements of each platform, and then performs semantic detection on the advertising content based on the platform content specification requirements of each target distribution platform to obtain semantic detection results. If it is found that the copy is not suitable, the advertising content is adapted and enhanced to obtain adapted advertising content, ensuring that the advertising content meets the platform specifications while maintaining high-quality expression. Ultimately, the automated publishing of the adapted advertising content is achieved, significantly improving the efficiency and accuracy of advertising publishing. This method integrates advanced natural language processing technology to support unified management across multiple platforms, reducing the complexity and error rate of manual operations. It is suitable for enterprises and individuals who need to publish advertisements efficiently and accurately, and promotes the development of the advertising industry towards intelligence.

Description

Technical Field

[0001] The present application relates to the field of advertising management technology, and more specifically, to an advertising publishing information management system and method. Background Art

[0002] In today's digital and networked era, advertising has become a crucial means for businesses to promote their brands, products, and services. With the development of internet technology, advertising channels have become increasingly diverse, encompassing numerous digital media platforms, including social media, search engines, and video platforms. However, each platform has its own unique content specifications and format requirements, posing challenges to the production and distribution of advertising content. Traditionally, advertising has relied on manual processes, requiring advertisers or agencies to tailor ad content to each platform to meet its specific requirements. This approach is not only time-consuming and labor-intensive, but also prone to errors, significantly limiting the efficiency and effectiveness of advertising.

[0003] While existing advertising publishing management systems achieve a certain degree of automated management of advertising content, most systems are only capable of handling simple ad content distribution tasks and lack the ability to conduct in-depth semantic analysis of advertising content. This means they are unable to automatically identify and adapt ad copy to ensure it complies with the specific specifications of each platform, such as complex rules such as word count limits, language style preferences, and sensitive word filtering. Furthermore, these systems often lack unified cross-platform management, forcing advertisers to switch between multiple systems to complete ad publishing on different platforms, further increasing operational complexity and management costs.

[0004] Therefore, we look forward to an optimized information management solution for advertising publishing. Summary of the Invention

[0005] This application is proposed to address the above technical issues. The embodiments of this application provide an information management system and method for advertising publishing, which integrates intelligent semantic detection and automated adaptation technology. First, it obtains the campaign planning copy and a list of target platforms, extracts the specifications of each platform, performs semantic analysis and adaptation optimization on the copy to ensure that the content meets the standards of each platform, and finally realizes automated publishing, thereby improving the efficiency and accuracy of creative campaign planning.

[0006] According to one aspect of the present application, a method for information management of advertising publication is provided, comprising: obtaining a list of advertising content uploaded by a user and a target distribution platform; extracting and recording the platform content specification requirements of each target distribution platform; performing semantic detection on the advertising content based on the platform content specification requirements of each target distribution platform to obtain a semantic detection result; in response to the semantic detection result that the text part is not compatible, performing content adaptation and enhancement on the advertising content to obtain adapted advertising content; connecting to each target distribution platform in the list of the target distribution platform to realize the automatic publication of the adapted advertising content; wherein, performing content adaptation and enhancement on the advertising content to obtain adapted advertising content comprises: providing Take the set of text semantic detection logic rules; perform semantic embedding coding on the text part in the advertisement content to obtain the original text part semantic embedding coding vector; perform semantic embedding coding on each text semantic detection logic rule to obtain a set of text semantic detection logic rule semantic embedding coding vectors; perform rule semantic saliency modulation on the set of text semantic detection logic rule semantic embedding coding vectors to obtain a set of saliency modulated text semantic detection logic rule semantic embedding coding vectors; input the set of original text part semantic embedding coding vectors and the saliency modulated text semantic detection logic rule semantic embedding coding vectors into a text adapter based on a large language model to obtain the adapted advertisement content.

[0007] In the above-mentioned advertising publishing information management method, based on the platform content specification requirements of each target distribution platform, semantic detection is performed on the advertising content to obtain a semantic detection result, including: intelligently classifying the advertising content to obtain a text part, a picture part and a video part; extracting the text semantic detection logic rules from the platform content specification requirements to obtain a set of text semantic detection logic rules; using the set of text semantic detection logic rules to verify the text parts one by one to obtain the semantic detection result.

[0008] In the above-mentioned advertising publishing information management method, the copy semantic detection logic rules are extracted from the platform content specification requirements to obtain a set of copy semantic detection logic rules, including: accessing the interface of the target distribution platform and pulling the latest copy semantic detection logic rules from the target distribution platform; updating the copy semantic detection logic rules based on the latest copy semantic detection logic rules.

