Intelligent content generation method and device oriented to private domain operation, computer equipment and storage medium

Abstract

The invention relates to the field of artificial intelligence, and discloses a private domain operation-oriented intelligent content generation method and device, computer equipment and a storage medium, and the method comprises the steps: processing community portrait information through a content generation component and a private domain component slot, and obtaining intelligent marketing content containing interaction elements; processing the intelligent marketing content through a multi-dimensional auditing model to obtain a multi-dimensional auditing result; if the multi-dimensional auditing result is automatic passing, a semantic label is added to the intelligent marketing content, and a target user is matched for the intelligent marketing content based on the semantic label; and distributing the intelligent marketing content to the target user. According to the method, the efficiency, the safety, the accuracy and the user stickiness of private domain operation can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, and in particular to an intelligent content generation method and device for private domain operation, a computer device and a storage medium. BACKGROUND

[0002] In the current private domain operation and content creation practice, enterprises generally face technical bottlenecks such as low efficiency, high cost, and weak collaboration. On the one hand, content production is highly dependent on manual work, making it difficult to meet large-scale and personalized communication needs. On the other hand, the review process is mostly done manually, which has problems such as low efficiency, inconsistent standards, and delayed response, which restricts the rapid release and compliance protection of content. At the same time, user behavior data is scattered in multiple platforms and systems, and due to the lack of unified data standards and integration mechanisms, it forms a "data island", hindering the accurate construction of user portraits and the intelligent optimization of marketing strategies. In addition, existing community operation tools have single functions and lack effective splitting incentives and interaction guidance mechanisms, making it difficult to improve user activity and brand stickiness.

[0003] Although AIGC (Artificial Intelligence Generated Content) technology has developed rapidly in recent years, its application is mostly limited to isolated content generation links and has not been deeply integrated with the overall private domain operation ecosystem. SUMMARY

[0004] The embodiments of the present application provide an intelligent content generation method and device for private domain operation, a computer device and a storage medium to improve the efficiency, security, accuracy and user stickiness of private domain operation.

[0005] An intelligent content generation method for private domain operation, comprising: processing community portrait information through a content generation component and a private domain component slot to obtain intelligent marketing content containing interactive elements; processing the intelligent marketing content through a multi-dimensional review model to obtain a multi-dimensional review result; if the multi-dimensional review result is automatically passed, adding a semantic tag to the intelligent marketing content and matching target users for the intelligent marketing content based on the semantic tag; distributing the intelligent marketing content to the target users.

[0006] Optionally, the processing of community portrait information through a content generation component and a private domain component slot to obtain intelligent marketing content containing interactive elements comprises: loading the community portrait information; constructing a marketing content skeleton based on the community portrait information; filling the marketing content skeleton to obtain preliminary marketing content; The context-aware component slot matching algorithm inserts a private domain component into the preliminary marketing content to obtain the intelligent marketing content.

[0007] Optionally, the context-aware component slot matching algorithm inserts a private domain component into the preliminary marketing content to obtain the intelligent marketing content, including: generating a plurality of candidate slot schemes through a preset private domain component library; calculating a context-aware score of each candidate slot scheme; the context-aware score includes a structure adaptation score and a component fusion fluency score; processing the context-aware score through a quality score model to obtain a content quality score of the candidate slot scheme; determining the candidate slot scheme with the highest content quality score as a target slot scheme; determining the intelligent marketing content according to the target slot scheme and the preliminary marketing content.

[0008] Optionally, the quality score model includes:

[0009] wherein, is the content quality score; is a generation quality score of the preliminary marketing content; is the structure adaptation score; is a structure adaptation weight; is the component fusion fluency score; is a component fusion fluency weight.

[0010] Optionally, the processing of the intelligent marketing content through the multi-dimensional review model to obtain a multi-dimensional review result includes: processing the intelligent marketing content through a compliance risk rule to obtain a compliance risk score; processing the intelligent marketing content through a brand association rule to obtain a brand association score; processing the intelligent marketing content through a content quality rule to obtain a content quality score; determining the multi-dimensional review result according to the compliance risk score, the brand association score, and the content quality score.

[0011] Optionally, the determination of the multi-dimensional review result according to the compliance risk score, the brand association score, and the content quality score includes: if the compliance risk score is greater than or equal to a first automatic pass threshold, the brand association score is greater than or equal to a second automatic pass threshold, and the content quality score is greater than or equal to a third automatic pass threshold, the multi-dimensional review result is an automatic pass; if the compliance risk score is less than a first automatic reject threshold, the brand association score is less than a second automatic pass reject threshold, or the content quality score is less than a third automatic reject threshold, the multi-dimensional review result is an automatic reject; if the compliance risk score is between the first automatic reject threshold and the first automatic pass threshold, the brand association score is between the second automatic reject threshold and the second automatic pass threshold, and the content quality score is between the third automatic reject threshold and the third automatic pass threshold, the multi-dimensional review result is a manual review.

