A Big Data-Based Business Audit Integration System

Abstract

This invention relates to the field of financial auditing technology, specifically disclosing a big data-based integrated auditing system, comprising: a data acquisition and preprocessing module, a standardized feature extraction module, a transaction link analysis module, a risk comprehensive scoring calculation module, and an intelligent audit report generation module. Compared to existing technologies that rely on static rule matching and manual review, especially in decentralized trading environments where transaction data is cross-platform, account identities are concealed, and transaction paths are complex, existing technologies cannot accurately identify high-risk transactions and abnormal fund flows. This invention achieves efficient, real-time, and dynamic financial auditing through real-time streaming data processing, multi-scale time series alignment, transaction path risk propagation analysis, and reinforcement learning to optimize risk scoring.

Description

Technical Field

[0001] This invention belongs to the field of financial auditing technology, and in particular relates to a business audit integration audit based on big data. Background Technology

[0002] Currently, decentralized finance (DeFi) auditing is developing rapidly, and its openness and permissionless nature have driven the globalization of financial transactions. However, existing auditing methods have many shortcomings when facing DeFi auditing, mainly manifested in: decentralized transactions and a lack of effective auditing and regulatory tools; high-frequency trading and smart contract operations increasing auditing difficulty; limited ability of traditional auditing methods to identify complex transaction patterns; difficulty in tracking cross-platform transactions, leading to risk diffusion; and, due to a lack of effective risk control measures on some decentralized exchanges, attackers can create new accounts in batches to conduct abnormal transactions, and existing auditing systems, due to insufficient data correlation, struggle to accurately identify these risky accounts. Therefore, there is an urgent need for an intelligent auditing method that can operate efficiently in the DeFi auditing environment, identify complex transaction patterns, track cross-platform fund flows, and automatically generate audit reports, thereby improving the security, compliance, and transparency of DeFi transactions. This invention provides a big data-based business auditing integration method that combines real-time streaming data processing, dynamic feature optimization, reinforcement learning risk assessment, and transaction path analysis. Even in complex transaction environments with high concurrency, cross-platform operation, and multiple smart contract interactions, it can still achieve accurate decentralized financial transaction auditing, improve the detection capability of high-risk transactions, reduce false alarm rates, and enhance audit efficiency, thereby meeting the needs of decentralized financial business auditing for real-time, dynamic, and intelligent auditing. Summary of the Invention

[0003] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a big data-based integrated audit system and method. This aims to solve the technical problem that existing technologies rely on static rule matching and manual review, especially in decentralized trading environments where transaction data is cross-platform, account identities are concealed, and transaction paths are complex. As a result, existing technologies cannot accurately identify high-risk transactions and abnormal fund flows.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a business audit integration system based on big data, comprising:

[0005] The data acquisition and preprocessing module is used to collect decentralized finance audit data. It uses Apache Kafka and Flink to process the real-time data stream of decentralized finance audit data and remove abnormal data to obtain preprocessed financial audit data.

[0006] The standardized feature extraction module is used to process the preprocessed financial audit data using a multi-scale time series alignment method and perform double-standard normalization to obtain the financial audit feature matrix.

[0007] The transaction link analysis module is used to construct a multi-dimensional transaction network based on the financial industry audit feature matrix, calculate the similarity of transaction patterns based on the multi-dimensional transaction network and combine it with cross-platform pattern matching analysis to obtain risky transaction groups; and obtain transaction path risk data by performing path detection on risky transaction groups.

[0008] The risk comprehensive score calculation module is used to obtain transaction data, account behavior data and fund flow data on the transaction path based on the transaction path risk data, and calculate the risk comprehensive score.

[0009] The intelligent audit report generation module is used to generate interpretable audit reports based on a comprehensive risk score combined with semantic analysis.

[0010] Preferably, in the data acquisition and preprocessing module, the decentralized finance audit data includes transaction data, account behavior data, and fund flow data; transaction data includes transaction amount, transaction time, counterparty, and transaction type; account behavior data includes user login frequency, device change, and IP change records; and fund flow data includes account fund inflows and account fund outflows.

