Network live broadcast tax management method and system based on data analysis

Abstract

The invention discloses a network live broadcast tax management method and system based on data analysis, and relates to the field of data analysis and tax management, and the method comprises the steps: capturing transaction data and multi-source heterogeneous data, carrying out the preprocessing of the transaction data, and carrying out the dynamic tax calculation according to the preprocessed transaction data, a transaction record is generated by using a block chain technology, a tax prediction model is established based on the transaction record generated by the block chain technology, a tax burden prediction report is obtained, a structured data set is constructed based on the tax burden prediction report and multi-source heterogeneous data, and a user portrait is constructed by analyzing the structured data set and combining a tax revenue rule. And customizing a tax policy for the anchor by using an AI recommendation engine, and recording feedback of the user to the tax policy in the block chain. According to the method, the preprocessed transaction data is used for dynamic tax calculation, the user portrait is constructed, and the AI recommendation engine is used for customizing the tax strategy for the anchor in combination with the tax rule, so that the personalized level of the service is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the fields of data analysis and tax management, and particularly to a network live broadcast tax management method and system based on data analysis. Background Art

[0002] In recent years, with the rapid development of Internet technology, network live broadcast, as a new form of media dissemination, has rapidly emerged globally, giving rise to a huge business ecosystem. Along with this phenomenon come complex tax management challenges. The traditional tax management system appears inadequate when dealing with such new types of economic activities, especially in dynamic tax calculation and the personalized mechanisms for individual users have many limitations.

[0003] In the tax calculation link of traditional methods, most adopt static models and fail to fully consider the impacts of factors such as market fluctuations and policy changes, resulting in inaccurate prediction results. The current tax management platforms lack personalized mechanisms for individual users and are difficult to provide personalized tax strategies based on user portraits and behavior preferences, affecting the user experience and service quality. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a network live broadcast tax management method based on data analysis to solve the problems of insufficiently dynamic and accurate tax calculation and lack of personalized mechanisms.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a network live broadcast tax management method based on data analysis, which includes capturing transaction data and multi-source heterogeneous data, and preprocessing the transaction data;

[0008] Performing dynamic tax calculation based on the preprocessed transaction data, and generating a transaction record using blockchain technology;

[0009] Establishing a tax prediction model based on the transaction record generated by blockchain technology to obtain a tax burden prediction report;

[0010] Constructing a structured data set based on the tax burden prediction report and multi-source heterogeneous data;

[0011] Through the analysis of the structured data set, constructing a user portrait and combining with tax laws, and using an AI recommendation engine to customize tax strategies for the anchor;

[0012] Recording the user's feedback on the tax strategy in the blockchain.

[0013] As a preferred solution of the network live broadcast tax management method based on data analysis according to the present invention, wherein: an edge device is deployed at the anchor end to capture transaction data;

[0014] The multi-source heterogeneous data includes user behavior data, external economic indicators, and unstructured data;

[0015] The unstructured data includes text data, multimedia data, and social network data;

[0016] After performing data cleaning, formatting processing, and numerical standardization processing on the transaction data, it is encrypted and uploaded to the cloud server using the TSL security protocol;

[0017] As a preferred solution of the network live broadcast tax management method based on data analysis according to the present invention, wherein: according to the preprocessed transaction data, dynamic tax calculation is performed, and a transaction record is generated using blockchain technology, specifically including the following steps,

[0018] Based on the preprocessed transaction data received by the cloud server, a FaaS platform is pre-deployed and a neural network model is configured;

[0019] When the cloud server receives the preprocessed transaction data, it triggers the execution of the FaaS function;

[0020] When the FaaS function executes, it verifies the format of the preprocessed transaction data, and extracts the numerical features, categorical features, and time series features in the transaction data;

[0021] The extracted numerical features, categorical features, and time series features are converted in format and input into the neural network model, and the payable tax amount is output through layer-by-layer transmission of neurons;

[0022] The information of the payable tax amount is constructed into a transaction request, the anchor account digitally signs the transaction request, and the signed transaction request is sent to the blockchain network for verification and consensus processing by the nodes;

[0023] After verification and consensus processing by the nodes, the transaction request is written into the distributed ledger of the blockchain to generate a transaction record.

