Marketing investment effect intelligent prediction and optimization method and system

Abstract

This invention relates to the field of human-computer interaction technology, and more particularly to a method and system for intelligent prediction and optimization of marketing campaign effectiveness. The method includes collecting historical campaign data, effectiveness data, and relevant information from social media platforms, and constructing a correspondence between historical campaign data and effectiveness data as a historical knowledge base, and using the relevant information as an information knowledge base. Semantic features and scenario information are extracted from the historical knowledge base and the information knowledge base, and the semantic features are categorized, tagged, and a content database is generated. By matching user input plans with the content database and proactively generating a list of business constraint information for questioning, the method can identify ambiguities and missing links in the plan, guiding users to improve it through data-driven approaches. This allows the marketing campaign plan to reach its optimal state through a limited number of human-computer interactions before deployment, thereby improving the accuracy of marketing campaign effectiveness prediction and the scientific nature of the plan.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, and in particular to a method and system for intelligent prediction and optimization of marketing campaign effectiveness. Background Technology

[0002] With the deepening development of digital marketing, enterprises have increasingly rich data resources at their disposal, including historical campaign records, user behavior data, and massive amounts of information on social media platforms. All of this data is recorded, and how to effectively utilize this data to develop efficient and accurate marketing campaign plans has become the key for enterprises to stand out in fierce market competition.

[0003] Currently, most marketing campaign development still heavily relies on the personal experience and intuition of marketing planners. Planners typically need to manually export data from multiple systems, perform tedious data processing and analysis using spreadsheets and other tools, and then combine this with their understanding of market trends to conceive a plan. This process is not only time-consuming and labor-intensive, but also inefficient. Furthermore, due to the limitations and subjectivity of personal experience, it is difficult to comprehensively and objectively assess the complex impact of multi-dimensional factors (such as content creation, target audience, timing of placement, and budget allocation) on the final results. This leads to low predictive accuracy and unclear optimization directions. The traditional campaign development process is static and one-off; once the plan is finalized, it is immediately implemented. Potential risks or areas for improvement are only revealed after launch, impacting the final marketing effectiveness. Summary of the Invention

[0004] The technical problem this invention aims to solve is that, due to the limitations and subjectivity of personal experience, marketing planners find it difficult to comprehensively and objectively assess the complex impact of multi-dimensional factors (such as content creativity, target audience, timing of placement, budget allocation, etc.) on the final effect, resulting in low accuracy in predicting the plan and a vague direction for optimization. The traditional plan formulation process is static and one-off; once the plan is determined, it will be directly launched, and potential risks or points for optimization in the plan will only be known after the launch, affecting the final marketing effect.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: Firstly, the intelligent prediction and optimization method for marketing campaign effectiveness of the present invention includes:

[0006] Step S1: Collect historical campaign data, performance data, and relevant information from social media platforms, and construct the correspondence between historical campaign data and performance data as a historical knowledge base, and use the relevant information as an information knowledge base;

[0007] Step S2: Extract semantic features and scene information from the historical knowledge base and the information knowledge base, classify the semantic features, add tag information, and generate a content database;

[0008] Step S3: The user inputs a marketing campaign plan, the semantic features and scenario information in the marketing campaign plan are extracted, and the semantic features and scenario information of the marketing campaign plan are matched with the content database to obtain tag information. Based on the tag information and scenario information, a list of business constraint information is generated and questions are asked to the user.

[0009] Step S4: Based on the information content of the user's answer to the business constraint information list, update the marketing campaign plan, and repeat steps S2 and S3 until the completeness of the requirements reaches the preset threshold or the user actively stops asking questions. Then, output the last updated marketing campaign plan as the final plan.

[0010] Step S5: Based on the correspondence between historical campaign data and performance data, predict the performance of the latest updated marketing campaign plan, obtain the predicted performance data, and adjust the marketing campaign plan when the actual performance data deviates from the prediction by more than a threshold.

[0011] Preferably, the historical deployment data includes supply-side data, demand-side data, population data, and time data;

[0012] The supply-side data includes content data, content format, distribution volume, budget data, and KPI data;

[0013] The demand-side data includes search terms, upstream and downstream vocabulary, similar words, search habits, and search paths;

[0014] The crowd data includes crowd segments, crowd segment characteristics, crowd segment names, and the number of crowd segments;

[0015] The time data includes marketing milestones, dissemination cycles, campaign timing, and recent trending topics;

[0016] The performance data includes readership, pageviews, likes, comments, search conversion rate, and revenue data.

[0017] Preferably, based on the search terms, upstream and downstream vocabulary, and similar words in the historical delivery data as keywords, relevant information from social media platforms is collected, and the searched relevant information is used as an information knowledge base.

