Brand promotion task multi-dimensional index evaluation and automatic grading system

Abstract

The invention discloses a multi-dimensional index evaluation and automatic grading system for brand promotion tasks, which relates to the technical field of brand promotion, and comprises a data processing module used for acquiring task feature data from a promotion task database, including dynamic features of target population division, delivery channel selection and budget scale distribution, removing duplicate records and missing values through a data cleaning technology, and converting the feature data into a structured feature data set with a mean value of 0 and a variance of 1 by adopting a standardization technology to obtain the structured feature data set; according to the multi-dimensional index evaluation and automatic grading system for the brand promotion tasks, intelligent grading, optimal scheduling and dynamic adjustment of the promotion tasks are realized, the task execution efficiency and the resource utilization rate are improved, and more accurate decision support is provided for promotion activities.

Description

Technical Field

[0001] The present invention relates to the technical field of brand promotion, and specifically relates to a multi-dimensional index evaluation and automatic grading system for brand promotion tasks. Background Art

[0002] Promotion task management is a key area for modern enterprises to enhance their market competitiveness. The core lies in optimizing resource allocation and effect evaluation through scientific methods to achieve efficient market promotion. Reasonable task division and quality evaluation can not only improve promotion efficiency but also provide clear strategic guidance for decision-makers.

[0003] However, the current methods for promotion task management have significant limitations. Most solutions rely on manual experience or simple rule-based division, lacking systematic and intelligent support, resulting in inaccurate task grading results and difficulty in intuitively reflecting the task quality level. This extensive management method cannot meet the refined requirements in a complex market environment. In promotion task management, task automatic division and quality representation face multiple challenges. The primary problem is the diversity and complexity of task characteristics. Promotion tasks involve multiple dimensions, such as target population, delivery channels, and budget scale. The dynamic changes of these characteristics make it difficult for a single division standard to adapt. The resulting technical problem is how to extract key information from multi-dimensional characteristics and achieve automatic division. Further, due to the lack of a unified quality evaluation system, it is difficult to generate intuitive representation results for the divided tasks, leading to decision-makers being unable to quickly understand the task priorities and potential values. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-dimensional index evaluation and automatic grading system for brand promotion tasks to solve the problems existing in the prior art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A multi-dimensional index evaluation and automatic grading system for brand promotion tasks, comprising: A data processing module, configured to obtain task feature data from a promotion task database, including dynamic features of target population division, delivery channel selection, and budget scale allocation, remove duplicate records and missing values through data cleaning technology, and convert the feature data into a structured feature data set with a mean of 0 and a variance of 1 using standardization technology to obtain a structured feature data set; A status monitoring module, configured to obtain a task execution status log based on the structured feature data set. If the click-through rate or conversion rate in the task execution status log is continuously lower than a preset threshold, adjust the weight coefficients of target population preference and channel matching degree based on historical data fusion through feature weight update technology, and generate an optimized feature vector set using an incremental update strategy to obtain an optimized feature vector set; The task grouping and grading module is used to group tasks based on the optimized feature vector set by repeating the clustering analysis technique according to the target population preferences and channel matching degrees, and re-grade them based on the adjusted weight coefficients through the decision tree classification technique to generate an updated task grading result; The visualization module is used to generate a new task priority distribution map and a potential value heat map for the updated task grading result by using visualization techniques. The distribution map reflects the adjusted priority ratio, and the heat map is drawn based on the updated target population coverage rate and channel conversion rate to determine the dynamically adjusted task priority and potential value, and obtain a dynamically adjusted characterization result.

[0006] Preferably, the status monitoring module obtains a task execution status log according to the structured feature data set, including extracting key information on target population preferences, channel matching degrees, and budget constraints by using the principal component analysis technique for the structured feature data set, determining the weight coefficients of each feature dimension, generating a feature vector set including target population preferences, channel matching degrees, and budget constraints, and obtaining a feature vector.

[0007] Preferably, the status monitoring module obtains a task execution status log according to the structured feature data set, including if the weight coefficient in the feature vector set is higher than the preset threshold, then preliminarily grouping tasks based on target population preferences and channel matching degrees by using the clustering analysis technique to generate a task grouping set based on multi-dimensional dynamic features, and obtaining a task grouping set.

[0008] Preferably, the status monitoring module obtains a task execution status log according to the structured feature data set, including obtaining the feature vectors of each group of tasks from the task grouping set, and automatically grading the tasks based on target population preferences, channel matching degrees, and budget constraints by using the decision tree classification technique to generate a task grading result including high, medium, and low priorities, and obtaining a task grading result.

[0009] Preferably, the status monitoring module obtains a task execution status log according to the structured feature data set, including calculating the quality scores of each group of tasks through a quality assessment model for the task grading result. The model input includes the target population coverage rate, channel conversion rate, and budget utilization rate, generating a task quality score set, and obtaining a task quality score set.