[0024] As a preferred solution of the network live broadcast tax management method based on data analysis according to the present invention, wherein: based on the transaction record generated by blockchain technology, a tax prediction model is established to obtain a tax burden prediction report, specifically including the following steps,

[0025] Define the input as the transaction record generated by blockchain technology and the output as the tax burden prediction report;

[0026] Extract relevant transaction records from the transaction records generated by the blockchain,

[0027] Extract basic transaction features, user behavior features, time series features, network features, external economic indicator features, and risk assessment features based on relevant transaction records;

[0028] Perform standardization processing on the extracted features and perform feature splicing to construct a feature vector;

[0029] Assign weights to the feature vector through business rules and machine automatic learning, calculate the feature importance value, and define the importance value function;

[0030] Perform weighted summation on the feature importance values, and define a structured data analysis function to perform information filtering and normalization processing on the feature vector to obtain a comprehensive value;

[0031] Use the normalization function to perform normalization processing on the comprehensive value to obtain the initial tax burden prediction value, and its expression is:

[0032]

[0033] Among them, M represents the number of transaction record entries, j represents the jth transaction, w j represents the weight of the transaction, X j represents the feature vector of the jth transaction, H j () represents the importance value function, σ() represents the normalization function, β represents the weight coefficient, represents the value of the feature vector X after information filtering, represents the value of the feature vector X after normalization processing, and P represents the initial tax burden prediction value;

[0034] Combine the initial prediction value with external economic indicators, market trends, and time series features in the transaction records to obtain the final tax burden prediction value;

[0035] Analyze the final tax burden prediction value, and generate a tax burden prediction report by combining background information on industry dynamics and market trends, as well as suggestions and countermeasures for reasonable income planning and risk management.

[0036] As a preferred solution of the network live broadcast tax management method based on data analysis described in the present invention, wherein: Connect the data in the tax burden prediction report to multi-source heterogeneous data, including user behavior data, external economic indicator data, text data, multimedia data, and social network data in unstructured data, and integrate all data into structured data;

[0037] Summarize the integrated structured data into a unified data framework and process it in a standard format to obtain a structured data set.

[0038] As a preferred solution of the network live broadcast tax management method based on data analysis according to the present invention, specifically: by analyzing the structured data set, constructing user portraits and combining tax laws, using an AI recommendation engine to customize tax strategies for live streamers, which specifically includes the following steps,

[0039] Use statistical analysis on the structured data set to extract the correlation characteristics of statistics;

[0040] Use a machine learning model to identify and predict features;

[0041] Use a deep learning algorithm to extract non-linear features and high-level abstract features, and combine the extracted features to form a comprehensive feature vector;

[0042] Define the transaction frequency and transaction amount functions to quantify the user's transaction behavior, and the expression for the feature intensity integral is:

[0043]

[0044] Where F i represents the feature intensity integral of the i-th user, t0 and t f respectively represent the start and end points of the time interval, Q represents the comprehensive feature vector, α represents the attenuation coefficient, f() represents the transaction frequency function, N is the total number of users, and g j () represents the transaction amount function;

[0045] Regarding the feature intensity integral, identify user features, group users using hierarchical clustering, and use the methods of association rule mining and time series analysis to obtain user behavior preferences and construct user portraits;

[0046] Combine the constructed user portraits with tax regulations, and select a deep learning model as the architecture according to the task type, model complexity, interpretability, and existing resources;

[0047] Train the deep learning model architecture to obtain a trained prediction model, which generates tax strategies based on user portraits and structured data sets.

[0048] As a preferred solution of the network live broadcast tax management method based on data analysis according to the present invention, specifically: record the user's feedback on the tax strategy in the blockchain, which specifically includes the following steps,

[0049] Use a direct feedback mechanism and a user interaction platform to obtain the user's feedback on the tax strategy;

[0050] Record the tax strategy and the user's feedback on the tax strategy in the blockchain.