[0018] Preferably, the relevant information is used as an information knowledge base, specifically including:

[0019] Extract supply-side data, demand-side data, population data, and time data from the information knowledge base, and classify the extracted supply-side data, demand-side data, population data, and time data to obtain a classified information set. Further classify the information set according to the respective categories of supply-side data, demand-side data, population data, and time data to obtain a meta-data set. Extract semantic features from the meta-data set to obtain a semantic feature set, and classify the semantic features in the semantic feature set and add label information.

[0020] Scene information is constructed by combining the audience package name from the audience data, the nouns from the demand-side data, and the environment from the content data. The scene information includes the audience package name plus the nouns plus the environment.

[0021] Preferably, step S3, in which the user inputs a marketing campaign plan and extracts the semantic features and scenario information from the marketing campaign plan, specifically includes:

[0022] Step S31: Obtain the user's marketing campaign plan and extract the supply-side data, demand-side data, audience data, and time data of the marketing campaign plan;

[0023] Step S32: Classify the supply-side data, demand-side data, audience data, and time data extracted from the marketing campaign plan to obtain the information set after classification of the marketing campaign plan;

[0024] Step S33: After classifying the information set of marketing campaign plans, further classify it according to the categories of supply-side data, demand-side data, audience data, and time data to obtain the meta-data set of marketing campaign plans;

[0025] Step S34: Extract semantic features from the metadata of the marketing campaign plan to obtain the semantic feature set of the marketing campaign plan, establish the correspondence between the semantic feature set of the marketing campaign plan and the marketing campaign plan, and label the scenario information of the marketing campaign plan.

[0026] Preferably, the semantic features and scenario information of the marketing campaign are matched with a content database to obtain tag information. Based on the tag information and scenario information, a list of business constraint information is generated, and questions are asked to the user, specifically including:

[0027] Step S35: Perform a first match between the semantic features in the marketing campaign plan and the content database to obtain a first match result. Then, perform a second match between the first match result and the scene information corresponding to the marketing campaign plan to obtain a second match result. At the same time, obtain the tag information and scene information corresponding to the second match result.

[0028] Step S36: Classify the data according to the tag information corresponding to the secondary matching results. The classification content includes historical knowledge base and information knowledge base. Select supply-side data, demand-side data, population data and time data from the historical knowledge base obtained from the secondary matching results. Then generate an initial business constraint information list based on the supply-side data, demand-side data, population data and time data, as well as the corresponding scenario information.

[0029] Preferably, based on the user account that inputs the marketing campaign plan, the user's identity information is obtained. Based on the user's identity information and preset identity operation permissions, the content of the initial business constraint information list is filtered, and the filtered business constraint information list is displayed or sent to the user, and questions are asked to the user.

[0030] Preferably, step S4 specifically includes:

[0031] Step S41: The user answers the questions based on the business constraint information list and enters the information into the system;

[0032] Step S42: Extract the supply-side data, demand-side data, audience data, and time data of the answer content, and update the supply-side data, demand-side data, audience data, and time data of the answer content in the corresponding places in the marketing campaign plan;

[0033] Step S42: Input the updated marketing campaign plan again, and repeat steps S2 and S3 until the completeness of the requirements reaches the preset threshold or the user actively stops asking questions. Then, output the last updated marketing campaign plan as the final plan.

[0034] Preferably, step S5 specifically includes:

[0035] Step S51: Record the correspondence between supply-side data, demand-side data, population data, time data, and effect data respectively. Analyze the correspondence between supply-side data, demand-side data, population data, time data, and effect data according to the preset volatility. When the volatility of supply-side data, demand-side data, population data, or time data is less than the preset volatility, it is considered as no change. When the volatility of supply-side data, demand-side data, population data, or time data is greater than the preset volatility, record the correspondence between supply-side data, demand-side data, population data, or time data and effect data.

[0036] Step S52: Based on the correspondence between the recorded supply-side data, demand-side data, audience data or time data and effect data, calculate the effect data corresponding to the supply-side data, demand-side data, audience data or time data in the last updated marketing campaign plan, and use the calculation results as the predicted effect data.

[0037] Step S53: Record the actual performance data of the marketing campaign. When the deviation between the actual performance data and the prediction exceeds the threshold, adjust the supply-side data, demand-side data, audience data or time data and performance data in the marketing campaign.

[0038] Secondly, the intelligent prediction and optimization system for marketing campaign effectiveness includes:

[0039] The data acquisition module is used to collect historical campaign data, performance data, and social media platform information, and to construct the correspondence between the historical campaign data and performance data to form a historical knowledge base, and to use the social media platform information as an information knowledge base;

[0040] The feature analysis module is used to extract semantic features and scene information from the historical knowledge base and the information knowledge base, and to classify and add tags to the semantic features to generate a content database;

[0041] The interactive matching module is used to extract the semantic features and scenario information of the marketing campaign input by the user, and match the semantic features and scenario information of the marketing campaign with the content database to generate a list of business constraint information and ask questions to the user.

