Personalized operation strategy generation method and system based on data processing

Abstract

The invention discloses a personalized operation strategy generation method and system based on data processing, and the method comprises the steps: collecting and preprocessing multi-dimensional data, and storing the data through employing a distributed storage architecture; constructing a comprehensive user portrait from multiple angles according to the target multi-dimensional data, extracting static features and dynamic features, and screening the static features and the dynamic features to obtain target decision features; performing data analysis and mining on the target decision feature according to the user portrait, generating a data analysis result, and performing visual chart presentation; constructing a deep learning model, training by using historical data to obtain a target deep learning model, inputting a data analysis result into the target deep learning model, and outputting a prediction result; generating a personalized operation strategy according to the prediction result and the operation target, monitoring the operation strategy in real time, and adjusting the operation strategy in real time according to the monitoring result. According to the invention, intelligentization, precision and high efficiency of operation management are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to a method, system, terminal and computer-readable storage medium for generating a personalized operation strategy based on data processing. Background Art

[0002] With the rapid development of information technology, the amount of data in various industries has exploded. Enterprises have accumulated massive amounts of data during their operations, including business data, user data, market data, etc. However, traditional operating models find it difficult to effectively process and utilize this data, resulting in a lack of accuracy and timeliness in operational decisions and low operational efficiency.

[0003] Limitations of the operation model: The traditional scenario-based operation model of targeted user stratification makes it difficult to fully perceive the user strategy coverage and conversion flow, and cannot promptly discover blind spots not covered by the strategy, limiting the potential for business growth.

[0004] Algorithms are out of touch with operations: The R&D algorithm team and the strategic operations team often work independently. The optimization of the algorithm model lacks effective input of operational experience, resulting in slow algorithm iteration and possible unrobustness in actual applications. The operations team lacks automation and intelligent tools, relies too much on manual labor, and is unable to conduct in-depth mining and efficient operation of large amounts of user data.

[0005] Most existing operation management systems can only perform simple data statistics and analysis, and cannot deeply explore the complex relationships and potential value behind the data. Although deep learning algorithms have achieved remarkable results in the fields of image recognition and speech recognition, their application in the field of operation management is still in its infancy. In addition, the lack of an effective real-time feedback control mechanism makes it impossible for the operation system to adjust its strategy in time according to the actual operation situation, making it difficult to adapt to the dynamically changing market environment.

[0006] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0007] The main purpose of the present invention is to provide a method, system, terminal and computer-readable storage medium for generating personalized operation strategies based on data processing, aiming to solve the problem that the existing technology fails to effectively process and utilize the massive data of the enterprise, resulting in the lack of accuracy and timeliness of operation decisions and low operation efficiency.

[0008] To achieve the above object, the present invention provides a method for generating a personalized operation strategy based on data processing, and the method for generating a personalized operation strategy based on data processing comprises the following steps:

[0009] Collecting multi-dimensional data from a target business system or a target business platform, preprocessing the multi-dimensional data to obtain target multi-dimensional data, and using a distributed storage architecture to store the target multi-dimensional data;

[0010] Building a comprehensive user portrait from multiple perspectives based on the target multi-dimensional data, extracting static features and dynamic features from the target multi-dimensional data, and screening the static features and the dynamic features to obtain target decision features;

[0011] Performing data analysis and data mining on the target decision-making features according to the user portrait, generating data analysis results, and presenting the data analysis results in an intuitive visual chart;

[0012] Construct a deep learning model, train the deep learning model using historical data to obtain a target deep learning model, input the data analysis results into the target deep learning model, and output a prediction result;

[0013] According to the prediction results, combined with the operation objectives, a personalized operation strategy is generated based on strategy rules and constraints, the operation strategy is monitored in real time, and the operation strategy is adjusted in real time according to the monitoring results.

[0014] Optionally, the method for generating a personalized operation strategy based on data processing further comprises:

[0015] The coverage of user strategies and conversion flow paths are presented in a visual manner to form a user strategy map. The user strategy map is used to display the flow of different user groups under various operation strategies, as well as the connection between strategies, and graphically display the execution process and effects of user strategies.

[0016] Optionally, the method for generating a personalized operation strategy based on data processing further comprises:

[0017] Establish a data sharing platform to share the latest operational data. When problems arise during operations, different teams can use the data sharing platform to jointly analyze and solve the problems. Establish a problem feedback and tracking mechanism to ensure that problems are handled in a timely manner and convert solutions into measures for algorithm improvement and operational process optimization.

[0018] Optionally, the method for generating a personalized operation strategy based on data processing, wherein the multi-dimensional data is collected from a target business system or a target business platform, the multi-dimensional data is pre-processed to obtain target multi-dimensional data, and the target multi-dimensional data is stored in a distributed storage architecture, specifically comprising:

[0019] Collect multi-dimensional data from the target business system or target business platform through open platform interfaces, web crawlers or data sharing protocols, the multi-dimensional data including basic user information, transaction behavior, social interaction and market trends;

[0020] Preprocessing the multidimensional data to obtain target multidimensional data, wherein the preprocessing includes data cleaning and data conversion, wherein the data cleaning is used to clean the multidimensional data to remove duplicate, erroneous and incomplete data, and the data conversion is used to convert unstructured data in the multidimensional data into structured data;

[0021] A distributed storage architecture is used to store the target multi-dimensional data, and a data index and metadata management mechanism is established.

