Cross-platform advertisement content generation method and device, equipment and storage medium

Abstract

The invention provides a cross-platform advertisement content generation method and device, equipment and a storage medium, and the method comprises the steps: obtaining a current to-be-processed advertisement content copywriting, and determining the content feature information of a to-be-published platform corresponding to the advertisement content copywriting; generating to-be-adapted advertisement content corresponding to the advertisement content copywriting, wherein the modal features of the to-be-adapted advertisement content are aligned; and performing format conversion on the to-be-adapted advertisement content based on the content feature information to generate to-be-published advertisement content. According to the scheme, the production efficiency of the cross-platform advertisement content can be remarkably improved, the content better conforms to the platform characteristics of all platforms, the quality of the generated advertisement content is better, and the image-text collaboration is higher.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, device and storage medium for generating cross-platform advertising content. Background Technology

[0002] With the rapid development of artificial intelligence and digital media technologies, the intelligent generation and optimized delivery of advertising content has become an important research direction in the field of digital marketing. In the advertising process, due to the existence of numerous advertising platforms, it is necessary to place ads on multiple different platforms to improve advertising efficiency.

[0003] In related technologies, cross-platform ad content creation typically involves manual adaptation and adjustment, such as adjusting the layout of videos and images to suit the specific platform. However, each platform has different characteristics, and the adjustment process heavily relies on the personnel's understanding of platform features, requiring relatively extensive operational experience, which is detrimental to the efficient cross-platform ad content creation. Furthermore, the resulting text and images exhibit poor synergy, such as information fragmentation between ad text and images. Additionally, cultural differences between platforms drastically increase the difficulty of cross-platform ad content adaptation and adjustment. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for generating cross-platform advertising content, which solves the problems of low production efficiency and poor text-image coordination in the production results of cross-platform advertising content in related technologies. By determining the content feature information of the platform to be published, and performing adaptation processing on the generated advertising content that aligns with the various modal features based on the content feature information, the production efficiency of cross-platform advertising content can be significantly improved, the content is more in line with the platform characteristics of each platform, the quality of the generated advertising content is better, and the text-image coordination is stronger.

[0005] Firstly, this application provides a cross-platform advertising content generation method, including: Obtain the currently pending advertising content copy and determine the content feature information of the platform to be published corresponding to the advertising content copy; Generate the ad content to be adapted, which is aligned with each modal feature corresponding to the ad content copy; Based on the content feature information, the format of the advertisement content to be adapted is converted to generate the advertisement content to be published.

[0006] Optionally, determining the content feature information of the platform to be published corresponding to the advertising content copy includes: Determine the static content characteristics of the platform to be published corresponding to the advertising content copy. The static content characteristics include any one or more of video features, text features, image features, and audio features. Obtain historically published videos that meet preset conditions from the platform to be published, and determine the dynamic content characteristics of the historically published videos; Based on the static content features and the dynamic content features, a platform feature map is constructed to obtain content feature information.

[0007] Optionally, generating the ad content to be adapted, aligned with each modal feature corresponding to the ad content copy, includes: Semantic parsing of the advertising content yields multiple text semantic feature vectors of different dimensions; A cross-modal feature mapping matrix is ​​constructed based on the text semantic feature vector. The cross-modal feature mapping matrix represents the transformation rules from the text semantic feature vector to multiple different modalities. The cross-modal feature mapping matrix records the feature control codes corresponding to each modality. Based on the feature control codes corresponding to each modality and the corresponding modality generator, the content information of the corresponding modality is generated, and the content information corresponding to each modality is aligned to obtain the advertising content to be adapted.

[0008] Optionally, the format conversion of the advertisement content to be adapted based on the content feature information includes: Identify the key entities of the advertising content to be adapted; Based on the content feature information, the layout of the advertisement content to be adapted is transformed and the position of the key subject is adjusted for adaptation.

[0009] Optionally, the key entities for determining the advertising content to be adapted include: Extract the visually salient regions and textual semantic focus from the advertising content to be adapted. The visually salient regions include one or more visual elements, and the textual semantic focus includes one or more keywords. Calculate the weight scores of one or more of the visual elements and one or more of the keywords, and determine the location area of ​​the visual element or keyword with the highest weight score as the key subject.