[0009] In the above-mentioned advertising publishing information management method, the set of the text semantic detection logic rule semantic embedding coding vectors is subjected to rule semantic significance modulation to obtain a set of significance modulated text semantic detection logic rule semantic embedding coding vectors, including: calculating the rule autocorrelation significance weight value of each text semantic detection logic rule semantic embedding coding vector in the set of the text semantic detection logic rule semantic embedding coding vectors to obtain a set of rule autocorrelation significance weight values; based on the set of rule autocorrelation significance weight values, the set of the text semantic detection logic rule semantic embedding coding vectors is subjected to significance modulation to obtain the set of significance modulated text semantic detection logic rule semantic embedding coding vectors.

[0010] In the above-mentioned advertising publishing information management method, the rule autocorrelation significance weight value of each copy semantic detection logic rule semantic embedding coding vector in the set of the copy semantic detection logic rule semantic embedding coding vector is calculated to obtain a set of rule autocorrelation significance weight values, including: inputting the set of the copy semantic detection logic rule semantic embedding coding vector into the semantic detection logic rule context encoder based on the converter model to obtain a set of context copy semantic detection logic rule semantic embedding coding vectors; inputting each context copy semantic detection logic rule semantic embedding coding vector in the set of context copy semantic detection logic rule semantic embedding coding vectors into the rule autocorrelation significance measurement network to obtain the set of rule autocorrelation significance weight values.

[0011] In the above-mentioned advertising publishing information management method, each context copy semantic detection logic rule semantic embedding coding vector in the set of context copy semantic detection logic rule semantic embedding coding vectors is input into a rule autocorrelation significance measurement network to obtain a set of rule autocorrelation significance weight values, including: the rule autocorrelation significance measurement network calculates the rule autocorrelation significance weight value of each context copy semantic detection logic rule semantic embedding coding vector using the following formula, wherein the formula is: ;in, Represents the contextual copywriting semantic detection logic rule semantic embedding encoding vector, is the auxiliary reference weight vector for saliency measurement, is the significance measure linear connection weight matrix, represents matrix multiplication, is the saliency measure bias vector, Indicates the significant weight value of the regular autocorrelation.

[0012] The above-mentioned advertising publishing information management method also includes: before inputting each context copy semantic detection logic rule semantic embedding coding vector in the set of context copy semantic detection logic rule semantic embedding coding vectors into the rule autocorrelation significance measurement network, the semantic similarity incentive is pre-performed on each context copy semantic detection logic rule semantic embedding coding vector.

[0013] According to another aspect of the present application, an advertising publishing information management system is provided, including: a user uploaded content acquisition module, used to obtain advertising content uploaded by users and a list of target distribution platforms; a platform content specification requirement extraction module, used to extract and record the platform content specification requirements of each target distribution platform; an advertising content semantic detection module, used to perform semantic detection on the advertising content based on the platform content specification requirements of each target distribution platform to obtain a semantic detection result; an advertising content adaptation enhancement module, used to perform content adaptation enhancement on the advertising content to obtain adapted advertising content in response to the semantic detection result that the copy part is not adapted; an advertising content automatic publishing module, used to connect to each target distribution platform in the list of the target distribution platform to realize the automatic publishing of the adapted advertising content.

[0014] Compared with the existing technology, the advertising publishing information management system and method provided by this application first obtains the advertising content uploaded by the user and the list of its target distribution platforms, and extracts the content specification requirements of each platform. Then, based on the platform content specification requirements of each target distribution platform, the advertising content is semantically detected to obtain semantic detection results. If it is found that the copy is not suitable, the advertising content is adapted and enhanced to obtain adapted advertising content, ensuring that the advertising content meets the platform specifications while maintaining high-quality expression. Ultimately, the automated publishing of the adapted advertising content is achieved, significantly improving the efficiency and accuracy of advertising publishing. This method integrates advanced natural language processing technology to support unified management across multiple platforms, reducing the complexity and error rate of manual operations. It is suitable for enterprises and individuals who need to publish advertisements efficiently and accurately, and promotes the development of the advertising industry towards intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 The figure shows a flow chart of the advertising publishing information management method according to an embodiment of the present application.

[0017] Figure 2 The figure shows a flow chart of step S3 in the advertising publishing information management method according to an embodiment of the present application.

[0018] Figure 3 The figure shows a flow chart of step S32 in the advertising publishing information management method according to an embodiment of the present application.

[0019] Figure 4 The figure shows a flow chart of step S4 in the advertising publishing information management method according to an embodiment of the present application.

[0020] Figure 5 The figure shows a flow chart of step S42 in the advertising publishing information management method according to an embodiment of the present application.

[0021] Figure 6 The figure shows a structural diagram of an advertisement publishing information management system according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0023] In today's digital and networked era, advertising has become a crucial means for businesses to promote their brands, products, and services. With the development of internet technology, advertising channels are becoming increasingly diverse, encompassing numerous digital media platforms, including social media, search engines, and video platforms. However, each platform has its own unique content standards and format requirements, creating challenges for the production and distribution of advertising content.