[0051] Second aspect, the present invention provides a network live broadcast tax management system based on data analysis, including: a preprocessing module, a transaction record module, a tax prediction module, a data integration module, a strategy customization module, and a record module;

[0052] Preprocessing module: Capture transaction data and multi-source heterogeneous data, and preprocess the transaction data;

[0053] Transaction record module: Perform dynamic tax calculation based on the preprocessed transaction data, and generate transaction records using blockchain technology;

[0054] Tax prediction module: Based on the transaction records generated by blockchain technology, establish a tax prediction model and obtain a tax burden prediction report;

[0055] Data integration module: Based on the tax burden prediction report and multi-source heterogeneous data, construct a structured data set;

[0056] Strategy customization module: Through the analysis of the structured data set, construct a user profile and combine it with tax laws, and use an AI recommendation engine to customize tax strategies for the live broadcast host;

[0057] Record module: Record the user's feedback on the tax strategy in the blockchain.

[0058] Third aspect, the present invention provides a computer device, including a memory and a processor, where: when the computer program is executed by the processor, it implements any step of a network live broadcast tax management method based on data analysis as described in the first aspect of the present invention.

[0059] Fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, where: when the computer program is executed by the processor, it implements any step of a network live broadcast tax management method based on data analysis as described in the first aspect of the present invention.

[0060] The beneficial effects of the present invention are as follows: The present invention proposes a network live broadcast tax management method based on data analysis, uses the preprocessed transaction data for dynamic tax calculation, and generates immutable transaction records with the help of blockchain technology, enhancing transparency and credibility. Based on the transaction records generated by blockchain, a tax prediction model is established, and comprehensive analysis is carried out in combination with external economic indicators, market trends and internal factors to generate a detailed tax burden prediction report. Through in-depth analysis of the structured data set, a detailed user profile is constructed, and an AI recommendation engine is used to customize tax strategies for the live broadcast host in combination with tax laws, significantly improving the personalization level of the service, and having significant technical advantages and application prospects in the field of network live broadcast tax management. Description of the Drawings

[0061] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0062] Figure 1 It is a flowchart of a network live broadcast tax management method based on data analysis in Embodiment 1.

[0063] Figure 2 It is a schematic diagram of a network live broadcast tax management system based on data analysis in Embodiment 1. Specific Embodiments

[0064] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification.

[0065] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0066] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0067] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides a network live broadcast tax management method based on data analysis, including the following steps:

[0068] S1. Capture transaction data and perform preprocessing.

[0069] Specifically, it includes the following steps.

[0070] Set an embedded computer as an edge device, install a Linux operating component, a data acquisition tool, and a TLS security protocol, and deploy it in the host's working environment to ensure a reliable Internet connection near the device;

[0071] Use the live broadcast platform API interface and local transaction logs to capture transaction data, and set up a temporary data cache in the embedded computer to ensure that transaction data will not be lost in the case of unstable or disconnected networks;

[0072] Capture user behavior data from user behavior logs using front - end logging to optimize the user experience and enhance personalized services;

[0073] Capture external economic indicators by obtaining API keys from financial APIs and government statistical websites to ensure the accuracy and timeliness of data;

[0074] Capture text data through web scraping, capture multimedia data through file directory reading, and capture social network data through the social media listening tool SproutSocial;

[0075] Pre - process the captured transaction data, perform data cleaning on the transaction data through a hash table deduplication algorithm and validity check, remove outliers and fill in missing values;

[0076] Format the transaction data using field mapping and timestamp standardization to unify the format and standardize the timestamp;

[0077] Perform numerical standardization on the transaction data through currency unit conversion and numerical normalization operations, and encode categorical features to ensure data consistency;

[0078] Pre - enable the TLS security protocol on the embedded computer to ensure authentication and encryption during the communication process. Upload the pre - processed transaction data to the cloud server by batch upload. For critical transaction data, select real - time upload to ensure data timeliness, and set up a retry mechanism to achieve an automatic retry mechanism after upload failure to ensure the complete upload of data to the cloud server.

[0079] S2. According to the pre - processed transaction data, perform dynamic tax calculation and generate transaction records using blockchain technology.