[0012] Optionally, after the smart marketing content is distributed to the target user, the method further includes: collecting feedback data of the target user; updating the content generation component, the private domain component slot, and / or the multi-dimensional review model according to the feedback data.

[0013] An intelligent content generation device for private domain operation, comprising: an intelligent content generation module configured to process community portrait information through a content generation component and a private domain component slot to obtain smart marketing content containing interactive elements; a multi-dimensional review module configured to process the smart marketing content through a multi-dimensional review model to obtain a multi-dimensional review result; a matching customer module configured to, if the multi-dimensional review result is an automatic pass, add a semantic tag to the smart marketing content and match a target user for the smart marketing content based on the semantic tag; an intelligent content distribution module configured to distribute the smart marketing content to the target user.

[0014] A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the intelligent content generation method for private domain operation when executing the computer program.

[0015] A computer readable storage medium, storing a computer program, wherein the computer program is executed by a processor to implement the intelligent content generation method for private domain operation.

[0016] The intelligent content generation method, device, computer device and storage medium facing private domain operation have the advantages that the AIGC content generation is deeply fused with the private domain community portrait, intelligent and personalized efficient production of marketing content is realized, the creation efficiency is improved by more than 80%, the multi-dimensional audit model realizes automatic discrimination in compliance, brand tone and content quality, the audit efficiency is improved by 50%, safety and quality are considered, the intelligent matching mechanism based on semantic tags and user portraits significantly improves the content distribution accuracy, drives the synchronous growth of exposure rate, interaction rate and user conversion rate (increased by more than 20%), an automatic closed loop from generation, audit, matching to distribution is constructed, tool islands are broken, the deep integration of AIGC and the private domain ecology is realized, the artificial cost is effectively reduced, and a scalable digital private domain operation solution is formed. The intelligent content generation method, device, computer device and storage medium facing private domain operation can improve the efficiency, safety, accuracy and user stickiness of private domain operation. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 is an application environment schematic diagram of the intelligent content generation method facing private domain operation in an embodiment of the present application; Figure 2 is a flowchart of the intelligent content generation method facing private domain operation in an embodiment of the present application; Figure 3 is a schematic diagram of the intelligent content generation device facing private domain operation in an embodiment of the present application; Figure 4 is a schematic diagram of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0020] The intelligent content generation method facing private domain operation provided in the embodiments of the present application can be applied to Figure 1The application environment shown. Specifically, the smart content generation method for private domain operation is applied in a smart content generation system for private domain operation, which includes a client and a server as shown in Figure 1 communicate with each other through a network for generating and distributing smart marketing content. The client, also known as the user end, is a program that provides local services for customers corresponding to the server. The client can be installed on, but not limited to, various personal computers, notebook computers, smartphones, tablet computers and portable wearable devices. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0021] This embodiment can realize full-link automation from content creation to precise distribution. First, personalized smart marketing content that fits the community portrait is generated using AIGC technology, and the content quality and compliance are ensured through a multi-dimensional review model. After the review, the system automatically adds structured semantic tags to the smart marketing content, and matches the target user group that is most likely to be interested based on the semantic tags. Finally, the content is precisely delivered. This embodiment significantly improves marketing efficiency and personalization level.

[0022] In an embodiment, as shown in Figure 2 , a smart content generation method for private domain operation is provided, and the method is applied in Figure 1 a server as an example for illustration, including the following steps S10-S40.

[0023] S10, process the community portrait information through the content generation component and the private domain component slot to obtain smart marketing content containing interactive elements.

[0024] Understandably, the portrait information of the target community (such as "knowledge sharing type", "benefit promotion type", "hot spot interaction type") can be loaded first. The content generation component (for example, based on the general thousand questions, bean bag and other large models) automatically generates a preliminary marketing content framework, such as a product introduction script or a short video script, according to these community portrait information. Subsequently, the private domain component slot (such as "benefit entry", "interactive Q&A", "fission hook") will be automatically embedded into the most suitable position in the preliminary content through context perception algorithm, forming intelligent marketing content containing interactive elements. Here, the community portrait information refers to the user tag system constructed through data collection (such as community chat records, consumer behavior), including basic attributes, interest preferences, consumer ability and other dimensions. For example, the community portrait of Costco may include "value for money" and "family shopping" tags. The content generation component refers to the function module that automatically produces marketing materials such as scripts, pictures, and videos using AIGC models. The private domain component slot refers to the pre-set interactive element template (such as coupon pop-up, customer service entry, invitation prize button), which is intelligently inserted into the content through component slot matching algorithm to improve user conversion and interaction.

[0025] In an example, when generating content for a new sunscreen, AIGC first generates a script about the hazards of ultraviolet rays, and then the private domain slot automatically embeds the "scan code to get a trial pack" component at the end of the script, along with the interactive phrase "invite friends to sunscreen together, each get double points".