[0022] Optionally, the method for generating a personalized operation strategy based on data processing, wherein data analysis and data mining are performed on the target decision-making features according to the user portrait, data analysis results are generated, and the data analysis results are presented in an intuitive visual chart, specifically including:

[0023] Using a data analysis method to perform data analysis on the target decision-making features according to the user portrait, the data analysis method includes descriptive statistical analysis and exploratory data analysis to obtain user behavior and business operation status;

[0024] Using a deep data mining method to perform data mining on the target decision features according to the user portrait, the data mining includes association rule mining and sequence pattern mining to obtain data regularities;

[0025] Generate data analysis results based on user behavior, business operation status and data patterns, present the data analysis results in intuitive visual charts, and generate regular data analysis reports, which include user behavior reports and operation performance reports.

[0026] Optionally, the method for generating a personalized operation strategy based on data processing, wherein the deep learning model is constructed, the deep learning model is trained using historical data to obtain a target deep learning model, the data analysis result is input into the target deep learning model, and a prediction result is output, specifically comprising:

[0027] Building a deep learning model according to operational needs, the deep learning model includes a recurrent neural network for user behavior prediction, a long short-term memory network, and a convolutional neural network for user classification;

[0028] Customize the input layer, hidden layer and output layer structure of the deep learning model for different business scenarios;

[0029] Using large-scale historical data to train the deep learning model, using stochastic gradient descent or Adam optimization algorithm to adjust model parameters, so that the model converges to the optimal solution;

[0030] The model performance is evaluated and optimized through cross-validation and model evaluation indicators to prevent the model from overfitting or underfitting, and the target deep learning model is obtained. The data analysis results are input into the target deep learning model to output the prediction results.

[0031] Optionally, the method for generating a personalized operation strategy based on data processing, wherein generating a personalized operation strategy based on the prediction result, in combination with the operation goal, based on the strategy rules and constraints, monitoring the operation strategy in real time, and adjusting the operation strategy in real time according to the monitoring result, specifically includes:

[0032] Based on the prediction results and big data analysis insights of the target deep learning model, combined with the operation goals, the strategy rules and constraints are integrated into the strategy generation process to generate a personalized operation strategy;

[0033] Key operating indicators are monitored in real time. When the monitored indicators deviate from the expected targets, the operating strategies are adjusted and the target deep learning is optimized.

[0034] In addition, to achieve the above-mentioned purpose, the present invention further provides a personalized operation strategy generation system based on data processing, wherein the personalized operation strategy generation system based on data processing includes:

[0035] A data collection and processing module is used to collect multi-dimensional data from a target business system or a target business platform, pre-process the multi-dimensional data to obtain target multi-dimensional data, and store the target multi-dimensional data using a distributed storage architecture;

[0036] A user portrait and feature extraction module, used to construct a comprehensive user portrait from multiple angles based on the target multi-dimensional data, extract static features and dynamic features from the target multi-dimensional data, and screen the static features and the dynamic features to obtain target decision features;

[0037] A data analysis and visualization module, used to perform data analysis and data mining on the target decision-making features according to the user portrait, generate data analysis results, and present the data analysis results in an intuitive visualization chart;

[0038] A model building and prediction module is used to build a deep learning model, train the deep learning model using historical data to obtain a target deep learning model, input the data analysis results into the target deep learning model, and output a prediction result;

[0039] The operation strategy generation and adjustment module is used to generate a personalized operation strategy based on the prediction results, operation objectives, strategy rules and constraints, monitor the operation strategy in real time, and adjust the operation strategy in real time according to the monitoring results.

[0040] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a personalized operation strategy generation program based on data processing stored in the memory and executable on the processor, wherein the personalized operation strategy generation program based on data processing implements the steps of the personalized operation strategy generation method based on data processing as described above when the personalized operation strategy generation program based on data processing is executed by the processor.

[0041] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a personalized operation strategy generation program based on data processing, and when the personalized operation strategy generation program based on data processing is executed by a processor, the steps of the personalized operation strategy generation method based on data processing as described above are implemented.

[0042] In the present invention, multi-dimensional data is collected from a target business system or a target business platform, the multi-dimensional data is preprocessed to obtain target multi-dimensional data, and a distributed storage architecture is used to store the target multi-dimensional data; a comprehensive user portrait is constructed from multiple angles based on the target multi-dimensional data, static features and dynamic features are extracted from the target multi-dimensional data, and the static features and the dynamic features are screened to obtain target decision features; data analysis and data mining are performed on the target decision features based on the user portrait to generate data analysis results, and the data analysis results are presented in intuitive visual charts; a deep learning model is constructed, the deep learning model is trained using historical data to obtain a target deep learning model, the data analysis results are input into the target deep learning model, and a prediction result is output; based on the prediction result and in combination with the operation goal, a personalized operation strategy is generated based on policy rules and constraints, the operation strategy is monitored in real time, and the operation strategy is adjusted in real time according to the monitoring result. The present invention generates personalized operation strategies by integrating big data analysis, deep learning algorithms and real-time feedback control mechanisms. Combined with user strategy maps and team collaboration modules, it can effectively improve the operational efficiency and accuracy of enterprises, provide strong support for enterprises in the fierce market competition, and realize intelligent, precise and efficient operation management. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of a preferred embodiment of the method for generating personalized operation strategies based on data processing of the present invention;

[0044] Figure 2 It is a structural diagram of a preferred embodiment of the personalized operation strategy generation system based on data processing of the present invention;

[0045] Figure 3 It is a structural diagram of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0047] The method for generating a personalized operation strategy based on data processing described in the preferred embodiment of the present invention is as follows: Figure 1 As shown, the method for generating a personalized operation strategy based on data processing includes the following steps:

[0048] Step S10: collect multi-dimensional data from a target business system or a target business platform, pre-process the multi-dimensional data to obtain target multi-dimensional data, and use a distributed storage architecture to store the target multi-dimensional data.