[0080] Specifically, it includes the following steps:

[0081] Configure an HTTPS interface on the cloud server to receive the pre - processed transaction data and store it in the Redis database to ensure the reliability and response speed of subsequent processing;

[0082] Set up the Fass platform, package the trained neural network model into a Fass function, and pre - deploy it on the Fass platform. Configure the APIGateway service tool to receive the pre - processed transaction data on HTTPS and forward it to the FaaS function for processing;

[0083] When the FaaS function runs, the preprocessed transaction events received are verified using the JSONSchema tool, which improves the accuracy and efficiency of data processing, ensures the correct format, and extracts features;

[0084] Use Pandas and NumPy to extract numerical features from transaction data and perform normalization processing. Use Pandas and Scikit-learn to perform one-hot encoding or label encoding on categorical features. Use Pandas and the Datetime library to parse timestamps, extract time series features, and identify periodic patterns, ensuring the quality and consistency of the features and providing high-quality input data for subsequent model prediction;

[0085] Normalize the extracted numerical features so that the numerical features fall within a specific range and convert them into the format of a two-dimensional array;

[0086] Perform one-hot encoding, label encoding, and embedding encoding on the extracted categorical features, and convert them into the formats of a binary matrix, an integer array, and a low-dimensional dense vector;

[0087] Perform timestamp parsing and periodicity identification on the extracted time series features, and convert them into the formats of multiple new numerical features and periodic features;

[0088] Pass the converted numerical features, categorical features, and time series features to the input layer of the neural network model, and then pass them layer by layer through the neurons in each layer of the neural network. In each layer, perform a non-linear transformation through an activation function, and combine the existing weights and bias parameters in the model. Finally, reach the output layer and generate a prediction result. Perform an inverse transformation on the prediction result to extract the payable tax amount;

[0089] Combine the payable tax amount with the host ID, transaction amount, and timestamp to organize a structured data object, serialize it into a string using the JSON format, and build a transaction request based on the string to ensure that all necessary information is accurately included;

[0090] The transaction request is sent to the host account through the API interface. The host account randomly generates a private key to ensure transaction security using blockchain technology, and uses the private key to digitally sign the transaction request. Then, it is sent to the blockchain network through the smart contract interface, and the nodes verify the signature and perform consensus processing to ensure the security and immutability of the transaction;

[0091] After node verification and consensus processing, the signed transaction request is confirmed to be valid and packed into a new block. This block is synchronously propagated through all nodes in the blockchain network and finally written into the distributed ledger of the blockchain to generate an immutable and transparent transaction record.

[0092] S3. Establish a tax prediction model based on the transaction records generated by blockchain technology to obtain a tax burden prediction report.

[0093] Specifically, it includes the following steps:

[0094] Define the input as the transaction records generated by blockchain technology and the output as the tax burden prediction report.

[0095] From the transaction records generated by blockchain, use smart contracts and blockchain browsers to screen out the transaction records related to tax calculation, reducing the possibility of manual intervention and improving the speed and accuracy of data processing.

[0096] Directly extract the basic transaction features based on the relevant transaction records.

[0097] Extract user behavior features by analyzing user behavior data through clustering algorithms.

[0098] Extract time series features through time series analysis.

[0099] Extract network features through social network analysis.

[0100] Extract external economic indicator features through web scraping.

[0101] Extract risk assessment features through anomaly detection.

[0102] Standardize the extracted features to unify the scale, horizontally splice the standardized feature sets into a comprehensive feature matrix, and construct feature vectors, integrating multi-dimensional information and providing comprehensive data support for model training.

[0103] Combine business rules with a random forest machine learning model to ensure that the model conforms to the actual business logic and can fully utilize the power of data mining to discover hidden patterns. Use the feature vectors for model training, use historical transaction records as input, and use tax burden levels and risk categories as output. Obtain the importance scores of the feature vectors through model training, define an importance numerical function based on these scores, and map the scores to the same scale through normalization to obtain importance values. Dynamically adjust the weights of each feature vector according to the importance values to improve the accuracy and interpretability of the model and make it more adaptable to the changing data environment.

[0104] Sum the weighted importance values of the features, and define a structured data analysis function to perform information filtering and normalization processing on the feature vectors to obtain comprehensive values, ensuring the stability and consistency of the values and facilitating subsequent comparison and analysis.