[0024] Based on this, in this application, a method for information management of advertising publication is provided. Figure 1 The figure shows a flow chart of the advertising information management method according to the embodiment of the present application. Figure 1 As shown, the advertising publishing information management method includes: S1, obtaining the advertising content uploaded by the user and the list of target distribution platforms; S2, extracting and recording the platform content specification requirements of each target distribution platform; S3, based on the platform content specification requirements of each target distribution platform, performing semantic detection on the advertising content to obtain a semantic detection result; S4, in response to the semantic detection result that the copy part is not compatible, performing content adaptation and enhancement on the advertising content to obtain adapted advertising content; S5, connecting with each target distribution platform in the list of the target distribution platform to realize the automatic release of the adapted advertising content.

[0025] For example, in step S1, a list of advertising content uploaded by the user and the target distribution platform is obtained. It should be understood that an efficient advertising publishing process begins with the accurate identification and management of advertising content and its target platform. By providing a convenient upload path and an intuitive selection mechanism, not only the user's learning cost is reduced, but also data input problems caused by human error are reduced. For example, spelling errors may occur when manually entering the target platform name, but such problems can be effectively avoided by using a drop-down menu or checkbox. In addition, instant verification of uploaded content also helps to detect and resolve potential technical problems early, such as file corruption or format incompatibility, thereby avoiding complications that may arise in the subsequent processing stages.

[0026] In a specific embodiment, a user-friendly interface is first provided to allow advertisers or agencies to easily upload advertising content and specify a list of target platforms where they wish to publish advertisements. For example, it can be designed as a web-based application. After logging into the application, users can see a simple and intuitive interface with a dedicated area for uploading advertising content files (such as pictures, videos, and text). In order to simplify the operation process, a drag-and-drop upload function is also supported, allowing users to complete the upload by simply dragging local files directly to the designated area. At the same time, on the same interface, a drop-down menu or check box list will be provided, listing all optional target distribution platforms, from which users can select the corresponding platform according to their needs. In addition, in order to improve the user experience, commonly used target platforms can be intelligently recommended based on the user's usage habits or historical records, further reducing the user's selection time.

[0027] For example, in step S2, the platform content specification requirements of each target distribution platform are extracted and recorded. It should be understood that different advertising platforms have highly heterogeneous content specifications. Manual management and maintenance of these specifications is not only time-consuming and labor-intensive, but also prone to errors. Acquiring and parsing these specifications in an automated manner can not only significantly improve efficiency, but also ensure the accuracy and real-time nature of the specification information. For example, when a platform adjusts its advertising policy, it can respond quickly and automatically update the corresponding rule set to avoid problems with advertising content mismatch due to information lag. In addition, structured storage specifications also facilitate subsequent semantic detection and content adaptation processes, enabling the rapid location and application of relevant rules, thereby improving overall processing speed and accuracy.

[0028] In a specific embodiment, first, the latest content specification requirements are obtained from each target distribution platform through an API interface or web crawler technology. For example, for a social media platform, its content specifications may include the maximum word limit of the copy, the resolution requirements of the image, the length limit of the video, and specific sensitive word filtering rules. This information can be obtained by calling the official API provided by the platform, or in the absence of a public API, using web crawler technology to regularly crawl the platform's help documents or release guide pages. Taking Facebook as an example, you can call the Facebook Graph API to obtain its latest advertising specifications; for some niche platforms that do not provide APIs, you can set up scheduled tasks and use crawler tools to regularly access and parse their specification pages.

[0029] After obtaining these specifications, they need to be structured to facilitate subsequent semantic detection and content adaptation. The specification requirements of different platforms will be broken down into multiple operational rule sets, such as requirements for copywriting, image requirements, and video requirements. Each rule set contains specific parameters and conditions, such as the maximum word count for copywriting, recommended language style, minimum image resolution, video format requirements, etc. These rule sets will be stored in a specially designed database so that they can be called and updated at any time. For example, a relational database (such as MySQL) may be used to store these rules, while a NoSQL database (such as MongoDB) may be used to process unstructured data to ensure flexibility and efficiency.

[0030] For example, in step S3, semantic detection is performed on the ad content based on the platform content specifications of each target distribution platform to obtain semantic detection results. It should be understood that different advertising platforms have highly heterogeneous content specifications. For example, Facebook may require that content be no longer than 150 characters and must not contain certain sensitive words, while Instagram may require that content be concise. Failure of ad content to comply with these specific specifications may result in ad rejection, potentially leading to legal risks or damage to brand image. Therefore, semantic detection ensures that ad content strictly complies with the regulations of each platform, avoiding issues caused by non-compliance. Manually managing and maintaining these specifications is not only time-consuming and labor-intensive, but also prone to errors. Traditionally, ad publishing relies on manual processes, requiring advertisers or agencies to tailor ad content to meet the specific requirements of each platform. This approach is not only inefficient but also prone to human errors, such as spelling errors and formatting mismatches. Automated acquisition and parsing of these specifications significantly improves efficiency while ensuring the accuracy and real-time nature of the information provided. For example, when a platform adjusts its advertising policy, it can respond quickly and automatically update the corresponding rule set to avoid the problem of inappropriate advertising content due to information lag.