[0026] S20, processing the intelligent marketing content through a multi-dimensional audit model to obtain a multi-dimensional audit result.

[0027] Understandably, the intelligent marketing content can be audited through a multi-dimensional audit model to obtain a multi-dimensional audit result. The multi-dimensional audit model will analyze the text, image, audio and video of the intelligent marketing content, check whether there is a compliance risk (such as false propaganda, infringement), content quality and whether the brand tone is appropriate. After the audit is completed, a multi-dimensional audit result is output. The multi-dimensional audit result includes "automatic pass", "manual audit" or "automatic rejection". Among them, the multi-dimensional audit model can be an audit system that integrates computer vision, natural language processing and other multi-modal AI technologies. The multi-dimensional audit model can deeply understand the semantics of complex content such as videos and graphics, identify violations in pictures, voices and texts, and the audit efficiency is more than 50% higher than traditional methods.

[0028] S30, if the multi-dimensional audit result is automatic pass, adding semantic tags to the intelligent marketing content, and matching target users for the intelligent marketing content based on the semantic tags.

[0029] S40, distributing the intelligent marketing content to the target users.

[0030] Understandably, if intelligent marketing content is marked as "automatically approved," the system will automatically add structured semantic tags to it. These tags originate from AI analysis of the content's theme, keywords, and emotional tone (such as "#beauty," "#focus on formula efficacy," and "#promotion"). The system then matches these tags against a customer profile database to identify target users with high tag matching. Here, semantic tags are structured identifiers generated based on AI's understanding of the content's deeper meaning. Semantic tags more accurately reflect the core theme and emotional tendency of the content, thus achieving more precise "content finding people." Algorithms can calculate the similarity between content semantic tags and user interest tags to achieve accurate matching of target users.

[0031] Finally, personalized smart marketing content can be distributed to matched target users through pre-set private channels (such as WeChat groups and Moments) or public channels. The distribution strategy can be optimized based on user activity time, channel preferences, etc., to ensure the best reach.

[0032] This embodiment deeply integrates AIGC content generation with private domain community profiling, achieving intelligent, personalized, and efficient production of marketing content, improving creation efficiency by over 80%. A multi-dimensional review model automatically assesses compliance, brand tone, and content quality, improving review efficiency by 50% while balancing security and quality. An intelligent matching mechanism based on semantic tags and user profiles significantly improves content distribution accuracy, driving simultaneous growth in exposure, interaction, and user conversion rates (by over 20%). The overall system constructs an automated closed loop from generation, review, matching to distribution, breaking down tool silos, achieving deep integration of AIGC and the private domain ecosystem, effectively reducing labor costs, and forming a scalable and replicable digital private domain operation solution.

[0033] Optionally, step S10, namely, processing community profile information through the content generation component and private domain component slots to obtain intelligent marketing content containing interactive elements, includes: S101. Load the community profile information; S102. Construct a marketing content framework based on the community profile information; S103. Fill in the marketing content skeleton to obtain preliminary marketing content; S104. A context-aware component slot matching algorithm is used to insert private domain components into the initial marketing content to obtain the intelligent marketing content.

[0034] Understandably, pre-built profile data of the target community (e.g., "post-90s mothers in first-tier cities") can be retrieved from a Customer Data Platform (CDP) or SCRM system (Social Customer Relationship Management System), i.e., community profile information. This community profile information can be a structured set of tags. Community profile information includes, but is not limited to, static attributes (such as region and age), dynamic behaviors (such as browsing preferences and purchase frequency), and interest tags (such as focusing on formula efficacy and liking classic domestic products).

[0035] The content generation component selects the most suitable marketing content skeleton from the template library based on user profile tags. For example, if the profile indicates that the community is "rational decision-making type," a skeleton of "pain point questioning -> data comparison -> authoritative certification -> limited-time commitment" might be selected; if it is "impulsive consumption type," a skeleton of "celebrity endorsement -> atmosphere creation -> scarcity emphasis -> immediate purchase" might be used. The marketing content skeleton does not involve specific copy, but defines the narrative logic, paragraph structure, and core emotional turning points of the content, ensuring that the generated content conforms to the cognitive and decision-making path of the target audience.

[0036] AIGC engines (such as those based on GPT or Wenxin Yiyan models) generate specific copy based on the requirements of each node in the skeleton. For example, for the "pain point questioning" node, combined with the profile of "young white-collar workers staying up late," the statement "Do you always feel inefficient late at night, yet unable to put down your phone?" can be generated. During generation, the AIGC engine's output must be constrained to conform to the contextual logic, for example, by providing prompts to specify length, keyword inclusion, and prohibiting repetition.