[0105] The comprehensive value is normalized using a normalization function to eliminate the dimensional differences between different features and ensure that the predicted values are within a reasonable range. The expression for the initial tax burden prediction value is as follows:

[0106]

[0107] where M represents the number of transaction record entries, j represents the j-th transaction, w j represents the weight of the transaction, X j represents the feature vector of the j-th transaction, H j () represents the importance value function, σ() represents the normalization function, β represents the weight coefficient, represents the value of the feature vector X after information filtering, represents the value of the feature vector X after normalization, and P represents the initial tax burden prediction value;

[0108] The initial predicted value is combined with the GDP growth rate and inflation rate of external economic indicators, policy changes and market demand fluctuations of market trends, and seasonal changes and periodic trend characteristics of time series in transaction records, so that the prediction model is no longer limited to internal data, but can comprehensively consider the changes in the internal and external environments, and more accurately reflect the impact of the current economic environment on the tax burden, obtaining an accurate final tax burden prediction value;

[0109] The final tax burden prediction value is analyzed to identify potential risks and opportunities, and a detailed tax burden prediction report is generated by combining background information such as industry dynamics and market trends and reasonable suggestions for income planning and risk management, providing a reference basis for the anchor and assisting the anchor in making financial decisions.

[0110] S4. Construct a structured data set based on the tax burden prediction report and multi-source heterogeneous data.

[0111] Specifically, it includes the following steps:

[0112] The tax burden prediction data in the tax report is connected to user behavior data, external economic indicator data, text data, multimedia data, and social network data through an API interface to ensure that all relevant data is efficiently centralized together, providing a comprehensive information basis for subsequent analysis;

[0113] The above data is processed, standardized into a unified format, and data cleaning is performed to remove noise, duplicate data, and missing values, improving the accuracy and reliability of the data and providing a clean and high-quality data set for subsequent analysis;

[0114] Feature extraction is performed on the processed data, directly extracting features such as transaction amount, payable tax amount, and transaction date from the tax report burden prediction data;

[0115] Extract user type and viewing frequency characteristics in user behavior data by parsing user registration information and viewing logs;

[0116] Extract GDP growth rate and inflation rate characteristics in economic indicator data through financial APIs;

[0117] Extract sentiment analysis and keyword characteristics in text data through NLP natural language processing;

[0118] Extract color histograms, texture features, and sound spectrum features in multimedia data through computer vision and audio processing technologies;

[0119] Extract social influence and social relationship network characteristics in data from social networks through social network analysis algorithms;

[0120] By defining a unified data model and fields, integrate the extracted features into a structured data table in a standard format, and use the Talend ETL tool for data extraction, transformation, and loading, outputting structured data and storing it in a data warehouse to ensure data structuring and easy querying;

[0121] Select a suitable Hadoop technology stack, process the structured data in the data warehouse according to a unified data standard, and formulate a detailed metadata management strategy to build a unified data framework to ensure data consistency and traceability, and integrate the data in the data framework into a structured data set for subsequent analysis and modeling.

[0122] S5. Through the analysis of the structured data set, construct user portraits and combine tax laws, and use an AI recommendation engine to customize tax strategies for the anchors.

[0123] Specifically, it includes the following steps:

[0124] Use the statistical analysis tool Pandas for the structured data set to extract and calculate the Pearson correlation coefficient and Spearman rank correlation coefficient between features;

[0125] Use the random forest model in the Scikit-learn library for feature selection and predictive modeling, and extract transaction amount, user type, and transaction frequency as predictive features;

[0126] Use the deep learning framework of Keras to build a convolutional neural network CNN, and train the convolutional neural network CNN to extract edges and textures in images, fluctuations and trends in time series, complex interaction patterns and preferences in user behavior;

[0127] The PCA technique is used to find the principal components of the extracted features. The principal components capture the variation information in the extracted features, and a sufficient number of principal components are selected to ensure the maximum retention of variation information and a significant reduction in the dimension of the extracted features. Then, the original extracted features are projected onto the new coordinate system formed by the principal components, converted into a compact and redundancy-reduced comprehensive feature vector. This process not only simplifies the feature space and improves the computational efficiency but also preserves the main structure and patterns of the original data, thereby enhancing the performance and interpretability of the model;

[0128] Define a transaction frequency function to measure the number of transactions and transaction amounts of users at specific time nodes, so as to quantify user behavior, optimize business decisions, and improve computational efficiency;

[0129] Based on the above, the expression for the feature intensity integral is obtained as follows:

[0130]

[0131] where F i represents the feature intensity integral of the i-th user, t0 and t f represent the start and end points of the time interval respectively, Q represents the comprehensive feature vector, α represents the decay coefficient, f() represents the transaction frequency function, N is the total number of users, and g j () represents the transaction amount function;