[0042] The optimization iteration module is used to update the marketing campaign plan based on the user's response to the business constraint information list, and trigger the content database generation module and the interaction matching module to execute in a loop until the preset termination condition is met, and then output the final plan.

[0043] The performance monitoring module is used to compare and analyze the actual marketing campaign results with the predicted results. When the deviation between the actual results and the predictions exceeds a threshold, it automatically triggers model updates and strategy adjustments.

[0044] The beneficial effects of this invention are:

[0045] By constructing historical and information knowledge bases and utilizing semantic feature extraction and tagging technologies, marketing data is transformed into structured knowledge that can be understood and utilized by machines. This replaces traditional manual data integration and analysis methods, significantly improving the efficiency of marketing plan development. By matching user-input plans with a content database and proactively generating a list of business constraint information for questioning, the system can identify ambiguities and missing elements in the plan, guiding users to refine it through data-driven processes. This allows marketing campaign plans to reach their optimal state through a limited number of human-computer interactions before launch, thereby improving the accuracy of marketing campaign performance prediction and the scientific rigor of the plan. Attached Figure Description

[0046] Figure 1This is a basic flowchart illustrating a marketing campaign performance intelligent prediction and optimization method provided in one embodiment of the present invention.

[0047] Figure 2 This is a schematic diagram of the architecture of the historical knowledge base of the intelligent prediction and optimization method for marketing campaign effectiveness provided in one embodiment of the present invention.

[0048] Figure 3 This is a schematic diagram of the information knowledge base architecture of the marketing campaign effectiveness intelligent prediction and optimization method provided in one embodiment of the present invention.

[0049] Figure 4 This is a schematic diagram of the information knowledge base processing flow of the intelligent prediction and optimization method for marketing campaign effectiveness provided in one embodiment of the present invention. Detailed Implementation

[0050] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0051] Example 1, referring to Figures 1-4 As an embodiment of the present invention, a method for intelligent prediction and optimization of marketing campaign effectiveness is provided;

[0052] Step S1: Collect historical campaign data, performance data, and relevant information from social media platforms, and construct the correspondence between historical campaign data and performance data as a historical knowledge base. Use relevant information as an information knowledge base. Social media platforms include, but are not limited to, Xiaohongshu, Douyin, and Bilibili, and can also be extended to content platforms such as Weibo, Zhihu, and Douban. Use API interfaces to capture historical campaign data for specific product categories (such as "tissues").

[0053] The relevant information includes the content of the notes published by each competitor company, the note ID, the publication time, and the interaction data (number of reads, number of likes).

[0054] In this embodiment, by constructing a historical knowledge base and an information knowledge base, a solid data foundation is provided for the prediction and optimization of marketing campaign plans.

[0055] Step S2: Extract semantic features and scene information from the historical knowledge base and the information knowledge base, classify the semantic features, add tag information, and generate a content database; preferably, use existing NLP natural language processing models (such as BERT or Word2Vec) to extract semantic features from the historical knowledge base and the information knowledge base.

[0056] In this embodiment, the semantic features and scenario information of the multi-dimensional and multi-level historical knowledge base and information knowledge base are classified and metadata-processed, which can more precisely depict the various elements of the marketing campaign, thereby improving the accuracy of subsequent feature extraction and tagging, and laying the foundation for generating a high-quality content database.

[0057] Step S3: The user inputs a marketing campaign plan, the semantic features and scenario information in the marketing campaign plan are extracted, and the semantic features and scenario information of the marketing campaign plan are matched with the content database to obtain tag information. Based on the tag information and scenario information, a list of business constraint information is generated and questions are asked to the user.

[0058] In this embodiment, the static, one-off solution development process is transformed into a dynamic, self-improving workflow, realizing a shift from "people searching for data" to "data serving people." This not only lowers the professional threshold for marketing decisions but also ensures, through continuous iterative optimization, that the final solution maximizes the use of data value, achieving optimal marketing results and bringing significant practical application value to enterprises.

[0059] Step S4: Based on the information content of the user's answer to the business constraint information list, update the marketing campaign plan, and repeat steps S2 and S3 until the completeness of the requirements reaches the preset threshold or the user actively stops asking questions. Then, output the last updated marketing campaign plan as the final plan.

[0060] Step S5: Based on the correspondence between historical campaign data and performance data, predict the performance of the latest updated marketing campaign plan, obtain the predicted performance data, and adjust the marketing campaign plan when the actual performance data deviates from the prediction by more than a threshold.