[0010] Preferably, the status monitoring module obtains a task execution status log according to the structured feature data set, including generating a task priority distribution map and a potential value heat map by using visualization techniques based on the task quality score set. The distribution map shows the proportion of the number of tasks with high, medium, and low priorities, and the heat map is drawn based on the target population coverage rate and channel conversion rate to determine the task priority ranking and value distribution, and obtain a graded characterization result.

[0011] Preferably, the status monitoring module obtains a task execution status log based on the structured feature dataset, including extracting the task priority ranking from the hierarchical representation result, combining the potential value heat map, and optimizing and scheduling the tasks based on the task quality score and budget utilization rate through a sorting algorithm to generate a task execution sequence and obtain the task execution sequence.

[0012] Preferably, the status monitoring module obtains a task execution status log based on the structured feature dataset, including for the task execution sequence, collecting real-time data of task execution once per minute using real-time monitoring technology, including click-through rate, conversion rate, and budget consumption rate, and identifying traffic anomalies or sudden drops in conversion rate through an anomaly detection mechanism to generate a task execution status log and obtain the task execution status log.

[0013] Preferably, the data cleaning technology in the data processing module includes a duplicate record detection algorithm and a missing value filling algorithm, and the missing value filling algorithm uses a mean substitution method based on similar features.

[0014] Preferably, the standardization technology in the data processing module is the Z-score standardization method.

[0015] As can be seen from the above technical solutions, the present invention has the following beneficial effects: The multi-dimensional index evaluation and automatic grading system for brand promotion tasks obtains the target population, placement channels, and budget characteristics of the promotion tasks, extracts key information using principal component analysis to generate a feature vector set. It groups and grades the tasks using clustering analysis and decision tree technology, and calculates the quality score. Generates a task priority distribution map and a value heat map based on the grading results to determine the task execution sequence. During the execution process, it monitors the task status in real time, and when anomalies occur, dynamically adjusts the feature weights and re-performs task grading and priority ranking. The present invention realizes the intelligent grading, optimized scheduling, and dynamic adjustment of promotion tasks, improves the task execution efficiency and resource utilization rate, and provides more accurate decision-making support for promotion activities. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a connection diagram of the system modules of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] Such as Figure 1As shown in the figure, the present invention provides a technical solution: a multi-dimensional index evaluation and automatic grading system for brand promotion tasks, including: A data processing module, configured to obtain task feature data from a promotion task database, including dynamic features of target population division, delivery channel selection, and budget scale allocation, remove duplicate records and missing values through data cleaning technology, and convert the feature data into a structured feature data set with a mean of 0 and a variance of 1 by using standardization technology, so as to obtain a structured feature data set; A status monitoring module, configured to obtain a task execution status log according to the structured feature data set. If the click-through rate or conversion rate in the task execution status log is continuously lower than a preset threshold, adjust the weight coefficients of target population preference and channel matching degree based on historical data fusion through feature weight update technology, and generate an optimized feature vector set by using an incremental update strategy, so as to obtain an optimized feature vector set; A task grouping and grading module, configured to group tasks based on target population preference and channel matching degree by using repeated clustering analysis technology according to the optimized feature vector set, and re-grade based on the adjusted weight coefficients by using decision tree classification technology, so as to generate an updated task grading result; A visualization module, configured to generate a new task priority distribution map and a potential value heat map by using visualization technology for the updated task grading result. The distribution map reflects the adjusted priority ratio, and the heat map is drawn based on the updated target population coverage rate and channel conversion rate, so as to determine the dynamically adjusted task priority and potential value, and obtain a dynamically adjusted representation result.

[0019] The system first performs cleaning and standardization processing on the dynamic feature data in the promotion task database through the data processing module to form a unified structured feature data set, which solves the problems of original data noise and inconsistent scales. The status monitoring module monitors the task execution status in real time, identifies tasks with poor performance, and adjusts the target population preference and channel matching degree based on historical performance data by using feature weight update technology to ensure that the optimized feature vector can better reflect the current market situation. The task grouping and grading module uses repeated clustering and decision tree algorithms to achieve automatic grouping and grading, effectively reducing manual intervention and improving the scientificity and adaptability of grading. Finally, the visualization module converts the complex data analysis results into intuitive visualization graphics to help decision-makers quickly identify task priorities and potential values, so as to guide subsequent resource allocation and strategy adjustment.

[0020] 1. Data standardization The original feature data X is processed by standardization, and the z-score standardization formula is used: ; Among them, It is the standardized characteristic data value. By subtracting the mean and dividing by the standard deviation, the data is transformed into a distribution with a mean of 0 and a variance of 1, which helps to eliminate the influence of different characteristic data scales. μ is the mean of the characteristic data, σ is the standard deviation of the characteristic data, and the parameters μ and σ are statistically calculated based on the training dataset to ensure data balance.