[0132] The feature intensity integral quantifies the trading behavior of users, identifies the highest integral value of users, the change trend of integrals, the average integral, and the volatility characteristics;

[0133] Use the hierarchical clustering algorithm of AgglomerativeClustering in Scikit-learn to group users;

[0134] Apply the Apriori algorithm to the grouped users for association rule mining to discover user behavior patterns, use ARIMA time series analysis to predict the future behavior trends of users, analyze the behavior patterns and trends, obtain the behavior preferences of users, and construct user portraits based on the behavior preferences of users;

[0135] Combine the constructed user portraits with relevant tax regulations, including the General Data Protection Regulation, anti-money laundering regulations, transparency and information disclosure regulations, and compliance review regulations, to ensure that every step in the data processing and analysis process complies with laws and regulations, especially the regulations on privacy protection and data security;

[0136] Evaluate specific application scenarios and task requirements, clarify the expected output of the deep learning model, and select the MLP multi-layer perceptron model in the deep learning model as the architecture based on the task type, model complexity, interpretability, and existing resources;

[0137] Create new features by combining features such as transaction frequency, transaction amount, and behavior preferences in the user profile, use the new features as the input of the MLP (Multi-Layer Perceptron) model, and conduct training and validation to ensure that it can effectively capture users' transaction behaviors and preferences. During the entire modeling process, continuously conduct compliance reviews to ensure that all operations comply with tax regulations, and provide transparency reports to business personnel to explain the working principle and decision-making process of the MLP model;

[0138] Deploy the trained model to the production environment, track the model's performance, and regularly update the model and user profile according to changes in user behavior and new regulatory requirements to ensure the continuous effectiveness and compliance of the solution;

[0139] Identify users with high-risk features through the model output, such as frequent large transactions, abnormal expenditure patterns, etc. These users may require more stringent tax reviews or compliance guidance, and use the model prediction results to ensure that all operations comply with the latest tax regulations, avoid legal risks, and automatically generate tax return forms according to the results identified by the user profile and the model;

[0140] Combine the tax return statements with the GDP growth rate, inflation rate, and user behavior in the unstructured dataset to obtain personalized tax strategies for the live streamers, which can not only improve the quality of business decisions but also ensure that all operations comply with the requirements of laws and regulations.

[0141] S6. Record users' feedback on tax strategies in the blockchain.

[0142] Specifically, it includes the following steps:

[0143] Build a user interaction platform through the technology stack of the React front-end framework, including functional modules such as user registration and login, feedback submission interface, and real-time notification, and collect user feedback through multi-factor authentication, timely feedback submission, and real-time notification;

[0144] At the same time, implement a direct feedback mechanism, including embedding feedback buttons or forms in tax return forms, sending regular questionnaires via email or push notifications, encouraging users to provide more detailed feedback, and setting up dedicated customer service channels to collect users' oral feedback and record it in the background;

[0145] Collect users' feedback on tax strategies through the above-mentioned user interaction platform and direct feedback mechanism;

[0146] Classify the feedback content, use NLP technology to convert the unstructured text of the feedback content into structured data, and anonymize sensitive information to protect users' privacy;

[0147] Select the Ethereum blockchain platform for the feedback content converted into structured data, develop smart contracts to define how to record tax strategies and user feedback, set up permission management and query interfaces, and fully test the contracts to ensure their correct functionality. Then create a block structure, and construct each user feedback as a transaction containing fields such as the hashed user ID, tax strategy ID, feedback summary, timestamp, etc.

[0148] And ensure the authenticity and integrity of the feedback through private key signature verification. To improve efficiency, adopt batch submission and compression encryption technologies to optimize the process of uploading feedback structured data to the blockchain.

[0149] Establish a detailed logging, performance monitoring, and regular auditing mechanism to ensure the stable operation and security of the platform. The entire process not only realizes the secure and transparent storage of user feedback but also enhances the credibility and user experience of the platform, providing a solid foundation for subsequent data analysis and decision-making.

[0150] This embodiment also provides a webcast tax management system based on data analysis, including: a preprocessing module, a transaction record module, a tax prediction module, a data integration module, a strategy customization module, and a record module.