[0061] In this embodiment, by iteratively optimizing steps S2 and S3 until the completeness of the demand reaches a preset threshold or the user actively terminates the question, the solution is ensured to reach its optimal state through a limited number of human-computer interactions before deployment. By matching user solutions with the content database and generating a list of business constraint information, a shift from passive analysis to proactive decision-making is achieved. This identifies missing elements in the solution, which helps improve the efficiency and accuracy of marketing deployment plan development.

[0062] Historical deployment data includes supply-side data, demand-side data, demographic data, and time data;

[0063] Supply-side data includes content data, content format, number of placements, budget data, and KPI data;

[0064] Demand-side data includes search terms, upstream and downstream vocabulary, similar words, search habits, and search paths;

[0065] Audience data includes audience segments, audience segment characteristics, audience segment names, and the number of audience segments;

[0066] Time-based data includes marketing milestones, communication cycles, campaign duration, and recent trending topics;

[0067] Performance data includes readership, page views, likes, comments, search conversion rate, and revenue.

[0068] In this embodiment, historical campaign data refers to a structured data set that records detailed information of all past marketing activities. It is used to analyze the inherent patterns of historical successes and failures. The historical campaign data is provided by the enterprise's data warehouse or advertising campaign management system. After data cleaning and formatting, it is transmitted to the knowledge base construction module. After being associated with the performance data, it forms a historical knowledge base and provides raw materials for subsequent feature extraction.

[0069] Furthermore, historical campaign data specifically includes supply-side data, demand-side data, audience data, and time data. Supply-side data refers to a set of parameters describing the attributes of the marketing campaign itself, including content data (such as text, images, and video links), content format (such as text, short videos, and live streams), number of campaigns (such as the number of ads), budget data, and KPI data (such as expected click-through rate and conversion rate). Demand-side data refers to a set of data reflecting the target users' search intent and behavioral paths, including search terms, upstream and downstream keywords (words searched by users before and after the target term), similar words, search habits, and search paths. Audience data refers to a set of profile information of the target audience, including audience segments (such as "maternal and infant - new mothers"), audience segment characteristics (such as age, region, and interest tags), audience segment names, and the number of audience segments. Time data refers to a set of information related to the time nodes of the marketing campaign, including marketing events (such as 618 and Double 11), the communication cycle, the campaign time, and recent hot topics.

[0070] Furthermore, performance data refers to a set of key performance indicators (KPIs) used to measure the results of historical campaigns, establishing a causal relationship between campaign actions and their effects. Performance data originates from backend reports of advertising platforms or third-party monitoring tools, and is linked to historical campaign data via a unique campaign ID, collectively forming a historical knowledge base. Specific performance data includes readership, page views, likes, comments, search conversion rates, and revenue data.

[0071] Based on search terms, upstream and downstream terms, and similar terms from historical campaign data, relevant information from social media platforms is collected, and the searched information is used as an information knowledge base.

[0072] In this embodiment, social media platform information refers to unstructured or semi-structured information reflecting public opinion and trending topics obtained from social media platforms such as Weibo, Douyin, and Xiaohongshu. This information is used to capture market dynamics and changes in user interests. Social media platform information is collected by calling the API interfaces provided by each platform or by using web crawling technology. After preliminary processing, it forms an information knowledge base, providing real-time and fresh external knowledge for the content database. By automatically collecting and standardizing multi-platform social media data and structuring information flow, the manpower required for content analysis and opportunity identification is reduced, and the efficiency of insight is improved.

[0073] Specifically, using relevant information as an information knowledge base includes:

[0074] Extract supply-side data, demand-side data, population data, and time data from the information knowledge base, and classify the extracted supply-side data, demand-side data, population data, and time data to obtain a classified information set. Further classify the information set according to the respective categories of supply-side data, demand-side data, population data, and time data to obtain a meta-data set. Extract semantic features from the meta-data set to obtain a semantic feature set, classify the semantic features in the semantic feature set, and add label information.

[0075] Contextual information is constructed by combining the audience package name from the audience data, the nouns from the demand-side data, and the environment from the content data. Contextual information includes the audience package name, nouns, and environment.