[0021] 2. State Monitoring and Threshold Judgment The system continuously monitors the click-through rate (CTR) and the conversion rate (CVR) to determine whether they are below the preset thresholds , .

[0022] Judgment condition: CTR < or CVR < ; where and are determined based on the industry average level and business expectations of past tasks. For example = 0.02, = 0.005. 3. Feature Weight Update When it is below the threshold, a dynamic update method based on historical data is used to adjust the weight: ; where is the updated feature weight. It represents the importance coefficient of the current feature in the model after being adjusted by the dynamic update method. This value will replace the previous weight and be used for subsequent grouping and grading calculations. represents the original weight, represents the performance score of the i-th task in the historical data, represents the performance metric of the corresponding feature, α is the update coefficient (usually taken as 0.7 - 0.9), N is the number of historical tasks, and the update coefficient α is obtained by empirical tuning or cross-validation to ensure that the new weight takes into account both historical performance and current adjustment.

[0023] 4. Incrementally Update Feature Vectors The feature vectors are updated using: ; where is the updated feature vector. It represents the adjusted representation of the current task in the feature space and is used for subsequent clustering and grading calculations. is the original feature vector. It represents the feature vector before update and reflects the previous attribute features of the task. is the incremental adjustment amount of the feature vector. It is calculated based on the deviation of the most recent data and reflects the trend of recent market or user behavior changes; where Calculate the deviation amount based on the most recent data to dynamically reflect the market change trend.

[0024] 5. Repeated clustering and grading Adopt the K-means or DBSCAN clustering algorithm and perform grouping based on the optimized feature vector set: ; Among them, C is the division result of clustering (the set of all categories). The goal is to find an optimal C. argmin is the operator for finding the minimum value, indicating that among all possible clustering divisions C, find the division that minimizes the objective function. K is the number of clusters, that is, the number of classification categories. is the set of the k-th category (cluster), which contains all data points belonging to this category. x is the data point, that is, a task feature vector in the optimized feature vector set. is the centroid of the k-th category (cluster), that is, the average position of all data points in this category. The initial value of K is determined according to the "elbow method".

[0025] 6. Decision tree grading Construct a decision tree based on the adjusted weight coefficient, and select the splitting feature according to the information gain. Information gain formula: ; Among them, IG(D,A) is the information gain, indicating how much the uncertainty of the data set is reduced after selecting the feature A to divide the data set D. The greater the information gain, the more suitable the feature A is as the splitting attribute of the decision tree. D is the current data set, which contains the data records (task feature data) for training. A is the feature, which is the candidate attribute used to divide the data set D, such as "channel matching degree" or "target population preference", etc. Entropy(D) is the entropy of the data set D, which measures the uncertainty or purity in the data set. V is the number of possible values of the feature A. is the corresponding sub-data set when the feature A takes the v-th value.

[0026] 7. Visualization output According to the updated grouping and grading results, draw a priority distribution diagram (bar chart or stacked bar chart) and a potential value heat map (two-dimensional or three-dimensional color mapping diagram) to provide an intuitive analysis basis.

[0027] Through automatic data cleaning and standardization, the system improves the quality and consistency of the input data; through real-time monitoring and dynamic weight update, the state feedback during the task execution process can be quickly reflected in the task model, enhancing the system's adaptability; advanced clustering and classification algorithms are used to automatically complete the grouping and grading of tasks, significantly reducing labor costs and improving decision-making efficiency; the visualization module further improves the readability and operability of the results, enabling managers to more intuitively grasp the task priorities and potential values, thereby optimizing the brand promotion strategy and enhancing the promotion effect and return on investment.

[0028] In practical applications, a large e-commerce platform plans to carry out a brand promotion activity during the "Double Eleven" period, targeting users aged 18 to 35 who prefer technology products and fashion items. The main advertising channels selected by the platform include social media (such as Douyin and Weibo), search engine advertising, and in-APP recommendation positions. The initial budget is 3 million yuan, and a dynamic allocation strategy is adopted to adapt to market feedback. The system first extracts feature data from the platform's "Double Eleven" promotion tasks in the past five years. After data cleaning and standardization, a structured input data set is formed. In the initial operation stage, the system finds that the conversion rate of some tasks is lower than the preset threshold of 0.5%. The status monitoring module is immediately activated. By using the feature weight update method, the weight coefficients of the target population preferences and channel matching degrees are dynamically adjusted. At the same time, the incremental update strategy is applied to optimize the feature vector to reflect the latest market changes. Subsequently, the system uses the K-means algorithm to cluster the optimized feature vectors, dividing the tasks into three categories: high value, medium value, and low value. Then, a decision tree is constructed based on the adjusted weight coefficients to re-grade each task. Finally, the visualization module generates an updated task priority distribution map and a potential value heat map, clearly showing the priority ratios of each task and the corresponding market potential. According to these analysis results, the management team optimizes the resource allocation strategy and increases the overall conversion rate to 1.2% in the subsequent execution, achieving an approximate 30% increase in the return on investment, fully verifying the high efficiency and adaptability of the system in actual brand promotion tasks.