[0010] Optionally, the step of performing layout transformation on the advertisement content to be adapted and positional adjustment of the key subject based on the content feature information includes: A layout strategy template is generated based on the static and dynamic content features of the platform feature map. The layout strategy template includes layout parameter information, canvas partition definition, visual flow rules, suggested element types for each area, and recommended position range for the main body. Based on the layout strategy template, the layout of the advertisement content to be adapted is transformed, and the position of the key subject is adapted and adjusted.

[0011] Optionally, the step of performing layout transformation of the advertisement content to be adapted and position adaptation adjustment of the key subject based on the layout strategy template includes: Calculate the matching degree between the key subject and other elements of the advertisement content to be adapted in different subject recommendation position intervals; The main layout position of the key subject is determined by the recommended position range of the subject with the highest matching degree, and the secondary layout positions of other elements are determined based on the canvas partition definition, visual flow rules and suggested element types of each area. Using the layout parameter information as a constraint, the elements of the advertisement content to be adapted are laid out to the corresponding main layout position and the secondary layout position.

[0012] Secondly, this application provides a cross-platform advertising content generation device, comprising: The acquisition module is used to acquire the text of the advertisement currently awaiting processing. The feature determination module is used to determine the content feature information of the platform to be published corresponding to the advertising content copy. The content generation module is used to generate the advertising content to be adapted, which is aligned with the various modal features corresponding to the advertising content copy. The content adaptation module is used to convert the format of the advertisement content to be adapted based on the content feature information, and generate the advertisement content to be published.

[0013] Thirdly, this application also provides a cross-platform advertising content generation device, the device comprising: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the cross-platform advertising content generation method as described in any of the preceding first aspects.

[0014] Fourthly, this application also provides a storage medium for storing computer-executable instructions, which, when executed by a computer processor, are used to perform the cross-platform advertising content generation method as described in any of the first aspects above.

[0015] In the solution provided in this application embodiment, the current advertising content text to be processed is obtained, and the content feature information of the platform to be published corresponding to the advertising content text is determined. Then, the advertising content to be adapted, which is aligned with each modal feature of the advertising content text, is generated. Then, the format of the advertising content to be adapted is converted based on the content feature information to generate the advertising content to be published. Since each modal feature is kept aligned during the content generation process, the text-image synergy is better. Moreover, the corresponding format conversion is performed based on the determined content feature information of the platform to be published, which can efficiently complete the generation of cross-platform advertising content. This can significantly improve the production efficiency of cross-platform advertising content, make the content more in line with the platform characteristics of each platform, and generate higher quality advertising content. Attached Figure Description

[0016] Figure 1 This is a flowchart of a cross-platform advertising content generation method provided in an embodiment of this application; Figure 2 This is a flowchart illustrating a method for generating corresponding advertising content to be adapted, as provided in an embodiment of this application. Figure 3 This is a flowchart of another advertising content generation method provided in the embodiments of this application, such as... Figure 3 As shown; Figure 4 This is a block diagram of the module structure of a cross-platform advertising content generation device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a cross-platform advertising content generation device provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as being processed sequentially, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. A process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0018] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0019] Cross-platform advertising can generally be understood as relatively complex digital advertising. Currently, cross-platform adaptation for digital ads requires repeated manual adjustments, such as generating vertical videos for platform A or designing separate graphic cards for platform B. Each platform has distinct characteristics and target audiences, making adjustments heavily reliant on the operational experience of the personnel making the adjustments. For example, content created by novice optimization specialists typically has a click-through rate 38% lower than the industry average. Furthermore, this manual generation method suffers from poor text-image coordination; for instance, authoritative research shows that 62% of ad text and images exhibit information fragmentation. In addition, the styles and formats of content published on different platforms vary, and the cultural differences among audiences are significant. Directly applying this to multiple different platforms can lead to a substantial drop in conversion rates. Therefore, this application provides a cross-platform advertising content generation method to address the aforementioned problems.

[0020] The cross-platform advertising content generation method provided in this application embodiment can be executed by a device with computing power, such as a server, laptop, or desktop computer.

[0021] Figure 1 This is a flowchart illustrating a cross-platform advertising content generation method provided in an embodiment of this application. Figure 1 As shown, this cross-platform ad content generation method includes: Step S101: Obtain the current advertising content copy to be processed, and determine the content feature information of the platform to be published corresponding to the advertising content copy.