[0037] Through a context-aware component slot matching algorithm, private domain components (such as benefit entry points and interactive Q&A) are automatically inserted at appropriate locations (such as paragraph ends or transition points) to form intelligent marketing content. It is essential to ensure that components blend naturally with the context; for example, inserting a "Customized Recipe Collection" link after explaining "fitness and diet." The component slot matching algorithm can understand the deep semantics and intent of the content, determining whether the content is currently in the stage of "introducing a problem," "explaining a viewpoint," or "calling for action," thereby deciding where and what kind of components (such as questionnaires, coupons, or customer service entry points) are most natural and effective. Private domain components can be reusable interactive touchpoint modules, serving as hooks to guide users into the brand's private traffic pool, such as "benefit collection entry points," "user survey questionnaires," "viral sharing buttons," and "dedicated customer service consultations."

[0038] This embodiment loads community profile information and constructs a content skeleton to achieve structured and scenario-based generation of marketing content. Combined with a context-aware component slot matching algorithm, it intelligently embeds private domain components such as benefit entry points, interactive prompts, and referral hooks, making the content naturally highly interactive and conversion-oriented. The entire process dynamically couples user group characteristics with content elements, significantly improving the relevance and appeal of intelligent marketing content, laying a high-quality content foundation for subsequent precise distribution and closed-loop operation.

[0039] Optionally, step S104, namely, the context-aware component slot matching algorithm inserting a private domain component into the initial marketing content to obtain the intelligent marketing content, includes: S1041. Generate multiple candidate slot schemes by pre-setting a private domain component library; S1042. Calculate the context-aware score for each of the candidate slot schemes; the context-aware score includes the structure adaptability score and the component fusion smoothness score. S1043. Process the context-aware score using a quality scoring model to obtain the content quality score of the candidate slot scheme; S1044. The candidate slot scheme with the highest content quality score is determined as the target slot scheme. S1045. Determine the intelligent marketing content based on the target slot scheme and the preliminary marketing content.

[0040] Understandably, a rich private domain component library can be pre-set, containing various types of interactive element templates, such as "limited-time coupon pop-up," "follow official account card," "add WeChat customer service button," "user survey questionnaire," and "invite friends for benefits." Upon receiving initial marketing content about "summer sunscreen," the system will initially filter out several relevant components from the component library (such as "get sunscreen samples" and "skin test appointment") based on the content theme and pre-set rules. For each component, it will generate multiple possible combinations of insertion positions and display formats, forming several candidate slot schemes. For example, Scheme 1 might insert a "Get a trial pack now" button at the end of the article; Scheme 2 might embed a "free skin test" pop-up window when "UV damage" is mentioned in the article.

[0041] A context-aware score can be calculated for each candidate slot design. A context-aware score is an algorithm-based quantitative evaluation metric used to measure the degree to which a slot design matches the content context (including logical structure and language style). In some examples, the context-aware score includes a structure fit score and a component integration fluency score.

[0042] The structural fit score is used to evaluate how well the slot scheme fits the logical structure of the content. For example, inserting a "purchase link" after the "solution" paragraph will score highly, while inserting a "customer service phone number" in the middle of the "pain point introduction" paragraph will score poorly because it disrupts the reading flow. The structural fit scoring algorithm analyzes the paragraph divisions, logical transitions, and semantic emphasis of the content.

[0043] Component fusion fluency scoring assesses the degree to which a component integrates semantically and in tone with the surrounding text. For example, inserting a playful, internet-slang-style "welfare entry" into a serious science article would result in a low fluency score. The component fusion fluency scoring algorithm uses natural language processing techniques to analyze the sentiment and word choice style of the context.

[0044] The quality scoring model (which can be a trained machine learning model) can receive the two scores from S1042 and combine them (or not) with other business metrics (such as the component's historical click-through rate, user lifetime value, etc.) to output a final content quality score. The system then selects the solution with the highest score as the target slot solution. The quality scoring model is a comprehensive evaluation and decision-making model; its evaluation dimensions can be customized according to business objectives and adjusted based on user feedback.

[0045] Finally, based on the selected target slot scheme, the corresponding private domain components are precisely embedded into the designated positions of the initial marketing content. Necessary formatting adjustments and language fine-tuning are then made to ensure a seamless visual and semantic integration, ultimately outputting ready-to-use intelligent marketing content. For example, a QR code for "Scan the code to receive sun protection tips" and its accompanying text are seamlessly inserted as annotation cards at the end of the article. This embodiment achieves intelligent, precise, and natural integration of private domain components and marketing content through algorithm-driven automated decision-making. While significantly improving the efficiency of intelligent marketing content production, it also significantly improves the click-through rate and conversion rate of interactive components by optimizing user experience, thereby truly unleashing the enormous potential of AIGC in the marketing field.

[0046] Optionally, the quality scoring model includes:

[0047] in, Rate the quality of the content; The quality score for the generated preliminary marketing content; Score the structural fit. Weights for structural fit; The smoothness of component integration is scored; Assign weights to component integration smoothness.