[0029] Based on the structured feature data set, the status monitoring module obtains the task execution status log, including using the principal component analysis technique to extract the key information of the target population preferences, channel matching degrees, and budget constraints for the structured feature data set, determining the weight coefficients of each feature dimension, generating a feature vector set including the target population preferences, channel matching degrees, and budget constraints, and obtaining the feature vector set.

[0030] 1. Principal Component Analysis (PCA) Dimensionality Reduction Perform PCA processing on the structured feature data set X by calculating the covariance matrix: ; Perform eigen - decomposition on the covariance matrix: ; where Σ is the covariance matrix, which describes the covariance relationship between features and is used to measure the linear correlation between features. n is the number of data samples. X is the matrix representation of the structured feature dataset, with rows representing samples and columns representing features, usually standardized (mean 0, variance 1). is the transpose matrix of matrix X (rows and columns are interchanged). Q is the eigen - vector matrix, and the column vectors are the eigen - vectors of Σ, representing the directions of the principal components. is the eigenvalue matrix, which contains the eigenvalues corresponding to each principal component and reflects the contribution degree of this principal component to the total variance (the larger, the more important). is the transpose of the eigen - vector matrix Q.

[0031] 2. Feature dimension selection Select the first k principal components whose cumulative contribution rate reaches a preset threshold (e.g., 95%) to eliminate redundant and noisy data and ensure maximum information retention.

[0032] 3. Determination of weight coefficients Determine the weight coefficients of each feature according to the selected principal - component loadings : ; k is the number of selected principal components or the number of features considered. is the weight coefficient of the j - th feature. It represents the importance ratio of feature j in the overall feature vector and is used for subsequent feature - vector construction. is the loading of feature j in the direction of the principal component. It reflects the contribution degree or correlation of this feature in the corresponding principal component.

[0033] 4. Construction of the feature - vector set Combine the target - population preference, channel - matching degree, and budget constraint according to the determined weight coefficients to form a feature vector: ; where, is the feature vector of the i - th task. It contains the weighted combination of three features: target - population preference, channel - matching degree, and budget constraint, and is used for subsequent analysis (e.g., clustering, classification); are the standardized feature values of the target - population preference, channel - matching degree, and budget constraint of the i - th task respectively; are the corresponding weight coefficients.

[0034] Finally, the obtained feature - vector set will be used for subsequent status monitoring, grouping, and grading.

[0035] This method effectively reduces the feature dimension, reduces noise interference, and improves the efficiency and accuracy of feature extraction by introducing principal component analysis technology. By adopting a weight assignment strategy based on principal component loadings, it ensures the reasonable contribution of each feature dimension to the overall analysis result, overcoming the deviation problem that may be brought about by artificially setting weights. The construction of the feature vector set enables the system to transform complex multi-dimensional features into concise vector representations, facilitating subsequent clustering, classification, and visualization analysis, and improving the system's adaptive ability and the execution efficiency of promotion tasks.

[0036] Based on the structured feature dataset, the status monitoring module obtains the task execution status log. If the weight coefficient in the feature vector set is higher than the preset threshold, the tasks are initially grouped based on the target population preference and channel matching degree through clustering analysis technology, generating a task grouping set based on multi-dimensional dynamic features, and obtaining the task grouping set.

[0037] 1. Weight threshold judgment First, the system monitors the weight coefficients of each feature in the feature vector set . If there exists: ; where, is the weight coefficient of the j-th feature; is the preset threshold (set according to business experience or historical data, for example, 0.4). When the weight exceeds this threshold, it indicates that the current feature has a significant impact on the task performance, triggering subsequent clustering grouping.

[0038] 2. Extract the feature dimensions for grouping Select the target population preference and channel matching degree as the main features for initial grouping.

[0039] Construct the feature sub-vector: ; where, is the sub-feature vector of the i-th task, used for initial task grouping, including two weighted features of target population preference and channel matching degree; is the weight coefficient corresponding to the target population preference, reflecting the importance of this feature to the overall feature representation; is the weight coefficient corresponding to the channel matching degree, reflecting the importance of this feature; is the normalized feature value of the target population preference, indicating the feature performance of the i-th task in this dimension; is the normalized feature value of the channel matching degree, indicating the feature performance of the i-th task in this dimension.

[0040] 3. Perform clustering analysis Use the K-means or DBSCAN algorithm to cluster all tasks' ​

[0041] 4. Generate the task grouping set After clustering is completed, the task grouping set is output , where each contains promotion tasks with similar performances, facilitating subsequent classification and optimization processing.