[0031] In one example, if Figure 2 As shown, step S3, based on the platform content specification requirements of each target distribution platform, performs semantic detection on the advertising content to obtain a semantic detection result, including: S31, intelligently classifies the advertising content to obtain a text part, a picture part and a video part; S32, extracts the text semantic detection logic rules from the platform content specification requirements to obtain a set of text semantic detection logic rules; S33, uses the set of text semantic detection logic rules to verify the text parts one by one to obtain the semantic detection result.

[0032] Specifically, considering that advertising content usually exists in a mixed form, such as pictures containing text descriptions, web pages with embedded videos, or PDF files with text and pictures side by side. In such diverse forms of expression, advertising copy, as the core element that directly conveys brand information, promotional content, and user action instructions (such as "click to buy"), its semantic accuracy directly affects the effectiveness and compliance of advertising. For example, an e-commerce platform may prohibit the use of "absolute terms" (such as "best" and "first"), while social media platforms have strict restrictions on sensitive topics. In contrast, the normative requirements for pictures and videos are more focused on technical parameters (such as size, format) or explicit illegal content (such as bloody scenes), their semantic complexity is lower, and the platform rules change less frequently. Therefore, in this application, the advertising content is intelligently classified to obtain the text part, the picture part, and the video part, and semantic detection is mainly performed on the text part.

[0033] Specifically, uploaded ad content is first intelligently classified into text, image, and video components. This means identifying the format of the uploaded content through file header information, extension, or MIME type. For example, a docx file is classified as text, a jpeg as an image, and an mp4 as a video. For composite files (such as PDFs containing mixed text and images), document parsing tools such as Apache PDFBox are used to separate them into independent units, extracting the text, image, and video components.

[0034] After completing ad content classification, the platform extracts semantic detection logic rules from the platform's content specification requirements to form a set of semantic detection logic rules. During this process, the platform's interfaces are connected to pull the latest semantic detection logic rules and update the existing rule set based on these rules. For example, if a platform recently adjusted its sensitive word list, these changes will be automatically synchronized to ensure that the rule set is always up to date. This way, detection logic can be dynamically adjusted before each ad is released to ensure that the ad content meets the latest platform requirements.

[0035] In one example, if Figure 3As shown, step S32, extracting the copy semantic detection logic rules from the platform content specification requirements to obtain a set of copy semantic detection logic rules, includes: S321, accessing the interface of the target distribution platform, and pulling the latest copy semantic detection logic rules from the target distribution platform; S322, updating the copy semantic detection logic rules based on the latest copy semantic detection logic rules. In this way, the scalability and adaptability can be enhanced. With the continuous expansion of advertising channels, new platforms and specifications emerge in an endless stream, and the traditional manual management model is difficult to keep up with the speed of this change. In this application, through flexible API interface technology, the specification requirements of the new platform can be easily integrated, and the dynamic update of the specifications can be achieved without a lot of manual intervention. For example, if an emerging short video platform launches a new set of advertising specifications, it only needs to add support for the platform API to quickly incorporate it into the management system without redesigning the entire architecture.

[0036] Next, the copy parts will be verified one by one using a set of copy semantic detection logic rules. Specifically, the copy will be scanned word by word, and regular expressions or other pattern matching methods will be used to find parts that match the rules. For example, if an e-commerce platform prohibits the use of absolute terms, you can search for sentences containing keywords such as "best" and "first" in the copy and mark them as potential problems. Similarly, for restrictions on sensitive topics, you can use a predefined list of sensitive words to check whether the copy contains relevant words, and determine whether it violates the rules based on the context. That is, the semantic detection result is to determine whether the copy part is adapted.

[0037] Exemplarily, in step S4, in response to the semantic detection result indicating that the copy is partially incompatible, the ad content is adapted and enhanced to obtain adapted ad content. It should be understood that manually adjusting ad copy to suit the requirements of multiple platforms is not only time-consuming and labor-intensive, but also prone to errors. Traditionally, advertisers or agencies need to modify ad copy for each platform individually, which not only increases workload but also can lead to human errors such as spelling errors and formatting mismatches. However, an automated content adaptation and enhancement process can complete adaptation for multiple platforms in a short period of time, significantly improving the success rate of ad placement. First, a set of semantic detection logic rules is extracted. These rules encompass the platform's specific requirements for copy, such as word count limits, language style preferences, and sensitive word filtering. Next, these rules, along with the ad copy, are input into a copy adapter based on a large language model. The copy adapter is a complex deep learning model that generates new copy that meets platform specifications based on the input rules and copy content.