[0049] Specifically, multi-dimensional data is collected from target business systems (such as user transaction systems, membership management systems, marketing platforms, etc. within the enterprise) or target business platforms (such as social media, industry data providers) through open platform interfaces, web crawlers or data sharing agreements, and data connections are established with various business systems within the enterprise. For example, connection to user databases, transaction databases, etc. is made through database connection tools, and a data collection schedule is set to ensure timely acquisition of data; for external data, such as collecting user comment data from social media platforms, web crawler technology is used, and the platform's data usage rules are followed to regularly crawl relevant data; the multi-dimensional data includes data such as basic user information, transaction behavior, social interaction and market trends.

[0050] The multi-dimensional data is preprocessed to obtain target multi-dimensional data. The preprocessing includes data cleaning and data conversion. The data cleaning is used to clean the multi-dimensional data and remove duplicate, erroneous and incomplete data. For example, by setting data format specifications and logical verification rules, data records that do not meet the requirements are eliminated. In the data cleaning process, a data cleaning script is written to identify and process abnormal values ​​in the data (such as transaction data that exceeds the normal price range), and the rationality of the data is determined by combining business rules and data statistical features; the data conversion is used to convert unstructured data in the multi-dimensional data into structured data, and convert unstructured data (such as text comments, images) into structured data to facilitate subsequent analysis, such as using natural language processing technology to extract keywords and emotional tendencies in texts, and extract features from images, such as using a natural language processing library to perform emotional analysis on user comments, and converting comments into positive, negative, and neutral emotional label data; using a distributed storage architecture (such as Hadoop HDFS, NoSQL database, etc.) to store the target multi-dimensional data to ensure the scalability and high availability of data; and establishing a data index and metadata management mechanism to facilitate rapid query and retrieval of data.

[0051] Step S20: construct a comprehensive user portrait from multiple angles based on the target multi-dimensional data, extract static features and dynamic features from the target multi-dimensional data, and screen the static features and the dynamic features to obtain target decision features.

[0052] Specifically, based on the integrated data (i.e., the target multi-dimensional data), a comprehensive user portrait is constructed from multiple aspects such as the user's demographic characteristics, behavioral preferences, consumption habits, and psychological characteristics. For example, by analyzing the user's purchase history and browsing behavior, a user's consumption interest map is depicted. Data mining techniques such as cluster analysis and association rule mining are used to classify and label users to form different user group portraits. In the construction of user portraits, the user's consumption preference score is calculated through a comprehensive analysis of the user's transaction data, browsing behavior data, etc. For example, the user's main consumption category is determined by calculating the user's purchase frequency and amount of different categories of goods. Clustering algorithms are used to classify users. For example, the K-Means algorithm is used to divide users into different groups such as high-consumption frequent users and low-consumption occasional users, and each group is assigned a corresponding portrait label.

[0053] Extract valuable features from user data, including static features (such as age, gender) and dynamic features (such as recent purchase frequency, browsing time). Use feature selection algorithms (such as information gain, chi-square test, etc.) to select the features that have the greatest impact on operational decisions from the static features and dynamic features, reduce data dimensions, and improve the operational efficiency of subsequent models. In the feature extraction process, analyze the user's behavioral time series data to extract time interval features (such as the time interval between two purchases by a user) and frequency features (such as the number of times a user browses per week). Use feature selection algorithms to screen the extracted features, and by calculating the information gain of the features, select the most valuable features for user classification and behavior prediction, remove redundant features, and improve data processing efficiency.

[0054] Step S30: perform data analysis and data mining on the target decision features according to the user portrait, generate data analysis results, and present the data analysis results in an intuitive visual chart.

[0055] Specifically, a variety of data analysis methods are used to analyze the target decision-making characteristics according to the user portrait, and the data analysis methods include descriptive statistical analysis and exploratory data analysis to obtain user behavior and business operation status, and fully understand user behavior and business operation status. For example, by statistically analyzing the purchase amount distribution of different user groups, the characteristics of high-value user groups are discovered. Descriptive statistical analysis methods are used to perform statistics on user data, such as calculating the mean and standard deviation of user age, the median of purchase amount, etc., to understand the basic characteristics and behavior distribution of users.

[0056] Use deep data mining methods to conduct deep data mining on the target decision-making features based on the user portrait. The data mining includes association rule mining (discovering the associated purchase relationship between commodities) and sequence pattern mining (analyzing the sequence of user behavior) to obtain data rules (mining the hidden rules behind the data). Use association rule mining algorithms to mine the association relationship between commodities, such as using the Apriori algorithm to find out the combination of commodities that are often purchased together, to provide a basis for commodity recommendation strategies.