[0042] is the k-th task group, and the tasks within the group are similar in characteristics such as target population preference and channel matching degree, facilitating subsequent unified optimization and management.

[0043] By performing automated grouping on tasks with significant weight features, the system can quickly identify task sets with similar target population preferences and channel matching degrees, avoiding subjective biases caused by manual division, and improving the efficiency and scientific nature of task management. Clustering analysis reduces the complexity of the high-dimensional feature space while retaining the most critical information, making subsequent task optimization, classification, and resource allocation more accurate. In addition, the dynamic trigger mechanism (weight exceeding the threshold) ensures that the system can respond in a timely manner when significant changes occur in market characteristics, enhancing the adaptability and real-time nature of the system.

[0044] The status monitoring module obtains the task execution status log based on the structured feature data set, including obtaining the feature vectors of each group of tasks from the task grouping set, and performing automated classification of tasks based on target population preference, channel matching degree, and budget constraint through decision tree classification technology, generating a task classification result containing high, medium, and low priorities, and obtaining the task classification result.

[0045] 1. Feature vector extraction For each task i, extract the complete weighted feature vector from the previously generated task grouping set: ; where is the weighted feature vector of the i-th task, containing three weighted features for subsequent classification or other analyses; is the standardized feature value of the target population preference; is the standardized feature value of the channel matching degree; is the budget constraint (standardized feature value); is the corresponding feature weight coefficient.

[0046] 2. Classification rule generation Based on the decision tree training results, the system automatically generates classification rules. The final classification results include three categories: high priority, medium priority, and low priority, and the priority is determined according to the comprehensive performance of task features. For example: High priority: Both the population preference and the channel matching degree are high, and the budget constraint is moderate or low.

[0047] Medium Priority: At least one core feature performs moderately.

[0048] Low Priority: The main features are low or the budget is limited.

[0049] 3. Output the task classification result The priority label for each task is assigned to form the task classification result for subsequent resource allocation and optimization strategies.

[0050] This method uses an automated decision tree classification mechanism to avoid the subjective bias of manually setting rules. It can dynamically determine the splitting features and classification rules based on actual data to ensure the scientific nature and adaptability of classification. By introducing budget constraints as one of the classification features, the classification result takes into account both market potential and resource feasibility while considering. The classification output result is concise and clear (high, medium, low priority), which is convenient for the management team to quickly formulate subsequent strategies, significantly improving the task management efficiency and the return on investment in promotion.

[0051] Based on the structured feature dataset, the status monitoring module obtains the task execution status log, including calculating the quality score of each group of tasks through a quality assessment model for the task classification result. The model inputs include the target population coverage rate, channel conversion rate, and budget utilization rate, generating a set of task quality scores.

[0052] 1. Input indicator collection For each group of tasks , extract the following three core indicators: Target population coverage rate ( ): The proportion of the number of target populations covered to the expected population.

[0053] Channel conversion rate ( ): The conversion effect of the task group on the selected channel (number of conversions / number of clicks).

[0054] Budget utilization rate ( ): The proportion of the used budget to the planned budget.

[0055] 2. Standardization processing To eliminate the scale impact of different indicators, use min-max normalization or z-score standardization: or ; where x is the original indicator value, are the minimum and maximum values of the indicator, μ and σ are the mean and standard deviation, is the standardized or normalized indicator value. After conversion, this value is used for subsequent modeling or analysis to ensure the comparability of different features.

[0056] 3. Weight Coefficient Setting Set the weight of each indicator according to the enterprise strategy or empirical data , , , for example: = 0.4 (importance of population coverage); = 0.4 (conversion effect); = 0.2 (budget efficiency).

[0057] 4. Calculation of Task Quality Score The quality score of each group of tasks Calculation formula: ; where are the three input indicators after standardization.

[0058] 5. Generate a Set of Task Quality Scores Finally, form: ; Each represents the comprehensive quality score of the k-th group of tasks.

[0059] By introducing a multi-dimensional quality assessment model, the system can comprehensively quantify the actual execution effect of each group of tasks, avoid the misleading caused by a single indicator (such as conversion rate), and ensure the fairness and scientificity of the evaluation results. The standardization and weight allocation mechanism enables the scoring to be dynamically adjusted according to the actual goals of the enterprise, meeting the strategic needs of different stages. The generated set of task quality scores provides an objective and quantitative decision-making basis for subsequent resource optimization allocation, improving the overall promotion efficiency and return on investment.

[0060] Based on the structured feature dataset, the status monitoring module obtains the task execution status log, including generating a task priority distribution map and a potential value heat map using visualization technology according to the set of task quality scores. The distribution map shows the proportion of the number of high, medium, and low-priority tasks, and the heat map is drawn based on the target population coverage rate and channel conversion rate to determine the task priority ranking and value distribution, obtaining a hierarchical representation result.