[0038] In one example, if Figure 4As shown, step S4, in response to the semantic detection result that the copy part is not suitable, the advertising content is content-adapted and enhanced to obtain adapted advertising content, including: S41, extracting the set of copy semantic detection logic rules; S42, inputting the set of copy semantic detection logic rules and the copy part in the advertising content into a copy adapter based on a large language model to obtain the adapted advertising content.

[0039] In one example, if Figure 5 As shown, step S42, the set of the copy semantic detection logic rules and the copy part in the advertising content are input into the copy adapter based on the large language model to obtain the adapted advertising content, including: S421, semantic embedding coding of the copy part in the advertising content to obtain the original copy part semantic embedding coding vector; S422, semantic embedding coding of each copy semantic detection logic rule in the set of the copy semantic detection logic rules to obtain a set of copy semantic detection logic rule semantic embedding coding vectors; S423, rule semantic saliency modulation of the set of the copy semantic detection logic rule semantic embedding coding vectors to obtain a set of saliency modulated copy semantic detection logic rule semantic embedding coding vectors; S424, inputting the set of the original copy part semantic embedding coding vector and the saliency modulated copy semantic detection logic rule semantic embedding coding vector into the copy adapter based on the large language model to obtain the adapted advertising content.

[0040] First, considering that natural language text (such as advertising copy and platform rules) is highly unstructured and difficult to directly process by algorithms, semantic embedding coding can convert text into numerical vectors (i.e., points in a high-dimensional space), thereby structuring the semantic information and facilitating machine understanding and computation. Therefore, semantic embedding coding is performed on the text portion of the advertising content to obtain a semantic embedding coding vector for the original text portion. Specifically, the text is processed using a pre-trained language model (such as BERT or RoBERTa). These language models, based on the Transformer architecture, can understand the contextual information of the text portion of the advertising content and generate a high-dimensional vector representation, namely the semantic embedding coding vector for the original text portion. Next, the same embedding coding is performed on each copy semantic detection logic rule in the set of copy semantic detection logic rules to generate a set of semantic embedding coding vectors for the copy semantic detection logic rules. These rules may include word count limits, sensitive word filtering, and language style preferences. These rules are input one by one into the same language model to generate the corresponding semantic embedding coding vectors for the copy semantic detection logic rules.

[0041] In particular, considering that during the text content adaptation process, there are multiple semantic detection logic rules in the set of text semantic detection logic rules, and each of these rules has different focuses, when adapting text based on a large language model, it is preferable to determine the influence of each semantic detection logic rule in the set on the text adaptation process based on different text content and platform characteristics, so that the final adapted text content can meet platform rules while having greater expressiveness and content fidelity.

[0042] In one example, rule semantic saliency modulation is performed on the set of the text semantic detection logic rule semantic embedding coding vectors to obtain a set of saliency modulated text semantic detection logic rule semantic embedding coding vectors, including: calculating the rule autocorrelation saliency weight value of each text semantic detection logic rule semantic embedding coding vector in the set of the text semantic detection logic rule semantic embedding coding vectors to obtain a set of rule autocorrelation saliency weight values; based on the set of rule autocorrelation saliency weight values, significance modulation is performed on the set of the text semantic detection logic rule semantic embedding coding vectors to obtain the set of saliency modulated text semantic detection logic rule semantic embedding coding vectors.

[0043] In one example, the rule autocorrelation significance weight value of each copy semantic detection logic rule semantic embedding coding vector in the set of the copy semantic detection logic rule semantic embedding coding vector is calculated to obtain a set of rule autocorrelation significance weight values, including: inputting the set of the copy semantic detection logic rule semantic embedding coding vector into a semantic detection logic rule context encoder based on a converter model to obtain a set of context copy semantic detection logic rule semantic embedding coding vectors; inputting each context copy semantic detection logic rule semantic embedding coding vector in the set of context copy semantic detection logic rule semantic embedding coding vectors into a rule autocorrelation significance measurement network to obtain the set of rule autocorrelation significance weight values.

[0044] Specifically, there may be correlations between the various copy semantic detection logic rules. Directly using independently encoded copy semantic detection logic rule semantic embedding encoding vectors may ignore the contextual relationship between the rules. The transformer model (such as Transformer) can capture the long-distance dependency between the semantic embedding encoding vectors of the copy semantic detection logic rules through the self-attention mechanism, thereby enhancing the global consistency of the semantic representation. Inputting the set of the semantic embedding encoding vectors of the copy semantic detection logic rules into the semantic detection logic rule context encoder based on the transformer model can obtain the contextual copy semantic detection logic rule semantic embedding encoding vector containing global context information.