[0048] Understandably, content quality rating It is not determined by a single dimension, but by the quality score generated. Structural fit score Component integration smoothness score A weighted comprehensive score determined by three core elements.

[0049] Generate quality score The evaluation assesses the quality of the initial marketing content generated by AIGC, which can be scored based on dimensions such as relevance (whether it stays on topic), fluency (whether the language is natural and fluent), and appeal (whether the content is engaging). For example, a piece of copy introducing high-end skincare products... They will examine whether the use of professional terminology is accurate and whether the descriptions are beautiful and moving.

[0050] Structural fit score Assess whether the content conforms to the pre-set marketing logic framework. For example, for "knowledge sharing" content, its structure might be "pain point introduction - principle explanation - product solution - call to action". The assessment will evaluate whether the content fully covers these aspects and whether the logical progression is clear and reasonable.

[0051] Component integration smoothness score Assess the naturalness of how well private domain components (such as coupon pop-ups and customer service guidance buttons) blend with the content context. Examine whether the placement of the component is appropriate and whether the guiding language flows smoothly with the preceding and following text, rather than being abruptly inserted. For example, after explaining the importance of sun protection in summer, a component like "Get a sunscreen sample now" can be seamlessly integrated. It will be very high.

[0052] Structural fit weight Component integration smoothness weight The size reflects the different emphases of different marketing strategies. If the content strategy is brand-oriented, it may be adjusted upwards. They place greater emphasis on logical rigor; if the strategy is sales conversion-oriented, the adjustment may be higher. It places greater emphasis on the embedding effect of interactive components. In one example, , .

[0053] This embodiment introduces a configurable weighted fusion formula to dynamically combine the basic content quality with the context-aware structural fit score and the component fusion smoothness score, thereby achieving quantitative optimization of the slot scheme. This not only preserves the original content generation quality but also significantly enhances the rationality of private domain component embedding and the overall content synergy, thereby improving the professionalism and user acceptance of intelligent marketing content.

[0054] Optionally, step S20, namely, processing the intelligent marketing content through a multi-dimensional review model to obtain multi-dimensional review results, includes: S201. Process the intelligent marketing content through compliance risk rules to obtain a compliance risk score; S202. Process the intelligent marketing content through brand association rules to obtain a brand association score; S203. Process the intelligent marketing content through content quality rules to obtain a content quality score; S204. Determine the multi-dimensional audit result based on the compliance risk score, the brand association score, and the content quality score.

[0055] Understandably, a pre-defined compliance risk rule base can be used to scan intelligent marketing content and generate a compliance risk score. This rule base typically integrates platform policies (such as WeChat's operating guidelines), prohibitive clauses of advertising laws, and industry-specific regulations. For example, the system might detect the presence of absolute terms like "best deal" or "number one brand," or unauthorized use of celebrity images. Compliance risk rules can be a set of machine-readable regulatory rules and platform specifications based on text, images, and audio (video) formats. These rules automatically identify potential violations in intelligent marketing content using technologies such as keyword matching, semantic analysis, and multimodal recognition. The compliance risk score is a quantifiable safety indicator; a lower score indicates a higher risk of content violations.

[0056] Brand association rules can be a logical set that transforms abstract concepts such as brand positioning and brand image into quantifiable evaluation metrics. Brand association rules ensure that intelligent marketing content aligns with the brand's core values. They evaluate aspects such as brand tone, visual guidelines (e.g., logo usage, primary color scheme), and core audience preferences to generate a brand association score. This score measures the consistency between marketing content and the brand's long-term image and strategy. For example, a high-end, luxury brand will have a low brand association score if its marketing content uses overly internet-savvy or colloquial expressions, even if the content itself is interesting. The system compares elements in the intelligent marketing content (such as visual style, keywords, and emotional tone) with the guidelines defined in the brand manual to calculate their fit.

[0057] Content quality rules can be a set of criteria for judging the quality of content based on communication studies, user experience design, and copywriting skills. Content quality rules assess the readability, logical structure, visual appeal, informational value, and effectiveness of interactive guidance in intelligent marketing content, generating a content quality score. For example, the system might analyze whether the copywriting of intelligent marketing content is fluent and free of typos, whether the video is clear, whether the information points are clearly stated, and whether there are explicit instructions to guide users to comment or click. The content quality score can be a quantifiable indicator of whether the intelligent marketing content itself is "attractive," "useful," and "encouraging to interact." High-quality intelligent marketing content is the foundation for generating positive user interaction and conversions.

[0058] Finally, a multi-dimensional review result is generated based on the three scoring methods. The multi-dimensional review result includes three types: automatic approval, transfer to manual review, and automatic rejection.