[0022] The advertising copy defines the core message the advertisement aims to convey. For example, in a "summer sunscreen" advertisement, the copy might be "Refreshing and weightless." More complex cases may consist of text elements such as a headline, body text, slogan, and keyword list. Subsequently, based on this advertising copy, adapted advertising content is generated and formatted to obtain the final, publishable advertising content.

[0023] The "platform to be published" refers to the digital media channels or applications where advertising content is planned to be placed. Different platforms, such as social media, search engines, video websites, and news applications, have unique technical architectures, user behavior patterns, and content ecosystems. Content feature information is a structured description of the rules, constraints, and preferences of the platforms to be published on regarding content presentation and dissemination. This includes, but is not limited to, supported media formats (such as video aspect ratio, image resolution), text limitations (character count, title length), interactive functions (buttons, links, hashtags), algorithm recommendation preferences (content type, publication time, interaction metrics), and visual style. Optionally, corresponding content feature information is pre-set for each publishing platform. Once the platforms to be published for the advertising content copy are determined, this content feature information can be directly obtained.

[0024] Step S102: Generate the advertising content to be adapted, which is aligned with the modal features corresponding to the advertising content copy.

[0025] In this context, "modality" refers to the form in which information is presented, such as text, images, audio, and video. Modal feature alignment means maintaining consistency and harmony in the expression of core advertising information, brand elements, visual style, and emotional tone across different modalities. The generated ad content to be adapted is the content after alignment of all modal features, thereby achieving harmony between text and images.

[0026] Optionally, a multimodal contrastive learning approach based on the CLIP (Contrastive Language-Image Pre-training) model can be used to generate advertising copy, resulting in advertising content aligned with features across different modalities. The CLIP model learns from large-scale image and text pairs to achieve deep association between images and text in a unified semantic space. It employs a dual-encoder architecture, including independent image and text encoders. The model optimizes through contrastive learning, bringing matching image and text pairs closer together in the vector space, while disjointed pairs are moved further apart. This achieves image-text harmony in the advertising content. For example, for an advertisement for "summer sunscreen," the copy "refreshing and weightless" corresponds to an image showing water droplets sliding off the sunscreen surface.

[0027] Step S103: Based on the content feature information, convert the format of the advertising content to be adapted to generate the advertising content to be published.

[0028] After generating the advertising content to be adapted, the format is converted according to the determined content feature information of the platform to generate advertising content that meets the corresponding requirements and satisfies the characteristics of the platform, so as to be published on the platform to be published.

[0029] In one embodiment, an adaptive layout engine can be used to convert the format of the ad content to be adapted, generating the ad content to be published. This adaptive layout engine uses an adversarial generative network to maintain the visual focus of core elements at different proportions, ensuring that the key product logo is always positioned at the visual golden ratio point after conversion.

[0030] As described above, by acquiring the current ad content copy to be processed and determining the content feature information of the platform to be published corresponding to the ad content copy, the system generates ad content to be adapted with each modal feature aligned with the ad content copy. Then, based on the content feature information, the format of the ad content to be adapted is converted to generate the ad content to be published. Since each modal feature is kept aligned during the content generation process, the text and image synergy is better. Furthermore, by performing corresponding format conversion based on the determined content feature information of the platform to be published, the system can efficiently complete the generation of cross-platform ad content, significantly improve the production efficiency of cross-platform ad content, make the content more in line with the platform characteristics of each platform, and generate higher quality ad content.

[0031] Figure 2 This is a flowchart illustrating a method for generating corresponding advertising content to be adapted, as provided in an embodiment of this application. Figure 2 As shown, it includes: Step S201: Obtain the current ad content copy to be processed and determine the content feature information of the platform to be published corresponding to the ad content copy.

[0032] Step S202: Semantically analyze the advertising content copy to obtain multiple text semantic feature vectors of different dimensions. Construct a cross-modal feature mapping matrix based on the text semantic feature vectors. Generate content information of the corresponding modality based on the feature control code and the corresponding modality generator for each modality. Align the content information of each modality to obtain the advertising content to be adapted.