[0076] In this embodiment, scene information refers to a set of tags describing the specific environment or background in which the data is generated or applied. This is used to limit the scope of semantic features and improve matching accuracy. For example, in the case of family travel, the audience package name would be "Mom," "Dad," and "Child" (including keywords like "infant," "child," or "kid"), the noun would be "travel," and the environment would be "outdoor." This is the integrated scene information. Currently, scene information is generally described as "party scene" or "care scene." Such broad descriptions are often ambiguous. For example, is a dining party a birthday party, a company team-building party, a family gathering, or a hot pot or barbecue? Recommending based on such scenarios often doesn't yield the desired results and can even cause user resentment. For instance, recommending a birthday cake for a company team-building party would mean recommending cakes for the company team-building event. However, according to the scene information structure in this embodiment, the audience package would be "employees and bosses" plus "party" (noun) plus "restaurant" or "hotel" (environment). In this embodiment, the crowd information is first extracted, then action verbs (such as "eat", "drink", "do") and environmental nouns (such as "hot pot", "racing", "home") in the text are extracted, and finally combined to form the scene information.

[0077] In this embodiment, data is automatically acquired in real time from internal enterprise systems and external platforms, replacing the traditional method of manually collecting and organizing data. By automatically binding the deployment behavior with the effect data through preset association rules, a traceable and analyzable historical knowledge base is formed, eliminating the analysis bias caused by data silos and manual operation, and laying a high-quality data foundation for subsequent intelligent prediction.

[0078] In step S3, the user inputs a marketing campaign plan, and the extraction of semantic features and scenario information from the marketing campaign plan specifically includes:

[0079] Step S31: Obtain the user's marketing campaign plan and extract the supply-side data, demand-side data, audience data, and time data of the marketing campaign plan;

[0080] Step S32: Classify the supply-side data, demand-side data, audience data, and time data extracted from the marketing campaign plan to obtain the information set after classification of the marketing campaign plan;

[0081] Step S33: After classifying the information set of marketing campaign plans, further classify it according to the categories of supply-side data, demand-side data, audience data, and time data to obtain the meta-data set of marketing campaign plans;

[0082] Step S34: Extract semantic features from the metadata of the marketing campaign plan to obtain the semantic feature set of the marketing campaign plan, establish the correspondence between the semantic feature set of the marketing campaign plan and the marketing campaign plan, and label the scenario information of the marketing campaign plan.

[0083] In this embodiment, semantic features refer to a set of vectors or keywords extracted from data such as text and images that can express their core meaning, used to enable machines to deeply understand the data content. In this invention, semantic features are extracted by the content database generation module from historical and information knowledge bases using natural language processing (NLP) models (such as BERT and Word2Vec), and serve as the basis for subsequent matching.

[0084] Furthermore, the scene information is annotated by the content database generation module based on the data's metadata (such as timestamps and source platforms) and content context, and stored in the content database along with semantic features.

[0085] The semantic features and contextual information of the marketing campaign are matched with the content database to obtain tag information. Based on the tag and contextual information, a list of business constraint information is generated, and questions are asked to users, specifically including:

[0086] Step S35: Perform a first match between the semantic features in the marketing campaign plan and the content database to obtain the first match result. Then, perform a second match between the first match result and the scenario information corresponding to the marketing campaign plan to obtain the second match result. At the same time, obtain the tag information and scenario information corresponding to the second match result.

[0087] In this embodiment, the content database refers to a database that stores structured knowledge that has undergone semantic and tagging processing, serving as a knowledge source for matching user plans. The content database is constructed by the content database generation module, and its data comes from the two knowledge bases generated in step S1. It is output to the interaction matching module, using historical data (historical knowledge base) and external data (information knowledge base) as the basis for analyzing existing data (user-input marketing campaign plans).

[0088] Step S36: Classify the data according to the tag information corresponding to the secondary matching results. The classification content includes the historical knowledge base and the information knowledge base. Select supply-side data, demand-side data, population data and time data from the historical knowledge base obtained from the secondary matching results. Then, generate an initial business constraint information list based on the supply-side data, demand-side data, population data and time data, as well as the corresponding scenario information.

[0089] In this embodiment, the business constraint information list refers to a structured list of questions automatically generated based on the matching results between the user's plan and the content database, used to guide the user to improve the details of the plan.

[0090] In this embodiment, the marketing campaign plan input by the user first undergoes a processing flow similar to step S2, namely, extracting its supply-side data, demand-side data, audience data, and time data, and classifying and extracting semantic features to obtain the semantic feature set and scenario information of the marketing campaign plan. The business constraint information list contains missing data in the supply-side, demand-side, audience, and time data, such as missing marketing nodes, dissemination cycles, campaign times, and recent hot topics. Alternatively, preset questions can be added to the template of the business constraint information list, such as target audience, promotion channels, budget range, brand tone breakthrough, and acceptance of AI-generated materials. Asking questions helps users consider their limited perspectives and potential risks related to budget and tone in advance, ensuring improved feasibility.