[0061] First, the system extracts the scores of each task group from the previously obtained task quality score set and classifies these tasks into three categories: high priority, medium priority, and low priority according to the generated priority labels. Then, the system counts the number of tasks belonging to each priority category and calculates the proportion of each category. For example, calculate the proportion of high-priority tasks in the total number of tasks, and the same calculation method is used for medium-priority and low-priority tasks. These proportions are used to generate a priority distribution diagram, usually in the form of a bar chart or a pie chart, clearly showing the proportion of tasks with different priorities in the overall task set.

[0062] Next, the system draws a potential value heat map based on the target population coverage rate and channel conversion rate of each task. In the chart, the horizontal axis represents the target population coverage rate, and the vertical axis represents the channel conversion rate. The system plots the performance of each task on these two dimensions as points, and at the same time, according to the quality score of the task, the depth or type of color is used to reflect the value of the task. The higher the score, the darker the color or the closer it is to the color of high value.

[0063] Finally, the system integrates the results of the priority distribution diagram and the potential value heat map to generate an overall hierarchical representation result. This result provides visual and quantitative data support for subsequent resource allocation, task optimization, and strategy adjustment, enabling the management team to more intuitively identify the market potential and execution priorities of various tasks.

[0064] By integrating visualization technology, the system converts the complex multi-dimensional task evaluation results into an intuitive and easy-to-understand graphical representation, helping the management team quickly identify high-value task groups and rationally allocate resources. The priority distribution diagram provides proportion information of the overall task level, helping to judge the rationality of the overall layout of promotion resources; the potential value heat map reflects the coverage and conversion potential of individual tasks, assisting in formulating personalized optimization strategies. This process significantly improves the decision-making efficiency and controllability of promotion activities.

[0065] The status monitoring module obtains the task execution status log according to the structured feature data set, including extracting the task priority ranking from the hierarchical representation result, combining with the potential value heat map, and optimizing and scheduling the tasks based on the task quality score and budget utilization rate through a sorting algorithm to generate a task execution sequence.

[0066] 1. Priority ranking extraction First, the system reads the priority labels of each task from the hierarchical representation result, including high priority, medium priority, and low priority. All tasks are initially sorted according to the priority level, so that high-priority tasks are ranked first, and medium-priority and low-priority tasks are arranged in sequence behind.

[0067] 2. Combine with the potential value heat map Based on the preliminary sorting, the system further refers to the potential value heat map. This heat map reflects the performance of tasks in two dimensions: the coverage rate of the target population and the channel conversion rate, providing visual and numerical information on the market potential of tasks. The system incorporates this information into subsequent sorting considerations to ensure that tasks with high market value are given more priority in the execution position.

[0068] 3. Application of the sorting algorithm The system performs comprehensive sorting on the preliminary sorting results and the potential value heat map data. The main basis for sorting is two indicators: the task quality score and the budget utilization rate.

[0069] First, the system assigns a higher sorting weight to tasks with a high quality score; in the case of similar or equal quality scores, the budget utilization rate is further compared. Tasks with a higher budget utilization rate are executed first to improve the efficiency of fund use; if the quality score and the budget utilization rate are both equal, the coverage rate and conversion rate in the potential value heat map are referred to, and the task with better performance is selected first.

[0070] 4. Generate the task execution sequence After sorting, the system forms a task execution sequence and arranges the tasks in the optimized order. This sequence ensures that tasks with high priority, high market value, and high budget efficiency are executed first, and subsequent tasks are arranged in turn, forming a reasonable and efficient execution strategy.

[0071] This embodiment realizes comprehensive and intelligent task scheduling by integrating three dimensions: priority, market potential (potential value), and resource utilization efficiency. The sorting algorithm avoids the one-sidedness brought by a single indicator, ensuring that the task execution order not only conforms to the business priority but also takes into account the return on investment and budget efficiency, significantly improving the execution effect of the overall promotion activity and the efficiency of fund use. In addition, the automated sorting and scheduling process reduces manual intervention and improves the response speed and flexibility of the system.

[0072] The status monitoring module obtains the task execution status log based on the structured feature data set. For the task execution sequence, real-time monitoring technology is used to collect real-time data of task execution once every minute, including click-through rate, conversion rate, and budget consumption rate. Traffic anomalies or sudden drops in conversion rate are identified through the anomaly detection mechanism to generate the task execution status log and obtain the task execution status log.