[0033] In one embodiment, for complex and detailed advertising copy, such as advertising copy containing fields like main headlines, body text, calls to action, keyword lists, and brand tone descriptions, preprocessing is first performed. This includes standardizing and cleaning the text, such as removing irrelevant characters, correcting spelling, and standardizing terminology, and converting it into a uniform text sequence that can be processed by the generative model. This uniform text sequence is then input into a multi-level semantic encoder framework, which performs concept-level encoding, syntactic / structural-level encoding, deep semantic encoding, and style / sentiment encoding in parallel or sequentially. The process involves several key layers: Conceptual encoding utilizes keyword / entity recognition, topic modeling, or specific concept extraction models to extract discrete concepts such as core product characteristics, user benefits, scenarios, and emotional polarities from the text, encoding them as conceptual feature vectors; Syntactic / structural encoding analyzes the sentence structure, rhetorical devices, and paragraph logic of the text using language models, encoding stylistic and structural feature vectors; Deep semantic encoding uses large-scale pre-trained language models (such as the encoder parts of BERT and GPT) to obtain context-sensitive overall semantic embedding vectors; Style / sentiment encoding, based on brand tone descriptions and text content, uses style classification models or sentiment analysis models to encode feature vectors representing emotional tone and brand style. The outputs from these multiple dimensions are then normalized to form a structured set of text semantic feature vectors, resulting in multiple text semantic feature vectors with different dimensions.

[0034] Optionally, the system maintains a predefined shared semantic space. After training, this space can map information from different modalities (text, images, audio, etc.) to the same set of comparable latent semantic dimensions. For each obtained text semantic feature vector, a pre-trained cross-modal mapping function is used to convert it into the corresponding feature control code of the target modality in the shared semantic space. Then, all the converted feature control codes are merged to generate a two-dimensional cross-modal feature mapping matrix. The rows of the matrix correspond to different semantic dimensions (concepts, styles, sentiments, etc.), and the columns correspond to different target modalities, such as images, videos, audio, and motion effects. Each element C_{i,j} in the matrix is ​​a specific vector, representing the core control signal that drives the generation of content in the j-th modality based on the "i-th text semantic dimension".

[0035] Optionally, for each target modality (e.g., image modality), all columns corresponding to that modality are extracted from the cross-modal feature mapping matrix, and the feature control codes of multiple dimensions are fused (e.g., concatenated, weighted summation) to form a comprehensive feature control code for that modality. In one embodiment, a conditional generative model is pre-set as the generator for each supported modality (e.g., image diffusion model, video generation model, audio synthesis model, etc.), and the obtained comprehensive feature control code is input to the corresponding modality generator to synthesize content information. After obtaining the content information of each modality, consistency detection and evaluation are performed. Specifically, a cross-modal understanding model (e.g., CLIP model) can be used to calculate the cross-modal consistency score of the content information of each modality, which may include specific scores for semantic consistency, temporal synchronization, and style uniformity. If the consistency score is lower than the preset threshold, a feedback optimization loop is executed. For example, based on the dimension of inconsistency, the strength or content of the corresponding feature control code C_{i,j} in the cross-modal feature mapping matrix is ​​adjusted in reverse. The adjusted feature control code is then re-input into the modality generator for local regeneration. Correspondingly, when all modal content passes the consistency evaluation or reaches the maximum number of optimization iterations, the content information corresponding to each modality is aligned to obtain the ad content to be adapted.

[0036] Step S203: Based on the content feature information, convert the format of the advertising content to be adapted to generate the advertising content to be published.

[0037] As described above, by acquiring the current ad content copy to be processed and determining the content feature information of the platform to be published, the ad content copy is then semantically parsed to obtain multiple text semantic feature vectors of different dimensions. A cross-modal feature mapping matrix is ​​constructed based on the text semantic feature vectors. Content information of the corresponding modality is generated based on the feature control codes and modality generators corresponding to each modality. The content information corresponding to each modality is aligned to obtain the ad content to be adapted. Then, the format of the ad content to be adapted is converted based on the content feature information to generate the ad content to be published. Since the modal features are kept aligned during the content generation process, the text-image collaboration is better. Furthermore, the format conversion is performed based on the determined content feature information of the platform to be published, which can efficiently complete the generation of cross-platform ad content. This can significantly improve the production efficiency of cross-platform ad content, make the content more in line with the platform characteristics of each platform, and generate higher quality ad content.