[0057] Generate data analysis results based on user behavior, business operation status and data patterns, and present the data analysis results in intuitive visual charts (such as bar charts, line charts, Sankey diagrams, etc.) to facilitate the operation team and decision-makers to quickly understand the meaning of the data. In terms of data visualization, use data visualization software to draw user behavior path maps to show the user's browsing, clicking, purchasing and other behavior trajectories on the website or APP, helping the operation team understand the user's operating habits.

[0058] Generate regular data analysis reports, including user behavior reports and operational performance reports, etc. The analysis results are presented in the form of tables, charts and text descriptions through automated report generation tools. The report content includes user growth trends, user conversion rates of various channels, etc., to provide data support for operational decisions.

[0059] Step S40: construct a deep learning model, use historical data to train the deep learning model to obtain a target deep learning model, input the data analysis results into the target deep learning model, and output a prediction result.

[0060] Specifically, a deep learning model is constructed according to operational needs, and the deep learning model includes a recurrent neural network (RNN) for user behavior prediction, a long short-term memory network (LSTM) and a convolutional neural network (CNN) for user classification; the input layer, hidden layer and output layer structure of the deep learning model are customized for different business scenarios (such as user churn prediction, purchase intention prediction); the deep learning model is trained with large-scale historical data, and the model parameters are adjusted using stochastic gradient descent or Adam optimization algorithm to make the model converge to the optimal solution; the model performance is evaluated and optimized through cross-validation and model evaluation indicators (such as accuracy, recall rate, F1-score, etc.) to prevent the model from overfitting or underfitting, and a target deep learning model is obtained, and the data analysis results are input into the target deep learning model to output the prediction results.

[0061] Design the deep learning model architecture according to business needs. For example, for the user churn prediction problem, build a neural network model based on LSTM, take the user's historical behavior data as input, and predict whether the user will churn in the future. Divide the training set, validation set, and test set, generally in a ratio of 7:2:1. During the training process, use the stochastic gradient descent algorithm to optimize the model parameters, and adjust the model's hyperparameters (such as learning rate, number of hidden layer nodes, etc.) by evaluating the model's accuracy, recall rate, and other indicators on the validation set. Evaluate the trained model and calculate the model's performance indicators, such as F1-score, on the test set. If the model performance does not meet expectations, analyze the reasons and adjust the model architecture or reselect features and retrain.

[0062] Step S50: Generate a personalized operation strategy based on the prediction result, the operation goal, and the policy rules and constraints, monitor the operation strategy in real time, and adjust the operation strategy in real time according to the monitoring result.

[0063] Specifically, according to the prediction results of the target deep learning model and the insights from big data analysis, combined with operational goals (such as user growth, sales increase), the policy rules and constraints are integrated into the policy generation process to generate personalized operational strategies, for example, accurate retention strategies are formulated for users at risk of churn, and incentive consumption strategies are formulated for high-potential users; the strategic operations team converts operational experience into policy rules and constraints, and integrates them into the policy generation process to ensure the feasibility and effectiveness of the strategy. Based on the prediction results of the deep learning model and the insights from big data analysis, operational strategies are generated in combination with operational goals. For example, if the model predicts that a user has a high purchase intention, a personalized recommendation strategy is generated for the user to recommend related products or promotions.

[0064] Establish a real-time data monitoring system to monitor key operational indicators (such as user conversion rate and strategy execution effect) in real time. When the monitoring indicators deviate from the expected targets, timely feedback is provided to trigger strategy adjustments and model optimization. For example, if the actual conversion rate of a certain user group is lower than the predicted value, immediately analyze the reasons and adjust the operational strategy. Establish a real-time data monitoring system to obtain key operational indicator data in real time by embedding data monitoring points in the business system, such as the user's real-time purchase conversion rate and the strategy's click-through rate. When it is monitored that the operational indicators deviate from the expected targets, such as the user conversion rate is lower than the preset threshold, analyze the reasons and adjust the operational strategy. If it is because the recommended products do not meet the user's preferences, reuse the user portrait and deep learning model to adjust the product recommendation strategy.

[0065] Furthermore, the coverage of user strategies and the conversion flow paths are presented in a visual way to form a user strategy map. The user strategy map is used to display the flow of different user groups under various operation strategies, as well as the connection relationship between strategies. The execution process and effect of user strategies are graphically displayed using graph theory algorithms and visualization technology to help the operation team quickly locate problem areas and blind spots not covered by strategies. The user strategy map is constructed through graph databases and visualization technology to graphically display the flow process of users under different strategies. For example, nodes are used to represent different user groups, edges are used to represent the transfer paths of users under the action of strategies, and colors and widths are used to represent the size and transfer probability of user traffic.

[0066] Visually monitor each module and key link of the operation engine, including data collection progress, model training status, strategy execution effect, etc. Provide an interactive monitoring interface for the operation team and decision-makers, support real-time query, drill-down analysis and decision adjustment, so as to quickly respond to changes and problems in the operation process. In terms of visual monitoring, develop an interactive monitoring interface to achieve real-time display and interactive operation of data through front-end technology. The operation team can view the operation status of each module in real time on the interface, such as the progress of data collection, changes in the loss value of model training, etc., and can perform drill-down analysis to gain a deeper understanding of specific problems.