[0059] This embodiment achieves multi-dimensional and structured automated review by evaluating the performance of intelligent marketing content across three key dimensions: compliance risk, brand relevance, and content quality. Independent scoring mechanisms for each dimension ensure clear, configurable, and traceable review logic, balancing content security, brand consistency, and dissemination effectiveness. The review results are comprehensively determined based on multi-dimensional scoring, effectively replacing traditional manual review and improving efficiency by over 50%. Simultaneously, it lays the technical foundation for flexibly adapting review strategies to different content types (such as promotions, science popularization, and interactive content).

[0060] Optionally, step S204, namely determining the multi-dimensional review result based on the compliance risk score, the brand association score, and the content quality score, includes: S2041. If the compliance risk score is greater than or equal to the first automatic pass threshold, the brand association score is greater than or equal to the second automatic pass threshold, and the content quality score is greater than or equal to the third automatic pass threshold, then the multi-dimensional review result is automatically passed. S2042. If the compliance risk score is less than the first automatic rejection threshold, the brand association score is less than the second automatic pass / rejection threshold, or the content quality score is less than the third automatic rejection threshold, then the multi-dimensional review result is automatic rejection. S2043. If the compliance risk score is between the first automatic rejection threshold and the first automatic approval threshold, the brand association score is between the second automatic rejection threshold and the second automatic approval threshold, and the content quality score is between the third automatic rejection threshold and the third automatic approval threshold, then the multi-dimensional review result is to be transferred to manual review.

[0061] Understandably, each rating type has a corresponding automatic approval threshold and automatic rejection threshold. The multi-dimensional review result is automatically approved only when all three ratings exceed the corresponding automatic approval threshold. This means that the intelligent marketing content has reached high standards in terms of risk, brand fit, and content appeal, and can proceed directly to the next stage (such as semantic tag matching and distribution) without human intervention.

[0062] If any score falls below the corresponding automatic rejection threshold, the multi-dimensional review result will be an automatic rejection. This mechanism ensures that content with obvious problems is efficiently blocked, preventing it from entering the market or wasting human review resources.

[0063] When none of the three ratings are good enough to be automatically approved, nor bad enough to be immediately rejected, but rather fall between their respective "automatic rejection threshold" and "automatic approval threshold" (i.e., greater than or equal to the automatic rejection threshold, less than the automatic approval threshold), the system will determine the result as "transfer to manual review." This usually means that the intelligent marketing content is somewhat controversial or requires more complex judgment.

[0064] It's important to note that the automatic approval and automatic rejection thresholds generally differ for different types of marketing content. In one application example, the first automatic approval threshold for compliance risk scoring could be 0.9, and the first automatic rejection threshold could be 0.6; the second automatic approval threshold for brand association scoring could be 0.8, and the second automatic rejection threshold could be 0.4; and the third automatic approval threshold for content quality scoring could be 0.8, and the third automatic rejection threshold could be 0.5.

[0065] For smart marketing content that fails the review, the system not only provides scoring results across various dimensions, but also accurately identifies specific problems and generates actionable modification suggestions: if the brand relevance score is below the threshold, the system will identify paragraphs lacking brand keywords and suggest inserting recommended keywords in the corresponding positions; if the content quality score is substandard, the system will point out specific sentences with incoherent logic and recommend adding transitional sentences to enhance the fluency of the writing, ultimately outputting a structured optimization report containing a description of the problem, location information, and modification suggestions.

[0066] This embodiment achieves refined and tiered review decisions for intelligent marketing content by setting multi-dimensional dual threshold ranges (automatic approval / automatic rejection / manual review). While ensuring that high-risk content is blocked in a timely manner, it avoids over-reliance on manual review, improving review efficiency and the rationality of resource allocation. The threshold mechanism supports flexible configuration according to content type, taking into account compliance bottom line, brand tone and operational agility, forming a safe, controllable, efficient and intelligent content review closed loop.

[0067] Optionally, after step S40, i.e. after distributing the intelligent marketing content to the target user, the method further includes: S50. Collect feedback data from the target user; S60. Update the content generation component, the private domain component slot, and / or the multi-dimensional audit model based on the feedback data.

[0068] Understandably, after intelligent marketing content is distributed to target users, the system automatically collects user interaction feedback through various channels. For example, after pushing a marketing article about "summer skincare products," the system records whether users clicked to read, the reading time, whether they liked / disliked the article, whether they copied a discount code, and whether they clicked on private domain components in the article (such as the "Contact Customer Service" button). In some examples, more direct subjective evaluations can also be obtained by proactively pushing lightweight NPS (Net Promoter Score) questionnaires or satisfaction surveys.

[0069] After being cleaned, categorized, and analyzed, the collected feedback data is used to optimize content generation components, private domain component slots, and / or multi-dimensional review models. For example, for content generation components, if data analysis reveals that article titles starting with "question sentences" have a significantly higher open rate than those starting with "declarative sentences," this successful pattern will be refined into rules or used to fine-tune AIGC model prompts, ensuring that subsequent generated content uses more high-conversion-rate expressions. For private domain component slots, if it's found that "scan to receive materials" components have the highest acquisition rate when inserted in the middle of the article, but a very low acquisition rate when inserted at the end, the system will adjust the optimal insertion position of this type of component using a context-aware component slot matching algorithm to improve interaction and conversion. For multi-dimensional review models, if content with low scores in the "brand association" dimension but excellent user feedback (such as a successful "down-to-earth" marketing campaign) is found, the system can adjust the rules or weights of the brand association scoring, allowing the review model to better encompass and identify effective innovative forms, avoiding the mistaken rejection of high-quality content.