[0038] Figure 3 This is a flowchart of another advertising content generation method provided in the embodiments of this application, such as... Figure 3 As shown, it includes: Step S301: Obtain the current advertising content copy to be processed, determine the static content features of the platform to be published corresponding to the advertising content copy, obtain the historical published videos that meet the preset conditions in the platform to be published, determine the dynamic content features of the historical published videos, and construct a platform feature map based on the static content features and dynamic content features to obtain content feature information.

[0039] Among them, static content features include any one or more of video features, text features, image features, and audio features. Historically published videos that meet the preset conditions can be viral videos with clicks, plays, shares, etc. that exceed the preset number. Their corresponding features are dynamic content features, such as video rhythm features, narrative structure features, text style features, audio mode features, etc.

[0040] In one embodiment, the static content features of the platform to be published can be obtained from the static feature rule base of each platform, which is maintained in real time. This static feature rule base is a continuously maintained knowledge base that stores explicit rules and hard constraints extracted and parsed from the official developer documents, API specifications, and advertising policies of each platform. Static content features are generated by extracting data from this static feature rule base and processing it in a structured manner. For example, it includes format specifications (video encoding format (H.264 / AV1), resolution range, aspect ratio requirements, file size and duration limits, cover image size, supported audio encoding), content constraints (character limit for text fields, list of prohibited words, compliance requirements), etc.

[0041] In one embodiment, historically published videos that meet certain conditions, i.e., viral videos, are subjected to automated analysis to extract their inherent, non-written features. Specifically, this may involve using computer vision models to analyze the frequency of shot transitions, scene transformation patterns, and the time-varying curves of the subject's motion intensity; identifying common narrative templates through video subtitles (or speech recognition text) and keyframe sequences; analyzing the type, rhythm (BPM), and volume variation patterns of background music, as well as the characteristics of human speech rate and tone; and performing natural language processing on the titles and descriptions of highly interactive videos to summarize syntactic features and sentiment tendencies. Correspondingly, after obtaining the static and dynamic content features of the platform to be published, a platform feature map is constructed to obtain content feature information for subsequent conversion of advertising content formats. Optionally, the static and dynamic content features can be transformed into interconnected "nodes," and then strong correlations between nodes can be established through algorithmic analysis (e.g., cosine similarity, Bayesian networks, sequence pattern mining, LSTM / Transformer, etc.).

[0042] Step S302: Generate the advertising content to be adapted, which is aligned with the modal features corresponding to the advertising content copy.

[0043] Step S303: Determine the key subject of the ad content to be adapted, and perform layout transformation and position adaptation adjustment of the key subject of the ad content to be adapted based on the content feature information to generate the ad content to be published.

[0044] In one embodiment, when performing format conversion of the advertising content to be adapted, the key subject of the advertising content to be adapted is first determined. Optionally, one method of determination may be: extracting visually salient regions and textual semantic focus from the advertising content to be adapted, wherein the visually salient regions include one or more visual elements, and the textual semantic focus includes one or more keywords; then calculating the weight scores of one or more visual elements and one or more keywords, and determining the location region of the visual element or keyword with the highest weight score as the key subject. For the extraction of visually salient regions, saliency detection algorithms in computer vision can be used, such as the SAM algorithm based on deep learning to automatically identify the most eye-catching region in an image / video frame; the extraction of textual semantic focus can be achieved through natural language processing technology, combined with keyword extraction to identify words and phrases expressing core content in the text. The weight scores of visual elements can be calculated by: for each visual element, using a regional saliency scoring model, combining the element's spatial location, size, overlap with the visual focus, color contrast, and other features. The weighting of keywords can be achieved by using a text importance assessment model to score keywords based on their frequency in the text, position (e.g., in the title), part of speech (e.g., noun / verb), and semantic relevance to the core theme. Finally, the visual element or the area containing the keyword with the highest weighting score is identified as the key element.