[0045] Next, the set of semantic embedding vectors for the contextual copy semantic detection logic rules is input into a rule autocorrelation saliency measurement network. This network is a specially designed deep learning model that calculates a rule autocorrelation saliency weight for each rule. This method assigns a saliency weight to each rule based on its contextual information, prioritizing more important rules during the copy adaptation process. After obtaining the rule autocorrelation saliency weights, the set of semantic embedding vectors for the copy semantic detection logic rules is saliency modulated based on these weights. Specifically, the original rule semantic embedding vectors are weighted to give more important rules a greater influence during the copy adaptation process. This saliency modulation helps better process multiple semantic detection logic rules, ensuring that the resulting copy not only complies with platform specifications but also has high expressiveness and fidelity.

[0046] In one example, each contextual copy semantic detection logic rule semantic embedding encoding vector in the set of contextual copy semantic detection logic rule semantic embedding encoding vectors is input into a rule autocorrelation significance measurement network to obtain a set of rule autocorrelation significance weight values, including: the rule autocorrelation significance measurement network calculates the rule autocorrelation significance weight value of each contextual copy semantic detection logic rule semantic embedding encoding vector using the following formula, wherein the formula is: ;in, Represents the contextual copywriting semantic detection logic rule semantic embedding encoding vector, is the auxiliary reference weight vector for saliency measurement, is the significance measure linear connection weight matrix, represents matrix multiplication, is the saliency measure bias vector, Represents the rule autocorrelation significant weight value. Here, the significance measure linear connection weight matrix and the saliency measure bias vector Through parameterized learning, the importance of rules can be dynamically adjusted according to the semantics of the current advertising copy. It is the rule vector encoded by the converter model, which already contains the contextual association between the rules. Through matrix multiplication, the model further combines the semantics of the rules with the adaptation requirements of the current copy to capture complex semantic relationships. More specifically, the saliency measurement auxiliary reference weight vector As a global reference, it balances rule constraints and copywriting quality. These parameters can be obtained through supervised learning and back-propagation optimization.

[0047] Finally, the combination of the semantic embedding encoding vector of the original copy and the semantic embedding encoding vector of the semantic detection logic rule for the saliency-modulated copy is input into the large language model-based copy adapter to obtain the adapted advertising content. Specifically, the two input vector sets are first fused. This operation combines the semantic information of the original copy with the saliency weights of the rules to form a new comprehensive vector representation. The large language model-based copy adapter then leverages its powerful contextual understanding capabilities to comprehensively analyze the fused vector. It not only understands the literal meaning of the copy but also infers deeper information such as the copy's sentiment and thematic relevance based on contextual information. For example, it may identify that certain words in the copy do not conform to the style preferences of a particular platform and make adjustments accordingly. Finally, the copy adapter generates a new copy based on the fused vector. This process goes beyond simple text replacement or truncation; rather, it reorganizes the copy using natural language generation technology to ensure that it conforms to platform specifications while preserving the core message and emotional expression of the original copy. For example, if the original copy exceeds the limit of a platform, the copy adapter will generate a shorter version of the copy; if the copy contains sensitive words, the copy adapter will suggest replacing these words and ensure that the replaced copy can still convey the same emotion and intention.

[0048] In a preferred embodiment, the advertising publication information management method also includes: before inputting each context copy semantic detection logic rule semantic embedding coding vector in the set of context copy semantic detection logic rule semantic embedding coding vectors into the rule autocorrelation significance measurement network, pre-performing semantic similarity incentive on each context copy semantic detection logic rule semantic embedding coding vector.

[0049] Specifically, before inputting each contextual copy semantic detection logic rule semantic embedding coding vector in the set of contextual copy semantic detection logic rule semantic embedding coding vectors into the rule autocorrelation significance measurement network, it is expected to pre-perform semantic similarity incentives on each contextual copy semantic detection logic rule semantic embedding coding vector to enhance the consistent certainty of the set of rule autocorrelation significance weight values.

[0050] Specifically, firstly, the context copy semantic detection logic rule semantic embedding encoding vector is performed, for example, Calculate the second distance association matrix, expressed as: ;in, and represents the context copy semantic detection logic rule semantic embedding encoding vector and The eigenvalues ​​at the positions, Indicates the calculation of eigenvalues and between distance, Represents the two-distance correlation matrix The eigenvalue of the location.

[0051] Then obtain the two distance association matrix of Eigenvalues , to form the distance eigenvector .

[0052] That is, for each context copy semantic detection logic rule semantic embedding encoding vector, for example, , and the corresponding distance eigenvector is obtained .