[0067] Furthermore, a data sharing platform and mechanism should be established to share the latest operational data on the data sharing platform, so that the production and research algorithm team can obtain the latest operational data, and the strategic operation team can obtain the prediction results and analysis reports of the algorithm model, and encourage the strategic operation team to feedback the operational experience and business rules to the production and research algorithm team in the form of data features and policy constraints to promote the optimization and innovation of the algorithm model. A data sharing platform should be established, and data warehouse technology and data interface specifications should be adopted to ensure that the production and research algorithm team and the strategic operation team can easily share data. For example, the strategic operation team uploads the latest user feedback data to the sharing platform for the production and research algorithm team to use for algorithm optimization.

[0068] When problems arise during the operation process (such as algorithm prediction deviations, poor strategy execution), the data sharing platform is used by different teams to conduct joint analysis and problem solving (for example, the collaborative work platform is used to organize the production and research algorithm team and the strategy operation team to conduct joint analysis and problem solving), and a problem feedback and tracking mechanism is established to ensure that problems are handled in a timely manner and that solutions are converted into measures for algorithm improvement and operation process optimization. When operational problems arise, relevant teams are organized through the collaborative work platform to hold online meetings and problem discussions. For example, when the user churn rate predicted by the algorithm does not match the actual situation, the production and research algorithm team and the strategy operation team jointly analyze the data to identify possible causes of the deviation, such as changes in data features or adjustments to the operation strategy, and develop solutions.

[0069] In summary, the fully intelligent operation engine of the present invention can effectively improve the operational efficiency and accuracy of the enterprise by integrating big data analysis, deep learning algorithms and real-time feedback control mechanisms, combined with user strategy maps and team collaboration modules, and provide strong support for enterprises in the fierce market competition.

[0070] The present invention aims to build a fully intelligent operation engine, break the limitations of traditional operation models, strengthen the collaboration between the production and research algorithm team and the strategy operation team, and make full use of big data analysis and deep learning algorithms to achieve real-time feedback control, so as to improve the efficiency and accuracy of enterprise operations and tap business growth opportunities. Transform from a scenario-based operation model of targeted user stratification to a new operation model of user strategy map, through the strategy map, clearly perceive the user strategy coverage and conversion flow, quickly locate problems, help the algorithm find more blind spots not covered by the strategy, and find growth breakthroughs; production and research algorithm team: focus on model tuning, platform automation engineering, and can actively discover problems on the link for early warning or intelligent decision-making flow; strategy operation team: focus on user portrait construction, feature precipitation, data analysis and monitoring, and precipitate operational experience into the algorithm feature model, effectively improve the algorithm iteration speed and robustness, solve things that cannot be operated or paid attention to manually, and thus achieve the possibility of achieving marketing goals.

[0071] The innovative features and beneficial effects of the present invention are as follows:

[0072] (1) The present invention collects data by means of web crawlers, open platform interfaces, etc., processes unstructured data in combination with natural language processing, image feature extraction, and other technologies, and uses a distributed storage system (such as Hadoop HDFS, NoSQL database) for data storage. It solves the problem of a single data source and inability to process unstructured data in the traditional operating model, makes full use of the massive data resources in the big data era, and provides a data foundation for comprehensive insight into users and business operations. It can integrate diverse data from multiple business systems within the enterprise and external platforms, including structured and unstructured data. Through advanced data cleaning and conversion technologies, these data are converted into a format that can be used for analysis and stored in a distributed architecture to ensure the scalability and high availability of data.

[0073] (2) Based on data preprocessing, the present invention uses data analysis algorithms to mine data patterns, designs and trains deep learning models according to different business scenarios, and trains and optimizes the models through optimization algorithms (such as stochastic gradient descent, Adam, etc.). It overcomes the defect that the traditional operating model relies only on simple data analysis methods and cannot accurately predict complex user behaviors, and achieves accurate insights into user needs and behaviors. Combine big data analysis methods (such as association rule mining, sequence pattern mining, etc.) with deep learning algorithms (such as RNN, LSTM, CNN, etc.) to conduct comprehensive and in-depth analysis and prediction of user data. Big data analysis can mine hidden patterns in data, and deep learning algorithms can accurately predict user behavior.

[0074] (3) The present invention embeds data monitoring points in the business system, acquires data in real time, determines whether the indicators deviate by setting thresholds and comparison algorithms, and triggers strategy adjustment and model optimization mechanisms. This changes the situation of lagging strategy adjustments in traditional operating models, realizes dynamic optimization of operating strategies, and ensures that strategies can always adapt to changes in the market and users. A real-time data monitoring system is established to track key operating indicators (such as user conversion rate, strategy execution effect, etc.) in real time. Once the monitoring indicators deviate from the expected targets, they can be quickly fed back to the strategy generation module and the deep learning model module to adjust the operating strategy and optimize the model in a timely manner.

[0075] (4) The present invention combines the prediction results of the deep learning model with business rules, uses algorithms to generate personalized strategies, and introduces operational experience as a constraint in the strategy generation process. This solves the problem of inaccurate strategy coverage in the traditional targeted user stratification operation model, and can formulate appropriate operation strategies based on the characteristics of each individual user, thereby improving marketing effectiveness and user experience. Based on the prediction results of the deep learning model and the insights from big data analysis, personalized operation strategies are generated in combination with operational goals. At the same time, the strategic operation team can convert operational experience into strategy rules and constraints, and integrate them into the strategy generation process to ensure the feasibility and effectiveness of the strategy.