[0045] In one embodiment, after identifying the key subject, the subsequent location adaptation adjustment can be performed as follows: A layout strategy template is generated based on the static and dynamic content features of the platform feature map. This layout strategy template includes layout parameter information, canvas partition definitions, visual flow rules, suggested element types for each region, and recommended subject position ranges. Then, the layout conversion of the advertising content to be adapted and the location adaptation adjustment of the key subject are performed based on this layout strategy template. Optionally, when generating the layout strategy template, after querying the platform feature map of the platform to be published, the rules corresponding to the nodes of the static content features are extracted to directly define the basic canvas parameters and partitions. Simultaneously, the records corresponding to the nodes of the associated dynamic content features, such as the visual hotspot distribution and element combination rules of highly interactive videos, are retrieved. These records are then converted into visual flow rules, suggested element types for each region, and recommended subject position ranges using graph inference algorithms (e.g., GNN-based algorithms, Node2Vec-based algorithms based on paths / walks), to obtain the layout strategy template.

[0046] In one embodiment, the method for adapting the layout of the ad content to be adapted and adjusting the position of the key subject based on the layout strategy template can be as follows: The matching degree between the key subject and other elements of the ad content to be adapted is calculated for different subject recommendation position ranges. The subject recommendation position range with the highest matching degree is determined as the main layout position of the key subject. The secondary layout positions of other elements are determined based on the canvas partition definition, visual flow rules, and suggested element types for each area. Using layout parameter information as constraints, the elements of the ad content to be adapted are laid out to the corresponding main and secondary layout positions. This yields the final ad content to be published. Optionally, a pre-trained matching evaluation model (such as the visual aesthetics and composition evaluation models A-LAMP and NIMA; and the visual relationship detection models VCTree and GPS-Net) can be used to render the key subject sequentially onto typical coordinates within each subject's recommended location range. The matching evaluation model then comprehensively calculates the visual integration, compositional balance, and semantic association strength between the subject and other elements such as background, text, and logo at that location, outputting a comprehensive matching score for each candidate range. The subject's recommended location range with the highest matching degree (i.e., the highest comprehensive matching score) is determined as the main layout location for the key subject. Subsequently, using this key subject's location as a fixed anchor point, and based on "visual flow rules" (such as Z-shaped and F-shaped paths) and "canvas partitioning definitions," a constraint satisfaction problem solver determines appropriate secondary layout locations for each other element, ensuring that the element types of the final generated advertisement conform to the regional recommendations and that the overall composition is harmonious.

[0047] As described above, by acquiring the current ad content copy to be processed and determining the content feature information of the platform to be published corresponding to the ad content copy, the system generates ad content to be adapted with each modal feature aligned with the ad content copy. Then, based on the content feature information, the format of the ad content to be adapted is converted to generate the ad content to be published. Since each modal feature is kept aligned during the content generation process, the text and image synergy is better. Furthermore, by performing corresponding format conversion based on the determined content feature information of the platform to be published, the system can efficiently complete the generation of cross-platform ad content, significantly improve the production efficiency of cross-platform ad content, make the content more in line with the platform characteristics of each platform, and generate higher quality ad content.

[0048] Figure 4 This is a block diagram of the module structure of a cross-platform advertising content generation device provided in this application embodiment. This device is used to execute a cross-platform advertising content generation method provided in the above embodiment, and has corresponding functional modules and beneficial effects for executing the method. Figure 4 As shown, the device specifically includes: The acquisition module 101 is used to acquire the currently pending advertisement content text; Feature determination module 102 is used to determine the content feature information of the platform to be published corresponding to the advertising content copy; Content generation module 103 is used to generate advertising content to be adapted, which is aligned with each modal feature corresponding to the advertising content copy. The content adaptation module 104 is used to perform format conversion on the advertising content to be adapted based on the content feature information, and generate advertising content to be published.

[0049] As can be seen from the above solution, by obtaining the current ad content copy to be processed and determining the content feature information of the platform to be published corresponding to the ad content copy, the ad content to be adapted, which is aligned with the modal features of the ad content copy, is then generated. Based on the content feature information, the format of the ad content to be adapted is converted to generate the ad content to be published. Since the modal features are kept aligned during the content generation process, the text and image synergy is better. Furthermore, the corresponding format conversion based on the determined content feature information of the platform to be published can efficiently complete the generation of cross-platform ad content, significantly improve the production efficiency of cross-platform ad content, make the content more in line with the platform characteristics of each platform, and generate higher quality ad content.