[0073] During the system operation, for each task executed in a predefined order, the status monitoring module automatically collects three pieces of real-time data at a one-minute interval, namely click-through rate, conversion rate, and budget consumption rate. The click-through rate is calculated by dividing the number of times an advertisement is clicked within a certain period by the number of times it is displayed, resulting in a percentage that measures the attractiveness of the advertisement. The conversion rate is calculated by dividing the number of users who achieve the expected conversion behavior by the number of users who click on the advertisement, obtaining the percentage of the promotion's effectiveness. The budget consumption rate is calculated by dividing the currently consumed budget amount by the total budget predefined for the task, getting the current budget usage ratio. Before the collected data enters the analysis process, it undergoes data cleaning, removing duplicate data, filling in missing values, and eliminating abnormal extreme values to ensure the accuracy of the analysis. Subsequently, the system performs anomaly detection on these three metrics for each task. The anomaly detection mechanism compares the currently collected data with the average level of the task's historical data or the set normal range. If it is found that the click-through rate or conversion rate fluctuates significantly in a short period, or the budget consumption rate shows an unexpected abnormal change, such as the conversion rate being significantly lower than the previous average in two consecutive time periods, the system will identify this as an abnormal situation. Once an anomaly is detected, the system immediately generates a task execution status log containing the time, task identifier, name of the abnormal metric, current metric value, and type of anomaly. This log will be continuously updated, serving as the basis for task adjustment, suspension, or optimization, and at the same time, it will be fed back to the scheduling module and management personnel to support timely response and reduce potential risks.

[0074] Through real-time monitoring and anomaly detection, the system can immediately grasp the actual performance of task execution, quickly respond when abnormal situations occur, and effectively avoid budget waste and performance decline. The automatically generated task execution status log provides detailed data support for subsequent data analysis, task adjustment, and system learning, enhancing the transparency and controllability of the overall promotion activity. In addition, the minute-level data collection frequency balances the response speed and system resource consumption, ensuring the efficiency and economy of monitoring.

[0075] The data cleaning techniques in the data processing module include duplicate record detection algorithms and missing value filling algorithms, and the missing value filling algorithm adopts the mean substitution method based on similar features.

[0076] During the data processing of the system, duplicate record detection is first performed on the structured feature dataset. The system combines and matches multiple key fields such as task name, target population identifier, channel code, budget information, etc. to identify records with exactly the same content or consistent in key fields in the data. When duplicate records are found, the system only retains one valid record and deletes other duplicates to avoid errors in subsequent analysis caused by data redundancy. After deduplication, the system scans all fields of the dataset to identify records with missing values and marks these missing data units. Next, for each missing data unit, based on the other feature values already existing in the record, the system searches for complete records with similar features in the entire dataset. The criterion for judging similarity can be based on the distance calculation between features, such as calculating the difference between feature values, or matching according to the similarity ratio of categorical features, ensuring that the selected complete records are highly similar to the missing records in key attributes such as target population category, budget range, and channel type. After finding the set of similar records, the system extracts the data corresponding to the missing fields in these records, adds up all the values of these data, and divides by the number of records to obtain the average value of this field. Subsequently, the system replaces the missing value with this average value, thus completing the data filling. This method can ensure that the filled value fully reflects the data distribution with similar attributes to the missing record, avoiding the bias that may be brought by simply using the average value of all data. After completing the filling of all missing values, the system performs consistency verification to check the data type, value range, and logical relationship of each field to ensure the integrity and correctness of the data, providing a high-quality data basis for subsequent data analysis and model training.

[0077] This data cleaning method significantly reduces data redundancy and improves data processing efficiency by automatically detecting and deleting duplicate records. Using the mean substitution method based on similar features for missing value filling can better maintain the representativeness and accuracy of the data compared to traditional global mean filling or simply deleting missing records, reducing analysis errors caused by missing value processing. This processing method improves the reliability of subsequent feature extraction, model training, and task evaluation, laying a solid data foundation for scientific decision-making in brand promotion tasks.

[0078] The standardization technique in the data processing module is the Z-score standardization method.

[0079] During the data processing of the system, after data cleaning is completed and the features to be processed are determined, the system will standardize the data of each feature. First, the system will add up the values of the feature in all records to obtain the total sum, and then divide this total sum by the number of records to calculate the average value of the feature. Next, the system will calculate the standard deviation. The specific method is as follows: for each data point, subtract the average value just calculated from the value of the data point to obtain the difference, and then square all the differences to obtain a set of squared differences. The system will add up all these squared differences, divide this sum by the number of records minus one, and finally perform a square root operation on this result to obtain the standard deviation of the feature. After calculating the average value and the standard deviation, the system will transform each piece of data: subtract the average value from the original value of the data point to obtain the deviation value, and then divide this deviation value by the standard deviation to finally obtain the standardized value. Through this method, the data of all features are transformed into a distribution with a mean of zero and a standard deviation of one, so that the scale differences between different features are eliminated, and the contributions of all features become fairly comparable. This standardization process is particularly suitable for subsequent data analysis and model training, and can prevent features with a large value range from occupying too high a weight in the algorithm, thereby improving the accuracy and stability of subsequent operations such as principal component analysis, clustering, and classification.