[0076] (5) The present invention uses graph theory algorithms and visualization technology to construct a user strategy map, and implements a visual monitoring interface for each operation link through front-end development technology. It makes up for the lack of intuitive visualization tools in the traditional operation model, helps the operation team quickly locate problem areas and blind spots not covered by the strategy, and facilitates decision-making and management. A user strategy map is constructed to present the coverage of user strategies and conversion flow paths in a visual manner. At the same time, each module and key link of the operation engine is visually monitored, providing an interactive monitoring interface for the operation team and decision-makers, supporting real-time query, drill-down analysis and decision adjustment.

[0077] (6) The present invention builds a data sharing platform, adopts data warehouse technology and data interface specifications to realize data sharing, and uses a collaborative work platform to organize teams to communicate and solve problems online. It solves the problem of the disconnection between algorithms and operations in the traditional operation model, so that the algorithm model can fully absorb the operation experience for optimization, and at the same time the operation strategy can get strong support from the algorithm model. A collaborative work mechanism between the production and research algorithm team and the strategy operation team is established, including data and experience sharing, collaborative problem solving, etc. Through the sharing platform and the collaborative work platform, close cooperation between the two teams is promoted, and operational efficiency and algorithm iteration speed are improved.

[0078] Furthermore, variations of the present invention may include:

[0079] The present invention uses federated learning technology to process data. Under this scheme, data can be trained on local devices or data sources, and only model parameters are exchanged without exchanging original data. In this way, collaborative analysis of decentralized data can be achieved while protecting data privacy. Different data sources (such as various departments of an enterprise or different partner companies) train models locally, and then encrypt and upload the model parameters to the central server for aggregation, and then return the aggregated parameters to the local server for updating, and iterate repeatedly until the model converges. Advantages and application scenarios: Suitable for industries with extremely high requirements for data privacy, such as finance and medical care. For example, customer credit risk assessment models are shared between banks without disclosing detailed data of each customer.

[0080] The present invention applies reinforcement learning algorithms to optimize strategy generation and real-time feedback control mechanisms. Reinforcement learning learns the optimal strategy through continuous trial and error by the intelligent agent in the environment, which can better adapt to dynamically changing user behaviors and market environments. The operating environment is regarded as the environment of reinforcement learning, and the operating strategy is regarded as the action of the intelligent agent. By defining reward functions (such as improved user conversion rate, increased profits, etc.), the intelligent agent can learn the optimal operating strategy through continuous strategy implementation and feedback. Advantages and application scenarios: In scenarios where the market environment changes rapidly, such as emerging businesses in the Internet industry, reinforcement learning can quickly adjust operating strategies to obtain the best results.

[0081] The present invention uses causal inference technology to formulate operational strategies. Causal inference can help determine the causal relationship between different factors (such as user characteristics, operational strategies), so as to formulate more effective strategies. By constructing causal models, such as structural equation models or potential outcome models, the causal relationship between user behavior and operational strategies is analyzed. For example, determine the causal impact of a certain promotional activity (strategy) on user purchasing behavior (results), rather than just based on correlation. Advantages and application scenarios: In scenarios where an in-depth understanding of the causal mechanism of user behavior is required, such as the formulation of marketing strategies for high-end products, causal inference can be used to ensure the effectiveness of the strategy.

[0082] Competitors may design an operation engine based on simple rules without using deep learning and complex data analysis methods. For example, several fixed user levels are set according to the user's historical purchase amount and frequency, and fixed operation strategies are adopted for different levels. User stratification and strategy formulation are carried out by writing some simple if-then rules. For example, if the user's purchase amount > X and the purchase frequency > Y, the user is classified as a high-value user and given a fixed preferential strategy. Limitations and countermeasures: Although this solution is simple and easy to implement, it lacks deep insight into user behavior and flexibility of strategy. The present invention can respond by emphasizing its own advantages in accurate prediction of complex user behavior and dynamic strategy adjustment, such as demonstrating significant results in personalized marketing and improving user experience.

[0083] Only select some key data for analysis instead of integrating multi-source data. For example, only focus on the user's transaction data, ignore the user's social data and browsing behavior data, and use simple statistical analysis methods to formulate operational strategies. Perform basic statistical analysis on the selected local data, such as calculating the mean, standard deviation, etc., and then formulate operational strategies based on these statistical results. Limitations and countermeasures: This solution cannot fully understand the user and is prone to strategic errors. The present invention can promote itself as being able to tap into more potential user needs and market opportunities through multi-source data integration and in-depth analysis, thereby gaining a wider market share.

[0084] In the process of strategy generation and regulation, judgment and adjustment are mainly based on manual experience, and only simple data analysis tools are used in some links. For example, operators judge when to adjust marketing strategies based on their own experience, and only use some statistical software when analyzing basic user attributes. Operators regularly check operational data reports and decide whether to adjust strategies based on their own experience and intuition. When data analysis is required, they use some ready-made business analysis software with relatively simple functions. Limitations and countermeasures: This solution is limited by the limitations of manual experience and analysis tools, is inefficient and prone to human errors. The present invention can highlight its own advantages of real-time feedback regulation and automated strategy generation, and demonstrate through actual cases the lag and inaccuracy of semi-automated operations that combine manual and intelligent operations in response to complex market changes.