[0053] Then, for any two context copy semantic detection logic rules semantic embedding encoding vector and and its corresponding distance eigenvector, for example, and , use the inner product bounded linear overlap operator method in the distance space to calculate its Hilbert-Schmidt type normed distance number: ;in, represents the vector inner product, and represents the vector two norm, represents matrix addition, Represents the Hilbert-Schmidt normed distance number, thereby evaluating the similarity embedding quality of the semantic intrinsic endogenous space in the way of low-dimensional manifold similarity in high-dimensional distance space, so as to obtain a neighborhood vector with a high correlation with the semantic embedding encoding vector of the semantic detection logic rule of the specific context copywriting, and then optimize it through neighborhood weighting, that is: ;in, is a neighborhood vector that meets the conditions The number of is the distance hyperparameter, Represents the optimized contextual copy semantic detection logic rule semantic embedding encoding vector. Finally, each optimized contextual copy semantic detection logic rule semantic embedding encoding vector is input into the rule autocorrelation significance measurement network.

[0054] Exemplarily, in step S5, each target distribution platform in the list of target distribution platforms is connected to automatically publish the adapted advertising content. Specifically, the adapted advertising content is submitted directly to the corresponding platform using the APIs provided by each platform. For example, for Facebook and Instagram, images and text can be uploaded in bulk using the Facebook Graph API; while for Twitter, tweets can be published using its Tweet Posting API. Similarly, for e-commerce platforms, advertising content can be published to product detail pages or store homepages by calling the Amazon Product Advertising API or related interfaces on the Taobao Open Platform. The system's automation capabilities are particularly important in this process. Pre-configured scripts and task scheduling mechanisms allow publishing tasks to be automatically executed within a specified time period without manual intervention. For example, advertisers can choose to concentrate advertising within a specific time period to maximize exposure. Advertising content is pushed to each platform sequentially according to a preset schedule, and the publishing status is monitored to ensure that each ad is successfully launched. Furthermore, the published advertising content is continuously monitored and feedback is collected. For example, some platforms provide detailed advertising performance data, such as click-through rate and conversion rate. This data can be captured regularly and reports generated for advertisers’ reference. If it is found that certain ads are not performing as expected, the ad content can be further optimized based on the feedback data to improve its dissemination effect.

[0055] This application also provides an advertising information management system for implementing the above-mentioned advertising information management method. Specifically, Figure 6 The diagram shows a schematic diagram of the structure of the advertising information management system according to an embodiment of the present application. Figure 6 As shown, the advertising publishing information management system 600 includes: a user uploaded content acquisition module 610, which is used to obtain the advertising content uploaded by the user and the list of target distribution platforms; a platform content specification requirement extraction module 620, which is used to extract and record the platform content specification requirements of each target distribution platform; an advertising content semantic detection module 630, which is used to perform semantic detection on the advertising content based on the platform content specification requirements of each target distribution platform to obtain a semantic detection result; an advertising content adaptation enhancement module 640, which is used to perform content adaptation enhancement on the advertising content to obtain adapted advertising content in response to the semantic detection result that the text part is not adapted; an advertising content automatic publishing module 650, which is used to connect to each target distribution platform in the list of the target distribution platform to realize the automatic publishing of the adapted advertising content.

[0056] Here, those skilled in the art will appreciate that the specific operations of each module in the above-mentioned advertising information management system have been described in detail in the previous reference. Figures 1 to 5 The description of the advertising publishing information management method has been introduced in detail, and therefore, its repeated description will be omitted.

[0057] An embodiment of the present application also provides a computer-readable storage medium, which stores computer program code. When the computer program code runs on a computer, the computer executes the above-mentioned related method steps to implement an advertising publishing information management method provided by the above-mentioned embodiment.

[0058] The embodiment of the present application further provides a computer program product. When the computer program product is run on a computer, the computer is caused to execute the above-mentioned related steps to implement an advertisement publishing information management method provided by the above-mentioned embodiment.

[0059] Among them, the system, computer-readable storage medium or computer program product provided in the embodiments of the present application are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0060] It should be noted that the order of the above embodiments of the present application is only for description and does not represent the advantages or disadvantages of the embodiments.