[0091] Based on the user account that inputs the marketing campaign plan, the system obtains the user's identity information. According to the user's identity information and preset access permissions, the initial list of business constraints is filtered. The filtered list is then displayed or sent to the user, prompting questions. In this implementation, the system obtains the user's identity information (e.g., "Junior Planner," "Senior Manager") based on their account and filters the initial list of business constraints according to preset access permissions. For example, for a "Junior Planner," only basic questions about content format and campaign timing are displayed; while for a "Senior Manager," additional strategic questions about budget allocation and KPI setting are displayed. This system provides differentiated decision support based on user roles, ensuring efficient and targeted interaction. It simplifies the complex marketing decision-making process into a series of human-computer dialogues, lowers the decision-making threshold, and allows for faster access to effective information. It avoids asking users questions outside their knowledge domain, which could lead to user complaints, such as asking a technical person a financial question. Users may become impatient when faced with questions they cannot answer on the business constraint information list. Too many questions that they are simply unable to answer (for example, a junior employee is unaware of the company's legal risk threshold) can lead to a negative user experience. This embodiment takes this into full consideration, and instead of blindly asking users all questions, it asks questions based on the user's identity.

[0092] In this embodiment, through primary and secondary matching, marketing campaign plans and content databases with consistent semantic features and scenario information can be matched. This facilitates the proactive discovery of potential optimization points in users' marketing campaign plans and guides users' thinking through a checklist of business constraint information questions, replacing the traditional method of relying on expert experience for plan review. This provides differentiated decision support based on user roles, ensuring efficient and targeted interaction, simplifying the complex marketing decision-making process into a series of human-computer dialogues, and significantly lowering the decision-making threshold. In one embodiment, AI can also simulate experts to ask questions as supplementary inquiries to the user.

[0093] Step S41: The user answers the questions based on the business constraint information list and enters the information into the system;

[0094] Step S42: Extract the supply-side data, demand-side data, audience data, and time data of the answer content, and update the supply-side data, demand-side data, audience data, and time data of the answer content in the corresponding places in the marketing campaign plan;

[0095] Step S42: Re-enter the updated marketing campaign plan, and repeat steps S2 and S3 until the completeness of the demand reaches a preset threshold or the user actively terminates the question. Then, output the last updated marketing campaign plan as the final plan. Preferably, the preset threshold is used to end the loop when the completeness of the demand reaches the preset threshold (e.g., 95%), or when the user actively clicks the "Complete Optimization" button to terminate the question.

[0096] In this embodiment, a technical mechanism is employed to gradually bring the marketing campaign plan closer to its optimal state through a cyclical process of "user response - plan update - re-matching - asking again". By using a list of business constraint information to ask questions, users are forced to consider the business boundaries behind their needs (such as budget, risk, brand), which makes the user's implicit constraints explicit and defines the marketing campaign plan within the "feasibility domain", thereby improving the feasibility of the marketing campaign plan.

[0097] Increasing the number and depth of questions inevitably prolongs the interaction time. In particular, an increase in the number of interaction rounds can cause user impatience, leading them to ask questions even if they know the answer, or questions they cannot answer. This means there's a need to simultaneously satisfy two opposing requirements: providing quick answers (satisfying immediate user satisfaction) and delaying answers to allow for further questioning (ensuring the accuracy of the user's needs). Therefore, this implementation uses the most efficient questioning method: a question list, which is also a list of business constraint information. Furthermore, it filters out questions that users cannot answer based on their identity, as well as questions already explained in the user's marketing campaign. Users can complete the constraint information all at once or in batches by filling out a form. This process is asynchronous; users can pause at any time to reflect, consult resources, and then continue, improving interaction efficiency. This ensures that while reducing the number of interaction rounds, interaction efficiency is increased, guaranteeing a smooth user experience.

[0098] Step S51: Record the correspondence between supply-side data, demand-side data, population data, time data, and effect data respectively. Analyze the correspondence between supply-side data, demand-side data, population data, time data, and effect data according to the preset volatility. When the volatility of supply-side data, demand-side data, population data, or time data is less than the preset volatility, it is considered as no change. When the volatility of supply-side data, demand-side data, population data, or time data is greater than the preset volatility, record the correspondence between supply-side data, demand-side data, population data, or time data and effect data.

[0099] Step S52: Based on the correspondence between the recorded supply-side data, demand-side data, audience data or time data and effect data, calculate the effect data corresponding to the supply-side data, demand-side data, audience data or time data in the last updated marketing campaign plan, and use the calculation results as the predicted effect data.

[0100] Step S53: Record the actual performance data of the marketing campaign. When the deviation between the actual performance data and the prediction exceeds the threshold, adjust the supply-side data, demand-side data, audience data or time data and performance data in the marketing campaign.