[0050] In one possible embodiment, determining the content feature information of the platform to be published corresponding to the advertising content copy includes: Determine the static content characteristics of the platform to be published corresponding to the advertising content copy. The static content characteristics include any one or more of video features, text features, image features, and audio features. Obtain historically published videos that meet preset conditions from the platform to be published, and determine the dynamic content characteristics of the historically published videos; Based on the static content features and the dynamic content features, a platform feature map is constructed to obtain content feature information.

[0051] In one possible embodiment, the content generation module 103 is specifically used for: Semantic parsing of the advertising content yields multiple text semantic feature vectors of different dimensions; A cross-modal feature mapping matrix is ​​constructed based on the text semantic feature vector. The cross-modal feature mapping matrix represents the transformation rules from the text semantic feature vector to multiple different modalities. The cross-modal feature mapping matrix records the feature control codes corresponding to each modality. Based on the feature control codes corresponding to each modality and the corresponding modality generator, the content information of the corresponding modality is generated, and the content information corresponding to each modality is aligned to obtain the advertising content to be adapted.

[0052] In one possible embodiment, the content adaptation module 104 is specifically used for: Identify the key entities of the advertising content to be adapted; Based on the content feature information, the layout of the advertisement content to be adapted is transformed and the position of the key subject is adjusted for adaptation.

[0053] In one possible embodiment, the content adaptation module 104 is specifically used for: Extract the visually salient regions and textual semantic focus from the advertising content to be adapted. The visually salient regions include one or more visual elements, and the textual semantic focus includes one or more keywords. Calculate the weight scores of one or more of the visual elements and one or more of the keywords, and determine the location area of ​​the visual element or keyword with the highest weight score as the key subject.

[0054] In one possible embodiment, the content adaptation module 104 is specifically used for: A layout strategy template is generated based on the static and dynamic content features of the platform feature map. The layout strategy template includes layout parameter information, canvas partition definition, visual flow rules, suggested element types for each area, and recommended position range for the main body. Based on the layout strategy template, the layout of the advertisement content to be adapted is transformed, and the position of the key subject is adapted and adjusted.

[0055] In one possible embodiment, the content adaptation module 104 is specifically used for: Calculate the matching degree between the key subject and other elements of the advertisement content to be adapted in different subject recommendation position intervals; The main layout position of the key subject is determined by the recommended position range of the subject with the highest matching degree, and the secondary layout positions of other elements are determined based on the canvas partition definition, visual flow rules and suggested element types of each area. Using the layout parameter information as a constraint, the elements of the advertisement content to be adapted are laid out to the corresponding main layout position and the secondary layout position.

[0056] Figure 5 A schematic diagram of a cross-platform advertising content generation device provided in this application embodiment is shown below. Figure 5 As shown, the device includes a processor 201, a memory 202, an input device 203, and an output device 204; the number of processors 201 in the device can be one or more. Figure 5Taking a processor 201 as an example; the processor 201, memory 202, input device 203, and output device 204 in the device can be connected via a bus or other means. Figure 5 Taking a bus connection as an example, the memory 202, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions or modules corresponding to a cross-platform advertising content generation method in this embodiment. The processor 201 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 202, thereby realizing the aforementioned cross-platform advertising content generation method. The input device 203 can be used to receive input digital or character information and generate key signal inputs related to user settings and function control of the device. The output device 204 may include a display screen or other display device.

[0057] This application also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a cross-platform advertising content generation method, the method comprising: Obtain the currently pending advertising content copy and determine the content feature information of the platform to be published corresponding to the advertising content copy; Generate the ad content to be adapted, which is aligned with each modal feature corresponding to the ad content copy; Based on the content feature information, the format of the advertisement content to be adapted is converted to generate the advertisement content to be published.

[0058] It is worth noting that in the above-described embodiment of a cross-platform advertising content generation method system, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this application.

[0059] Note that the above are merely preferred embodiments and the technical principles applied in this application. Those skilled in the art will understand that the embodiments of this application are not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the embodiments of this application. Therefore, although the embodiments of this application have been described in detail through the above embodiments, the embodiments of this application are not limited to the above embodiments. More other equivalent embodiments may be included without departing from the concept of the embodiments of this application, and the scope of the embodiments of this application is determined by the scope of the appended claims.