[0151] Preprocessing module: Capture transaction data and multi-source heterogeneous data, and preprocess the transaction data.

[0152] Transaction record module: Based on the preprocessed transaction data, perform dynamic tax calculations and generate transaction records using blockchain technology.

[0153] Tax prediction module: Based on the transaction records generated by blockchain technology, establish a tax prediction model and obtain a tax burden prediction report.

[0154] Data integration module: Based on the tax burden prediction report and multi-source heterogeneous data, construct a structured data set.

[0155] Strategy customization module: Through the analysis of the structured data set, construct user portraits and combine tax laws, and use an AI recommendation engine to customize tax strategies for the live streamers.

[0156] Record module: Record the users' feedback on the tax strategies in the blockchain.

[0157] This embodiment also provides a computer device applicable to a webcast tax management method based on data analysis, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement a webcast tax management method based on data analysis as proposed in the above embodiment.

[0158] The computer device can be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, carrier networks, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0159] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for network live broadcast tax management based on data analysis proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read Only Memory (EPROM for short), Programmable Red-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, magnetic disk, or optical disc.

[0160] In summary, the present invention: uses the preprocessed transaction data for dynamic tax calculation, generates an immutable transaction record with the help of blockchain technology to enhance transparency and credibility, establishes a tax prediction model based on the transaction record generated by the blockchain, conducts comprehensive analysis by combining external economic indicators, market trends, and internal factors to generate a detailed tax burden prediction report, constructs a detailed user profile through in-depth analysis of the structured data set, and uses an AI recommendation engine to customize tax strategies for the live broadcaster in combination with tax laws, significantly improving the personalization level of the service, and having significant technical advantages and application prospects in the field of network live broadcast tax management.

[0161] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A tax management method for live webcasting based on data analysis, characterized by: include, Capture transaction data and multi-source heterogeneous data, and pre-process transaction data; Dynamic tax calculation is performed based on pre-processed transaction data, and transaction records are generated using blockchain technology; Based on the transaction records generated by blockchain technology, a tax forecasting model is established to obtain a tax burden forecast report; Construct a structured data set based on tax burden forecast reports and multi-source heterogeneous data; By analyzing structured data sets, we build user portraits and combine them with tax laws to customize tax strategies for anchors using an AI recommendation engine. Record user feedback on tax strategies in the blockchain.

2. The network live broadcast tax management method based on data analysis as claimed in claim 1, characterized in that: Capture transaction data and multi-source heterogeneous data, and pre-process the transaction data, including the following steps: Deploy edge devices at the anchor end to capture transaction data; The multi-source heterogeneous data includes user behavior data, external economic indicators and unstructured data; Unstructured data includes text data, multimedia data, and network data; After data cleaning, formatting, and numerical standardization, the transaction data is encrypted using the TSL security protocol and uploaded to the cloud server.

3. The network live broadcast tax management method based on data analysis as claimed in claim 2, characterized in that: Based on the pre-processed transaction data, dynamic tax calculation is performed and transaction records are generated using blockchain technology, which specifically includes the following steps: Based on the pre-processed transaction data received by the cloud server, the FaaS platform is pre-deployed and the neural network model is configured; When the cloud server receives the pre-processed transaction data, it triggers the FaaS function to execute; When the FaaS function is executed, it verifies the format of the preprocessed transaction data and extracts the numerical features, categorical features, and time series features in the transaction data. The extracted numerical features, categorical features and time series features are converted into formats and input into the neural network model, which is passed layer by layer through neurons to output the value of the tax payable; The information on the amount of tax payable is constructed into a transaction request. The anchor account digitally signs the transaction request and sends the signed transaction request to the blockchain network for verification and consensus processing by the nodes. After node verification and consensus processing, the transaction request is written into the distributed ledger of the blockchain to generate a transaction record.