[0085] In terms of collaboration between the production and research algorithm team and the strategic operations team, only the traditional communication and collaboration model is optimized, such as increasing the frequency of regular meetings, improving communication processes, etc., without involving the deep integration of data and algorithms. By formulating more detailed team collaboration specifications, communication and information sharing between teams are strengthened, but the essence of their independent work is not changed. Limitations and countermeasures: This solution cannot achieve the coordinated optimization of data and algorithms, and cannot give full play to the combined strength of the team. The present invention can emphasize that it has achieved deep integration and efficient collaboration between teams by establishing a collaborative platform and data sharing mechanism, and can quickly iterate operational strategies and algorithm models to gain an advantage in market competition.

[0086] Furthermore, if Figure 2 As shown, based on the above-mentioned personalized operation strategy generation method based on data processing, the present invention also provides a personalized operation strategy generation system based on data processing, wherein the personalized operation strategy generation system based on data processing includes:

[0087] The data collection and processing module 51 is used to collect multi-dimensional data from a target business system or a target business platform, pre-process the multi-dimensional data to obtain target multi-dimensional data, and store the target multi-dimensional data using a distributed storage architecture;

[0088] A user portrait and feature extraction module 52, configured to construct a comprehensive user portrait from multiple perspectives based on the target multi-dimensional data, extract static features and dynamic features from the target multi-dimensional data, and screen the static features and the dynamic features to obtain target decision features;

[0089] A data analysis and visualization module 53 is used to perform data analysis and data mining on the target decision-making features according to the user portrait, generate data analysis results, and present the data analysis results in an intuitive visualization chart;

[0090] A model building and prediction module 54 is used to build a deep learning model, train the deep learning model using historical data to obtain a target deep learning model, input the data analysis results into the target deep learning model, and output a prediction result;

[0091] The operation strategy generation and adjustment module 55 is used to generate a personalized operation strategy based on the prediction results, operation objectives, strategy rules and constraints, monitor the operation strategy in real time, and adjust the operation strategy in real time according to the monitoring results.

[0092] Furthermore, if Figure 3 As shown, based on the above-mentioned method and system for generating personalized operation strategies based on data processing, the present invention also provides a terminal accordingly, and the terminal includes a processor 10, a memory 20 and a display 30. Figure 3 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0093] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a personalized operation strategy generation program 40 based on data processing is stored on the memory 20, and the personalized operation strategy generation program 40 based on data processing can be executed by the processor 10, thereby realizing the personalized operation strategy generation method based on data processing in the present application.

[0094] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the personalized operation strategy generation method based on data processing.

[0095] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. The processor 10, the memory 20, and the display 30 of the terminal communicate with each other via a system bus.

[0096] In one embodiment, when the processor 10 executes the personalized operation strategy generation program 40 based on data processing in the memory 20, the steps of the personalized operation strategy generation method based on data processing are implemented.

[0097] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a personalized operation strategy generation program based on data processing, and when the personalized operation strategy generation program based on data processing is executed by a processor, the steps of the personalized operation strategy generation method based on data processing as described above are implemented.

[0098] In summary, the present invention provides a method, system, terminal and computer-readable storage medium for generating personalized operation strategies based on data processing, the method comprising: collecting multi-dimensional data from a target business system or a target business platform, preprocessing the multi-dimensional data to obtain target multi-dimensional data, and using a distributed storage architecture to store the target multi-dimensional data; constructing a comprehensive user portrait from multiple angles based on the target multi-dimensional data, extracting static features and dynamic features from the target multi-dimensional data, and screening the static features and the dynamic features to obtain target decision features; performing data analysis and data mining on the target decision features based on the user portrait, generating data analysis results, and presenting the data analysis results in intuitive visual charts; constructing a deep learning model, training the deep learning model with historical data to obtain a target deep learning model, inputting the data analysis results into the target deep learning model, and outputting prediction results; generating a personalized operation strategy based on the prediction results, in combination with the operation goals, based on policy rules and constraints, monitoring the operation strategy in real time, and adjusting the operation strategy in real time according to the monitoring results. The present invention generates personalized operation strategies by integrating big data analysis, deep learning algorithms and real-time feedback control mechanisms. Combined with user strategy maps and team collaboration modules, it can effectively improve the operational efficiency and accuracy of enterprises, provide strong support for enterprises in the fierce market competition, and realize intelligent, precise and efficient operation management.

[0099] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or terminal including the element.

[0100] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the program can be stored in a computer-readable storage medium that can be read by a computer, and the program can include the processes of the above-mentioned method embodiments when executed. The computer-readable storage medium can be a memory, a disk, an optical disk, etc.

[0101] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for generating a personalized operation strategy based on data processing, characterized in that: The method for generating a personalized operation strategy based on data processing includes: Collecting multi-dimensional data from a target business system or a target business platform, preprocessing the multi-dimensional data to obtain target multi-dimensional data, and using a distributed storage architecture to store the target multi-dimensional data; Building a comprehensive user portrait from multiple perspectives based on the target multi-dimensional data, extracting static features and dynamic features from the target multi-dimensional data, and screening the static features and the dynamic features to obtain target decision features; Performing data analysis and data mining on the target decision-making features according to the user portrait, generating data analysis results, and presenting the data analysis results in an intuitive visual chart; Construct a deep learning model, train the deep learning model using historical data to obtain a target deep learning model, input the data analysis results into the target deep learning model, and output a prediction result; According to the prediction results, combined with the operation objectives, a personalized operation strategy is generated based on strategy rules and constraints, the operation strategy is monitored in real time, and the operation strategy is adjusted in real time according to the monitoring results.