Claims

1. A method for generating cross-platform advertising content, characterized in that, include: Obtain the currently pending advertising content copy and determine the content feature information of the platform to be published corresponding to the advertising content copy; Generate the ad content to be adapted, which is aligned with each modal feature corresponding to the ad content copy; Based on the content feature information, the format of the advertisement content to be adapted is converted to generate the advertisement content to be published.

2. The cross-platform advertising content generation method according to claim 1, characterized in that, The step of determining the content feature information of the platform to be published corresponding to the advertising content copy includes: Determine the static content characteristics of the platform to be published corresponding to the advertising content copy. The static content characteristics include any one or more of video features, text features, image features, and audio features. Obtain historically published videos that meet preset conditions from the platform to be published, and determine the dynamic content characteristics of the historically published videos; Based on the static content features and the dynamic content features, a platform feature map is constructed to obtain content feature information.

3. The cross-platform advertising content generation method according to claim 1, characterized in that, The process of generating the ad content to be adapted, aligned with each modal feature corresponding to the ad content copy, includes: Semantic parsing of the advertising content yields multiple text semantic feature vectors of different dimensions; A cross-modal feature mapping matrix is ​​constructed based on the text semantic feature vector. The cross-modal feature mapping matrix represents the transformation rules from the text semantic feature vector to multiple different modalities. The cross-modal feature mapping matrix records the feature control codes corresponding to each modality. Based on the feature control codes corresponding to each modality and the corresponding modality generator, the content information of the corresponding modality is generated, and the content information corresponding to each modality is aligned to obtain the advertising content to be adapted.

4. The cross-platform advertising content generation method according to claim 2, characterized in that, The format conversion of the advertisement content to be adapted based on the content feature information includes: Identify the key entities of the advertising content to be adapted; Based on the content feature information, the layout of the advertisement content to be adapted is transformed and the position of the key subject is adjusted for adaptation.

5. The cross-platform advertising content generation method according to claim 4, characterized in that, The key entities for determining the advertising content to be adapted include: Extract the visually salient regions and textual semantic focus from the advertising content to be adapted. The visually salient regions include one or more visual elements, and the textual semantic focus includes one or more keywords. Calculate the weight scores of one or more of the visual elements and one or more of the keywords, and determine the location area of ​​the visual element or keyword with the highest weight score as the key subject.

6. The cross-platform advertising content generation method according to claim 4, characterized in that, The step of performing layout transformation of the advertisement content to be adapted and position adaptation adjustment of the key subject based on the content feature information includes: A layout strategy template is generated based on the static and dynamic content features of the platform feature map. The layout strategy template includes layout parameter information, canvas partition definition, visual flow rules, suggested element types for each area, and recommended position range for the main body. Based on the layout strategy template, the layout of the advertisement content to be adapted is transformed, and the position of the key subject is adapted and adjusted.

7. The cross-platform advertising content generation method according to claim 6, characterized in that, The process of transforming the layout of the advertisement content to be adapted and adjusting the position of the key subject based on the layout strategy template includes: Calculate the matching degree between the key subject and other elements of the advertisement content to be adapted in different subject recommendation position intervals; The main layout position of the key subject is determined by the recommended position range of the subject with the highest matching degree, and the secondary layout positions of other elements are determined based on the canvas partition definition, visual flow rules and suggested element types of each area. Using the layout parameter information as a constraint, the elements of the advertisement content to be adapted are laid out to the corresponding main layout position and the secondary layout position.

8. A cross-platform advertising content generation device, characterized in that, include: The acquisition module is used to acquire the text of the advertisement currently awaiting processing. The feature determination module is used to determine the content feature information of the platform to be published corresponding to the advertising content copy. The content generation module is used to generate the advertising content to be adapted, which is aligned with the various modal features corresponding to the advertising content copy. The content adaptation module is used to convert the format of the advertisement content to be adapted based on the content feature information, and generate the advertisement content to be published.

9. A cross-platform advertising content generation device, the device comprising: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the cross-platform advertising content generation method as described in any one of claims 1-7.

10. A storage medium storing computer-executable instructions, which, when executed by a computer processor, are used to perform the cross-platform advertising content generation method as described in any one of claims 1-7.

Images