[0061] The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous. The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referenced to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for managing advertising information, characterized in that: include: Obtain a list of advertising content uploaded by users and target distribution platforms; Extract and record the platform content specification requirements of each target distribution platform; Based on the platform content specification requirements of each target distribution platform, performing semantic detection on the advertisement content to obtain a semantic detection result; In response to the semantic detection result indicating that the copy is partially unmatched, performing content adaptation and enhancement on the advertisement content to obtain adapted advertisement content, wherein the content adaptation and enhancement on the advertisement content is performed based on a semantic similarity incentive of a set of copy semantic detection logic rules required by the platform content specification; Connecting to each target distribution platform in the list of target distribution platforms to achieve automatic publishing of the adapted advertising content; The content adaptation enhancement of the advertisement content to obtain adapted advertisement content includes: extracting the set of text semantic detection logic rules; performing semantic embedding coding on the text part in the advertisement content to obtain an original text part semantic embedding coding vector; performing semantic embedding coding on each text semantic detection logic rule to obtain a set of text semantic detection logic rule semantic embedding coding vectors; performing rule semantic saliency modulation on the set of text semantic detection logic rule semantic embedding coding vectors to obtain a set of saliency modulated text semantic detection logic rule semantic embedding coding vectors; inputting the set of original text part semantic embedding coding vectors and the saliency modulated text semantic detection logic rule semantic embedding coding vectors into a text adapter based on a large language model to obtain the adapted advertisement content; Among them, the set of the text semantic detection logic rule semantic embedding coding vectors is subjected to rule semantic saliency modulation to obtain a set of saliency modulated text semantic detection logic rule semantic embedding coding vectors, including: calculating the rule autocorrelation saliency weight value of each text semantic detection logic rule semantic embedding coding vector in the set of the text semantic detection logic rule semantic embedding coding vectors to obtain a set of rule autocorrelation saliency weight values; based on the set of rule autocorrelation saliency weight values, the set of the text semantic detection logic rule semantic embedding coding vectors is subjected to saliency modulation to obtain the set of saliency modulated text semantic detection logic rule semantic embedding coding vectors.

2. The advertising information management method according to claim 1, characterized in that: Based on the platform content specification requirements of each target distribution platform, semantic detection is performed on the advertisement content to obtain a semantic detection result, including: Intelligently classifying the advertisement content to obtain a text portion, an image portion, and a video portion; Extracting text semantic detection logic rules from the platform content specification requirements to obtain a set of text semantic detection logic rules; The text parts are verified one by one using the set of text semantic detection logic rules to obtain the semantic detection result.

3. The advertising information management method according to claim 2, characterized in that: Extracting text semantic detection logic rules from the platform content specification requirements to obtain a set of text semantic detection logic rules, including: Access the interface of the target distribution platform and pull the latest copywriting semantic detection logic rules from the target distribution platform; The text semantic detection logic rule is updated based on the latest text semantic detection logic rule.

4. The advertising information management method according to claim 3, characterized in that: Calculating the rule autocorrelation significant weight value of each text semantic detection logic rule semantic embedding coding vector in the set of text semantic detection logic rule semantic embedding coding vectors to obtain a set of rule autocorrelation significant weight values, including: Inputting the set of semantic embedding coding vectors of the text semantic detection logic rule into a semantic detection logic rule context encoder based on a transformer model to obtain a set of semantic embedding coding vectors of contextual text semantic detection logic rule; Each contextual copy semantic detection logic rule semantic embedding encoding vector in the set of contextual copy semantic detection logic rule semantic embedding encoding vectors is input into a rule autocorrelation significance measurement network to obtain a set of rule autocorrelation significance weight values.

5. The advertising information management method according to claim 4, characterized in that: Inputting each context copy semantic detection logic rule semantic embedding encoding vector in the set of context copy semantic detection logic rule semantic embedding encoding vectors into a rule autocorrelation significance measurement network to obtain a set of rule autocorrelation significance weight values, including: The rule autocorrelation significance measurement network calculates the rule autocorrelation significance weight value of each context copy semantic detection logic rule semantic embedding coding vector using the following formula: ;in, Represents the contextual copywriting semantic detection logic rule semantic embedding encoding vector, is the auxiliary reference weight vector for saliency measurement, is the linear connection weight matrix of the saliency measure, representing matrix multiplication, is the saliency measure bias vector, Indicates the significant weight value of the regular autocorrelation.

6. The advertising information management method according to claim 4, characterized in that: Also includes: Before inputting each contextual copy semantic detection logic rule semantic embedding coding vector in the set of the contextual copy semantic detection logic rule semantic embedding coding vector into the rule autocorrelation significance measurement network, semantic similarity excitation is pre-performed on each contextual copy semantic detection logic rule semantic embedding coding vector.

7. An advertisement publishing information management system, used to implement the advertisement publishing information management method according to any one of claims 1 to 6, characterized in that: include: A user uploaded content acquisition module is used to obtain the advertising content uploaded by the user and the list of target distribution platforms; Platform content specification requirements extraction module, used to extract and record the platform content specification requirements of each target distribution platform; An advertisement content semantic detection module, configured to perform semantic detection on the advertisement content based on the platform content specification requirements of each target distribution platform to obtain a semantic detection result; an advertisement content adaptation and enhancement module, configured to, in response to the semantic detection result indicating that the text portion is not adapted, perform content adaptation and enhancement on the advertisement content to obtain adapted advertisement content; The automatic advertising content publishing module is used to connect to each target distribution platform in the list of target distribution platforms to realize the automatic publishing of the adapted advertising content.

Images