[0101] By matching user input plans with a content database and proactively generating a list of business constraint information for questioning, the system can identify ambiguities and missing elements in the plans, guiding users to make data-driven improvements. This allows marketing campaign plans to reach their optimal state through a limited number of human-computer interactions before launch, thereby improving the accuracy of marketing campaign performance prediction and the scientific nature of the plans.

[0102] Example 2, refer to Figure 1-4 This is another embodiment of the present invention, which is based on the previous embodiment and further includes:

[0103] The data acquisition module is used to collect historical campaign data, performance data, and social media platform information, and to build a correspondence between historical campaign data and performance data to form a historical knowledge base, as well as to use social media platform information as an information knowledge base;

[0104] The feature analysis module is used to extract semantic features and scene information from historical knowledge bases and information knowledge bases, classify the semantic features and add tags to generate a content database;

[0105] The interactive matching module is used to extract the semantic features and scenario information of the marketing campaign input by the user, and match the semantic features and scenario information of the marketing campaign with the content database to generate a list of business constraint information and ask questions to the user.

[0106] The optimization iteration module updates the marketing campaign plan based on the user's response to the business constraint information list, and triggers the content database generation module and the interaction matching module to execute in a loop until the preset termination condition is met, and then outputs the final plan.

[0107] The performance monitoring module is used to compare and analyze the actual marketing campaign results with the predicted results. When the deviation between the actual results and the predictions exceeds a threshold, it automatically triggers model updates and strategy adjustments.

[0108] In this embodiment, by constructing a historical knowledge base and an information knowledge base, and by using semantic feature extraction and tagging technology, marketing data is transformed into structured knowledge that can be understood and utilized by machines, replacing the traditional manual data integration and analysis methods, and significantly improving the efficiency of marketing plan formulation.

[0109] By matching user input plans with a content database and proactively generating a list of business constraint information for questioning, the system can identify ambiguities and missing elements in the plans, guiding users to make data-driven improvements. This allows marketing campaign plans to reach their optimal state through a limited number of human-computer interactions before launch, thereby improving the accuracy of marketing campaign performance prediction and the scientific nature of the plans.

[0110] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0111] It should be noted that the above 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent prediction and optimization of marketing campaign effectiveness, characterized in that, include: Step S1: Collect historical campaign data, performance data, and relevant information from social media platforms, and construct the correspondence between historical campaign data and performance data as a historical knowledge base, and use the relevant information as an information knowledge base; Step S2: Extract semantic features and scene information from the historical knowledge base and the information knowledge base, classify the semantic features, add tag information, and generate a content database; Step S3: The user inputs a marketing campaign plan, the semantic features and scenario information in the marketing campaign plan are extracted, and the semantic features and scenario information of the marketing campaign plan are matched with the content database to obtain tag information. Based on the tag information and scenario information, a list of business constraint information is generated and questions are asked to the user. Step S4: Based on the information content of the user's answer to the business constraint information list, update the marketing campaign plan, and repeat steps S2 and S3 until the completeness of the requirements reaches the preset threshold or the user actively stops asking questions. Then, output the last updated marketing campaign plan as the final plan. Step S5: Based on the correspondence between historical campaign data and performance data, predict the performance of the latest updated marketing campaign plan, obtain the predicted performance data, and adjust the marketing campaign plan when the actual performance data deviates from the prediction by more than a threshold.

2. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 1, characterized in that: The historical deployment data includes supply-side data, demand-side data, population data, and time data; The supply-side data includes content data, content format, distribution volume, budget data, and KPI data; The demand-side data includes search terms, upstream and downstream vocabulary, similar words, search habits, and search paths; The crowd data includes crowd segments, crowd segment characteristics, crowd segment names, and the number of crowd segments; The time data includes marketing milestones, dissemination cycles, campaign timing, and recent trending topics; The performance data includes readership, pageviews, likes, comments, search conversion rate, and revenue data.

3. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 2, characterized in that: Based on the search terms, upstream and downstream vocabulary, and similar words in the historical campaign data, relevant information from social media platforms is collected, and the searched information is used as an information knowledge base.

4. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 3, characterized in that: Specifically, the information knowledge base includes the following: Extract supply-side data, demand-side data, population data, and time data from the information knowledge base, and classify the extracted supply-side data, demand-side data, population data, and time data to obtain a classified information set. Further classify the information set according to the respective categories of supply-side data, demand-side data, population data, and time data to obtain a meta-data set. Extract semantic features from the meta-data set to obtain a semantic feature set, and classify the semantic features in the semantic feature set and add label information. Scene information is constructed by combining the audience package name from the audience data, the nouns from the demand-side data, and the environment from the content data. The scene information includes the audience package name plus the nouns plus the environment.

5. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 4, characterized in that, In step S3, the extraction of semantic features and scenario information from the user-input marketing campaign plan specifically includes: Step S31: Obtain the user's marketing campaign plan and extract the supply-side data, demand-side data, audience data, and time data of the marketing campaign plan; Step S32: Classify the supply-side data, demand-side data, audience data, and time data extracted from the marketing campaign plan to obtain the information set after classification of the marketing campaign plan; Step S33: After classifying the information set of marketing campaign plans, further classify it according to the categories of supply-side data, demand-side data, audience data, and time data to obtain the meta-data set of marketing campaign plans; Step S34: Extract semantic features from the metadata of the marketing campaign plan to obtain the semantic feature set of the marketing campaign plan, establish the correspondence between the semantic feature set of the marketing campaign plan and the marketing campaign plan, and label the scenario information of the marketing campaign plan.

6. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 5, characterized in that: The semantic features and scenario information of the marketing campaign are matched with the content database to obtain tag information. Based on the tag information and scenario information, a list of business constraint information is generated, and questions are asked to the user, specifically including: Step S35: Perform a first match between the semantic features in the marketing campaign plan and the content database to obtain a first match result. Then, perform a second match between the first match result and the scene information corresponding to the marketing campaign plan to obtain a second match result. At the same time, obtain the tag information and scene information corresponding to the second match result. Step S36: Classify the data according to the tag information corresponding to the secondary matching results. The classification content includes a historical knowledge base and an information knowledge base. Select supply-side data, demand-side data, population data and time data from the historical knowledge base corresponding to the secondary matching results. Then, generate an initial business constraint information list based on the supply-side data, demand-side data, population data and time data, as well as the corresponding scenario information.

7. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 6, characterized in that, Based on the user account that inputs the marketing campaign plan, the system obtains the user's identity information, adjusts the user's identity information and preset identity operation permissions, filters the contents of the initial business constraint information list, displays or sends the filtered business constraint information list to the user, and asks questions to the user.

8. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 7, characterized in that: Step S4 specifically includes: Step S41: The user answers the questions based on the business constraint information list and inputs the answers into the system; Step S42: Extract the supply-side data, demand-side data, audience data, and time data from the answer content, and update the corresponding parts of the marketing campaign plan with the extracted supply-side data, demand-side data, audience data, and time data from the answer content. Step S43: Re-enter the updated marketing campaign plan, and repeat steps S2 and S3 until the completeness of the requirements reaches the preset threshold or the user actively stops asking questions. Then, output the last updated marketing campaign plan as the final plan.

9. The intelligent prediction and optimization method for marketing campaign effectiveness as described in claim 8, characterized in that: Step S5 specifically includes: Step S51: Record the correspondence between supply-side data, demand-side data, population data, time data, and effect data respectively. Analyze the correspondence between supply-side data, demand-side data, population data, time data, and effect data according to the preset volatility. When the volatility of supply-side data, demand-side data, population data, or time data is less than the preset volatility, it is considered as no change. When the volatility of supply-side data, demand-side data, population data, or time data is greater than the preset volatility, record the corresponding correspondence between supply-side data, demand-side data, population data, or time data and effect data. Step S52: Based on the correspondence between the recorded supply-side data, demand-side data, audience data or time data and effect data, calculate the effect data corresponding to the supply-side data, demand-side data, audience data or time data in the last updated marketing campaign plan, and use the calculation results as the predicted effect data. Step S53: Record the actual performance data of the marketing campaign. When the deviation between the actual performance data and the prediction exceeds the threshold, adjust the supply-side data, demand-side data, audience data or time data and performance data in the marketing campaign.

10. A marketing campaign performance intelligent prediction and optimization system, the system being used to execute the marketing campaign performance intelligent prediction and optimization method according to any one of claims 1-9, characterized in that, The system includes: The data acquisition module is used to collect historical campaign data, performance data, and social media platform information, and to construct the correspondence between the historical campaign data and performance data to form a historical knowledge base, and to use the social media platform information as an information knowledge base; The feature analysis module is used to extract semantic features and scene information from the historical knowledge base and the information knowledge base, and to classify and add tags to the semantic features to generate a content database; The interactive matching module is used to extract the semantic features and scenario information of the marketing campaign input by the user, and match the semantic features and scenario information of the marketing campaign with the content database to generate a list of business constraint information and ask questions to the user. The optimization iteration module is used to update the marketing campaign plan based on the user's response to the business constraint information list, and trigger the content database generation module and the interaction matching module to execute in a loop until the preset termination condition is met, and then output the final plan. The performance monitoring module is used to compare and analyze the actual marketing campaign results with the predicted results. When the deviation between the actual results and the predictions exceeds a threshold, it automatically triggers model updates and strategy adjustments.

Images