4. The network live broadcast tax management method based on data analysis as claimed in claim 3, characterized in that: Based on the transaction records generated by blockchain technology, a tax forecasting model is established to obtain a tax burden forecasting report, which specifically includes the following steps: The input is defined as the transaction record generated by blockchain technology, and the output is the tax burden forecast report; Extract relevant transaction records from the transaction records generated by the blockchain, Based on relevant transaction records, extract basic transaction characteristics, user behavior characteristics, time series characteristics, network characteristics, external economic indicator characteristics, and risk assessment characteristics; The extracted features are standardized and concatenated to construct feature vectors; Assign weights to feature vectors through business rules and machine learning, calculate feature importance values ​​and define importance value functions; The importance values ​​of the features are weighted and summed, and a structured data analysis function is defined to filter and normalize the feature vectors to obtain a comprehensive value. The normalization function is used to normalize the comprehensive value to obtain the initial tax burden forecast value, which is expressed as follows: Where M represents the number of transaction record entries, j represents the jth transaction, and w j represents the weight of the transaction, Xj represents the feature vector of the jth transaction, and H j () represents the importance numerical function, σ() represents the normalization function, β represents the weight coefficient, Represents the value of the feature vector X after information filtering, represents the normalized value of the eigenvector X, and P represents the initial tax burden forecast value; Combine the initial forecast with external economic indicators, market trends, and time series features in transaction records to get the final tax burden forecast; The final tax burden forecast value is analyzed, and a tax burden forecast report is generated based on background information on industry dynamics, market trends, and recommended measures for reasonable revenue planning and risk management.

5. The network live broadcast tax management method based on data analysis as claimed in claim 4, characterized in that: Based on the tax burden forecast report and multi-source heterogeneous data, a structured data set is constructed, which includes the following steps: Connect the data in the tax burden forecast report to multi-source heterogeneous data, including user behavior data, external economic indicator data, text data, multimedia data, and social network data in unstructured data, and integrate all data into structured data; The integrated structured data is aggregated into a unified data framework and processed according to a standard format to obtain a structured data set.

6. The network live broadcast tax management method based on data analysis as claimed in claim 5, characterized in that: By analyzing structured data sets, building user portraits and combining tax rules, we use the AI ​​recommendation engine to customize tax strategies for anchors. The specific steps include: Use statistical analysis to extract statistical correlation features from structured data sets; Use machine learning models to extract discriminative and predictive features; Use deep learning algorithms to extract nonlinear features and high-level abstract features, and combine the extracted features to form a comprehensive feature vector; Define transaction frequency and transaction amount functions to quantify user transaction behavior, and get the expression of feature strength integral as follows: where F i represents the feature strength integral of the i-th user, t0 and t f They represent the starting point and end point of the time interval, Q represents the comprehensive feature vector, α represents the attenuation coefficient, f() represents the transaction frequency function, N is the total number of users, and g j () represents the transaction amount function; Based on feature strength integration, user features are identified, users are grouped using hierarchical clustering, and user behavior preferences are obtained using association rule mining and time series analysis methods to build user portraits; Combine the constructed user profile with tax regulations and select a deep learning model as the architecture based on the task type, model complexity, interpretability, and existing resources; The deep learning model architecture is trained to obtain a trained prediction model that generates tax strategies based on user profiles and structured data sets.

7. The network live broadcast tax management method based on data analysis as claimed in claim 6, characterized in that: The user's feedback on the tax strategy is recorded in the blockchain, which includes the following steps: Get user feedback on tax strategies using direct feedback mechanisms and user interaction platforms; The tax strategy and user feedback on the tax strategy are recorded in the blockchain.

8. A network live broadcast tax management system based on data analysis, based on a network live broadcast tax management method based on data analysis according to any one of claims 1 to 7, characterized in that: Including pre-processing module, transaction record module, tax forecasting module, data integration module, strategy customization module, and record module; Preprocessing module: captures transaction data and multi-source heterogeneous data, and preprocesses transaction data; Transaction record module: performs dynamic tax calculation based on pre-processed transaction data and generates transaction records using blockchain technology; Tax forecasting module: Based on the transaction records generated by blockchain technology, a tax forecasting model is established to obtain a tax burden forecasting report; Data integration module: construct structured data sets based on tax burden forecast reports and multi-source heterogeneous data; Strategy customization module: By analyzing structured data sets, building user portraits and combining tax laws, the AI ​​recommendation engine is used to customize tax strategies for anchors; Recording module: Record user feedback on tax strategies in the blockchain.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the network live broadcast tax management method based on data analysis as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the network live broadcast tax management method based on data analysis as described in any one of claims 1 to 7 are implemented.

Images