2. The method for generating a personalized operation strategy based on data processing according to claim 1, characterized in that: The method for generating a personalized operation strategy based on data processing also includes: The coverage of user strategies and conversion flow paths are presented in a visual manner to form a user strategy map. The user strategy map is used to display the flow of different user groups under various operation strategies, as well as the connection between strategies, and graphically display the execution process and effects of user strategies.

3. The method for generating a personalized operation strategy based on data processing according to claim 1, characterized in that: The method for generating a personalized operation strategy based on data processing also includes: Establish a data sharing platform to share the latest operational data. When problems arise during operations, different teams can use the data sharing platform to jointly analyze and solve the problems. Establish a problem feedback and tracking mechanism to ensure that problems are handled in a timely manner and convert solutions into measures for algorithm improvement and operational process optimization.

4. The method for generating a personalized operation strategy based on data processing according to claim 1, characterized in that: The collecting of multi-dimensional data from a target business system or a target business platform, preprocessing of the multi-dimensional data to obtain target multi-dimensional data, and storing the target multi-dimensional data using a distributed storage architecture specifically includes: Collect multi-dimensional data from the target business system or target business platform through open platform interfaces, web crawlers or data sharing protocols, the multi-dimensional data including basic user information, transaction behavior, social interaction and market trends; Preprocessing the multidimensional data to obtain target multidimensional data, wherein the preprocessing includes data cleaning and data conversion, wherein the data cleaning is used to clean the multidimensional data to remove duplicate, erroneous and incomplete data, and the data conversion is used to convert unstructured data in the multidimensional data into structured data; A distributed storage architecture is used to store the target multi-dimensional data, and a data index and metadata management mechanism is established.

5. The method for generating a personalized operation strategy based on data processing according to claim 1, characterized in that: The step of performing data analysis and data mining on the target decision-making features according to the user portrait, generating data analysis results, and presenting the data analysis results in an intuitive visual chart specifically includes: Using a data analysis method to perform data analysis on the target decision-making features according to the user portrait, the data analysis method includes descriptive statistical analysis and exploratory data analysis to obtain user behavior and business operation status; Using a deep data mining method to perform data mining on the target decision features according to the user portrait, the data mining includes association rule mining and sequence pattern mining to obtain data regularities; Generate data analysis results based on user behavior, business operation status and data patterns, present the data analysis results in intuitive visual charts, and generate regular data analysis reports, which include user behavior reports and operation performance reports.

6. The method for generating a personalized operation strategy based on data processing according to claim 1, characterized in that: The constructing of the deep learning model, training the deep learning model with historical data to obtain a target deep learning model, inputting the data analysis result into the target deep learning model, and outputting a prediction result specifically includes: Building a deep learning model based on operational needs, the deep learning model includes a recurrent neural network for user behavior prediction, a long short-term memory network, and a convolutional neural network for user classification; Customize the input layer, hidden layer and output layer structure of the deep learning model for different business scenarios; Using large-scale historical data to train the deep learning model, using stochastic gradient descent or Adam optimization algorithm to adjust model parameters, so that the model converges to the optimal solution; The model performance is evaluated and optimized through cross-validation and model evaluation indicators to prevent the model from overfitting or underfitting, and the target deep learning model is obtained. The data analysis results are input into the target deep learning model to output the prediction results.

7. The method for generating a personalized operation strategy based on data processing according to claim 1, characterized in that: The generating of a personalized operation strategy based on the prediction result, the operation goal, the strategy rules and the constraint conditions, the real-time monitoring of the operation strategy, and the real-time adjustment of the operation strategy according to the monitoring result specifically include: Based on the prediction results and big data analysis insights of the target deep learning model, combined with the operation goals, the strategy rules and constraints are integrated into the strategy generation process to generate a personalized operation strategy; Key operating indicators are monitored in real time. When the monitored indicators deviate from the expected targets, the operating strategies are adjusted and the target deep learning is optimized.

8. A personalized operation strategy generation system based on data processing, characterized in that: The personalized operation strategy generation system based on data processing includes: A data collection and processing module is used to collect multi-dimensional data from a target business system or a target business platform, pre-process the multi-dimensional data to obtain target multi-dimensional data, and store the target multi-dimensional data using a distributed storage architecture; A user portrait and feature extraction module, used to construct a comprehensive user portrait from multiple angles based on the target multi-dimensional data, extract static features and dynamic features from the target multi-dimensional data, and screen the static features and the dynamic features to obtain target decision features; A data analysis and visualization module, used to perform data analysis and data mining on the target decision-making features according to the user portrait, generate data analysis results, and present the data analysis results in an intuitive visualization chart; A model building and prediction module is used to build a deep learning model, train the deep learning model using historical data to obtain a target deep learning model, input the data analysis results into the target deep learning model, and output a prediction result; The operation strategy generation and adjustment module is used to generate a personalized operation strategy based on the prediction results, operation objectives, strategy rules and constraints, monitor the operation strategy in real time, and adjust the operation strategy in real time according to the monitoring results.

9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a personalized operation strategy generation program based on data processing stored in the memory and executable on the processor. When the personalized operation strategy generation program based on data processing is executed by the processor, the steps of the personalized operation strategy generation method based on data processing as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a personalized operation strategy generation program based on data processing, and when the personalized operation strategy generation program based on data processing is executed by a processor, the steps of the personalized operation strategy generation method based on data processing according to any one of claims 1 to 7 are implemented.

Images