[0070] This embodiment collects feedback data from target users on the distributed content to dynamically optimize the content generation components, private domain component slots, and multi-dimensional review models, forming a closed-loop learning mechanism of "distribution-feedback-iteration" to continuously improve the relevance, interactive effects, and review accuracy of the content, thereby promoting the intelligent evolution of the private domain operation system.

[0071] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0072] In one embodiment, a smart content generation device for private domain operations is provided, which corresponds one-to-one with the smart content generation method for private domain operations described in the above embodiments. For example... Figure 3 As shown, this intelligent content generation device for private domain operations includes: The intelligent content generation module 10 is used to process community profile information through content generation components and private domain component slots to obtain intelligent marketing content containing interactive elements. The multi-dimensional review module 20 is used to process the intelligent marketing content through a multi-dimensional review model to obtain multi-dimensional review results; The customer matching module 30 is used to add semantic tags to the intelligent marketing content if the multi-dimensional review result is automatic approval, and to match target users for the intelligent marketing content based on the semantic tags; The intelligent content distribution module 40 is used to distribute the intelligent marketing content to the target user.

[0073] Optionally, the intelligent content generation module 10 includes: A profile information loading unit is used to load the community profile information; A content skeleton unit is constructed to build a marketing content skeleton based on the community profile information. A preliminary content unit is obtained, which is used to fill in the marketing content skeleton to obtain preliminary marketing content; A smart marketing content unit is generated, which is used to insert a private domain component into the initial marketing content based on a context-aware component slot matching algorithm to obtain the smart marketing content.

[0074] Optionally, the intelligent marketing content generation unit includes: A candidate slot scheme generation unit is used to generate multiple candidate slot schemes through a preset private domain component library; A context-aware scoring unit is used to calculate the context-aware score for each of the candidate slot schemes; the context-aware score includes a structure adaptability score and a component fusion smoothness score. A content quality scoring unit is obtained, which is used to process the context-aware scoring through a quality scoring model to obtain the content quality score of the candidate slot scheme; The target slot scheme determination unit is used to determine the candidate slot scheme with the highest content quality score as the target slot scheme. A marketing content unit is defined to determine the intelligent marketing content based on the target slot scheme and the preliminary marketing content.

[0075] Optionally, the quality scoring model includes:

[0076] in, Rate the quality of the content; The quality score for the generated preliminary marketing content; Score the structural fit. Weights for structural fit; The smoothness of component integration is scored; Assign weights to component integration smoothness.

[0077] Optionally, the multi-dimensional audit module 20 includes: The compliance scoring unit is used to process the smart marketing content according to compliance risk rules and obtain a compliance risk score. The brand scoring unit is used to process the intelligent marketing content through brand association rules to obtain a brand association score. The content scoring unit is used to process the intelligent marketing content according to content quality rules and obtain a content quality score. A multi-dimensional audit result unit is generated to determine the multi-dimensional audit result based on the compliance risk score, the brand association score, and the content quality score.

[0078] Optionally, the unit for generating multi-dimensional audit results includes: An automatic approval unit is configured to automatically approve the multi-dimensional review result if the compliance risk score is greater than or equal to a first automatic approval threshold, the brand association score is greater than or equal to a second automatic approval threshold, and the content quality score is greater than or equal to a third automatic approval threshold. An automatic rejection unit is used to automatically reject the multi-dimensional review result if the compliance risk score is less than the first automatic rejection threshold, the brand association score is less than the second automatic pass rejection threshold, or the content quality score is less than the third automatic rejection threshold. The manual review unit is configured to transfer the multidimensional review result to manual review if the compliance risk score is between the first automatic rejection threshold and the first automatic approval threshold, the brand association score is between the second automatic rejection threshold and the second automatic approval threshold, and the content quality score is between the third automatic rejection threshold and the third automatic approval threshold.

[0079] Optionally, the intelligent content generation device for private domain operations further includes: A user feedback data collection module is used to collect feedback data from the target user; The component / model update module is used to update the content generation component, the private domain component slot, and / or the multi-dimensional audit model based on the feedback data.

[0080] Specific limitations regarding the intelligent content generation device for private domain operations can be found in the limitations of the intelligent content generation method for private domain operations described above, and will not be repeated here. Each module in the aforementioned intelligent content generation device for private domain operations can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0081] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores data related to a smart content generation method for private domain operations. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a smart content generation method for private domain operations.