[0080] By adopting the Z-score standardization method, it is possible to effectively eliminate the scale differences between different features, make the contributions between features fairly comparable, and avoid large-value features from occupying too high a weight in the model. At the same time, this method has strong adaptability and is applicable to most machine learning and data analysis models. It has a positive effect on subsequent processes such as principal component analysis, feature weight update, clustering, and classification, improving the analysis accuracy and execution efficiency of the overall system. In addition, the Z-score standardization has relatively loose requirements for the data distribution and can maintain good standardization effects under various data feature distribution types.

[0081] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-dimensional index evaluation and automatic grading system for brand promotion tasks, characterized in that, Including: A data processing module, which is used to obtain task feature data from the promotion task database, including dynamic features such as target population division, placement channel selection, and budget scale allocation, remove duplicate records and missing values through data cleaning technology, and convert the feature data into a structured feature data set with a mean of 0 and a variance of 1 by using standardization technology, so as to obtain a structured feature data set; A status monitoring module, which is used to obtain a task execution status log according to the structured feature data set. If the click-through rate or conversion rate in the task execution status log is continuously lower than the preset threshold, the weight coefficients of target population preference and channel matching degree are adjusted based on historical data fusion through feature weight update technology, and an incremental update strategy is adopted to generate an optimized feature vector set, so as to obtain an optimized feature vector set; A task grouping and grading module, which is used to group tasks based on target population preference and channel matching degree by using repeated clustering analysis technology according to the optimized feature vector set, and re-grade them based on the adjusted weight coefficients by using decision tree classification technology, so as to generate an updated task grading result; A visualization module, which is used to generate a new task priority distribution map and a potential value heat map by using visualization technology for the updated task grading result. The distribution map reflects the adjusted priority ratio, and the heat map is drawn based on the updated target population coverage rate and channel conversion rate to determine the dynamically adjusted task priority and potential value, so as to obtain a dynamically adjusted representation result.

2. The multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 1, characterized in that: The status monitoring module obtains a task execution status log according to the structured feature data set, including extracting key information of target population preference, channel matching degree, and budget constraint by using principal component analysis technology for the structured feature data set, determining the weight coefficients of each feature dimension, and generating a feature vector set including target population preference, channel matching degree, and budget constraint, so as to obtain a feature vector set.

3. The multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 2, wherein: The status monitoring module obtains a task execution status log according to the structured feature data set, including if the weight coefficient in the feature vector set is higher than the preset threshold, then initially grouping tasks based on target population preference and channel matching degree by using clustering analysis technology, and generating a task grouping set based on multi-dimensional dynamic features, so as to obtain a task grouping set.

4. A multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 3, characterized in that: The status monitoring module obtains a task execution status log according to the structured feature data set, including obtaining the feature vectors of each group of tasks from the task grouping set, and automatically grading the tasks based on target population preference, channel matching degree, and budget constraint by using decision tree classification technology, and generating a task grading result including high, medium, and low priorities, so as to obtain a task grading result.

5. The multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 4, characterized in that: The status monitoring module obtains a task execution status log according to the structured feature data set, including calculating the quality scores of each group of tasks through a quality assessment model for the task grading result. The model input includes target population coverage rate, channel conversion rate, and budget utilization rate, and generating a task quality score set, so as to obtain a task quality score set.

6. The multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 5, characterized in that: The state monitoring module obtains the task execution status log according to the structured feature dataset, including generating a task priority distribution map and a potential value heat map by using visualization technology based on the task quality score set. The distribution map shows the proportion of the number of tasks with high, medium, and low priorities. The heat map is drawn based on the target population coverage rate and the channel conversion rate to determine the task priority ranking and value distribution, and obtain the hierarchical representation result.

7. The multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 6, characterized in that: The state monitoring module obtains the task execution status log according to the structured feature dataset, including extracting the task priority ranking from the hierarchical representation result, combining with the potential value heat map, and optimizing and scheduling the tasks based on the task quality score and budget utilization rate through a sorting algorithm to generate a task execution sequence and obtain the task execution sequence.

8. A multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 7, characterized in that: The state monitoring module obtains the task execution status log according to the structured feature dataset, including collecting the real-time data of task execution once per minute for the task execution sequence by using real-time monitoring technology, including click-through rate, conversion rate, and budget consumption rate, and identifying traffic anomalies or sudden drops in conversion rate through an anomaly detection mechanism to generate the task execution status log and obtain the task execution status log.

9. The multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 1, characterized in that: The data cleaning technology in the data processing module includes a duplicate record detection algorithm and a missing value filling algorithm. The missing value filling algorithm uses a mean substitution method based on similar features.

10. A multi-dimensional index evaluation and automatic grading system for brand promotion tasks according to claim 1, characterized in that: The standardization technology in the data processing module is the Z-score standardization method.

Images