[0082] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent content generation method for private domain operation described in the above embodiment; to avoid repetition, it will not be described again here. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the intelligent content generation device for private domain operation embodiment; to avoid repetition, it will not be described again here.

[0083] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When executed by a processor, the computer program implements the intelligent content generation method for private domain operation described in the above embodiment. To avoid repetition, this will not be described again here. Alternatively, when executed by a processor, the computer program implements the functions of each module / unit in the intelligent content generation device for private domain operation described in this embodiment. To avoid repetition, this will not be described again here.

[0084] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0085] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0086] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A smart content generation method for private domain operation, characterized in that, The method comprises the following steps: processing community portrait information through a content generation component and a private domain component slot to obtain intelligent marketing content containing interactive elements; processing the intelligent marketing content through a multi-dimensional audit model to obtain a multi-dimensional audit result; if the multi-dimensional audit result is automatic passing, adding a semantic label to the intelligent marketing content and matching target users for the intelligent marketing content based on the semantic label; distributing the intelligent marketing content to the target users.

2. The intelligent content generation method for private domain operation according to claim 1, characterized in that, The processing of the community portrait information through the content generation component and the private domain component slot to obtain the intelligent marketing content containing the interactive elements comprises the following steps: loading the community portrait information; constructing a marketing content skeleton based on the community portrait information; filling the marketing content skeleton to obtain preliminary marketing content; inserting a private domain component into the preliminary marketing content based on a context-aware component slot matching algorithm to obtain the intelligent marketing content.

3. The intelligent content generation method for private domain operation according to claim 2, characterized in that, The inserting of the private domain component into the preliminary marketing content based on the context-aware component slot matching algorithm to obtain the intelligent marketing content comprises the following steps: generating a plurality of candidate slot schemes through a preset private domain component library; calculating a context-aware score of each candidate slot scheme; the context-aware score comprises a structure adaptation score and a component fusion fluency score; processing the context-aware score through a quality score model to obtain a content quality score of the candidate slot scheme; determining a candidate slot scheme with the highest content quality score as a target slot scheme; determining the intelligent marketing content according to the target slot scheme and the preliminary marketing content.

4. The intelligent content generation method for private domain operation according to claim 3, characterized in that, The quality score model comprises: wherein, is the content quality score; a generated quality score for the preliminary marketing content; adapt the structure to the score; is the structure fitness weight; fusion fluency score for the assembly; A component fusion fluency weight.

5. The intelligent content generation method for private domain operation according to claim 1, characterized in that, The processing of the intelligent marketing content through the multi-dimensional audit model to obtain the multi-dimensional audit result comprises the following steps: processing the intelligent marketing content through a compliance risk rule to obtain a compliance risk score; processing the intelligent marketing content through a brand association rule to obtain a brand association score; processing the intelligent marketing content through a content quality rule to obtain a content quality score; determining the multi-dimensional audit result according to the compliance risk score, the brand association score and the content quality score.

6. The intelligent content generation method for private domain operation according to claim 5, characterized in that, The determination of the multi-dimensional audit result according to the compliance risk score, the brand association score and the content quality score comprises the following steps: if the compliance risk score is greater than or equal to a first automatic passing threshold, the brand association score is greater than or equal to a second automatic passing threshold, and the content quality score is greater than or equal to a third automatic passing threshold, the multi-dimensional audit result is automatic passing; if the compliance risk score is less than a first automatic rejection threshold, the brand association score is less than a second automatic passing rejection threshold, or the content quality score is less than a third automatic rejection threshold, the multi-dimensional audit result is automatic rejection; if the compliance risk score is between the first automatic rejection threshold and the first automatic passing threshold, the brand association score is between the second automatic rejection threshold and the second automatic passing threshold, and the content quality score is between the third automatic rejection threshold and the third automatic passing threshold, the multi-dimensional audit result is manual audit.

7. The intelligent content generation method for private domain operation according to claim 1, characterized in that, The distributing the intelligent marketing content to the target user further comprises: collecting feedback data of the target user; updating the content generation component, the private domain component slot, and / or the multi-dimensional review model according to the feedback data.

8. An intelligent content generation device for private domain operation, characterized in that, The method comprises: an intelligent content generation module configured to process community portrait information through a content generation component and a private domain component slot to obtain intelligent marketing content containing interactive elements; a multi-dimensional review module configured to process the intelligent marketing content through a multi-dimensional review model to obtain a multi-dimensional review result; a matching customer module configured to, if the multi-dimensional review result is automatic pass, add a semantic tag to the intelligent marketing content and match a target user for the intelligent marketing content based on the semantic tag; an intelligent content distribution module configured to distribute the intelligent marketing content to the target user.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the private domain operation-oriented intelligent content generation method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1-9. The computer program is executed by the processor to implement the private domain operation-oriented intelligent content generation method according to any one of claims 1 to 7.

Images