[0011] Preferably, in the data acquisition and preprocessing module, the step of processing the real-time data stream of decentralized finance audit data using Apache Kafka and Flink and removing abnormal data specifically includes the following steps:

[0012] Apache Kafka and Flink are used to process the real-time data stream of decentralized finance (DeFi) audit data to obtain the abnormal transaction feature vector Z of the DeFi audit data;

[0013] Based on the abnormal transaction feature vector Z and decentralized finance audit data X t The probability P of outlier data is calculated using the variational autoencoder method.

[0014] Set an outlier probability threshold ∈ th If P(X) t )<∈ th Then remove X t .

[0015] Preferably, in the standardized feature extraction module, the step of processing the preprocessed financial audit data using a multi-scale time series alignment method employs the following formula:

[0016]

[0017] in, X is the smoothed financial audit data for the i-th transaction at time j; i,t * For the i-th transaction at time t, financial industry audit data; w t α is the trading volume weighting factor; α is the time decay coefficient, used to control the impact of forward transactions on current financial data; T is a preset time window; exp(·) is the exponential function.

[0018] The steps of dual-standard normalization processing specifically include: Z-score standardization is applied to transaction data in the preprocessed financial audit data; and Min-Max normalization is applied to account behavior data in the preprocessed financial audit data.

[0019] Preferably, in the transaction link analysis module, the step of constructing a multi-dimensional transaction network based on the financial industry audit feature matrix specifically includes: introducing the financial industry audit feature matrix M = {W1, W2, ..., W...} generated by the standardized feature extraction module. n}, where W p Let n be the dynamic feature weight of the p-th transaction feature in the financial industry audit feature matrix, and n be the total number of transaction features. Construct a multi-dimensional transaction network G based on the financial industry audit feature matrix M: G = (V, E), E e,r =f(X) e ',X r ',M), where G is the multidimensional transaction network; V is the account node; E e,r X represents the transaction path between transaction e and transaction r; e ',X r 'Transaction data and account behavior data in financial audit data that have been normalized by the standardized feature extraction module.'

[0020] Preferably, in the transaction link analysis module, the step of calculating the similarity of transaction patterns based on the multi-dimensional transaction network and combining it with cross-platform pattern matching analysis to obtain risky transaction groups specifically includes:

[0021] The mean and variance of financial industry audit features are calculated based on the financial industry audit feature matrix M. The similarity of transaction patterns S is calculated based on the mean μ and variance σ of the financial industry audit features. i ; Among them, W p X' represents the dynamic feature weights corresponding to the pre-defined financial industry audit data. p This represents the financial industry audit feature value corresponding to the p-th transaction feature;

[0022] Using dynamic feature weights W pAfter weighting the financial industry audit feature matrix M, anomaly pattern scores are calculated using cross-platform transaction pattern matching: Where Q is the abnormal mode score, and A e,r Let be the path weight of the transaction path between the e-th transaction and the r-th transaction in the multidimensional transaction network G;

[0023] A preset abnormal pattern scoring threshold is set, and transactions with an abnormal pattern score greater than the abnormal pattern scoring threshold are classified as risky transactions.

[0024] The steps to obtain transaction path risk data by performing path detection on risky trading groups specifically include: introducing a constrained risk propagation coefficient exp(-βd) e,r ) Calculate suspicious fund flow paths to obtain transaction path risk data: R path =Σ e,r A e,r ·exp(-βd e,r )·Q i , where R path For transaction path risk data, d e,r Let β be the length of the transaction path, β be the risk propagation coefficient weight control factor, and exp(·) be the exponential function.

[0025] Preferably, the step of obtaining transaction data, account behavior data, and fund flow data along the transaction path based on the transaction path risk data, and calculating the comprehensive risk score in the risk comprehensive score calculation module specifically includes:

[0026] High-risk paths are identified based on transaction path risk data.

[0027] Detailed transaction data, including transaction data, account behavior data, and fund flow data, is extracted from high-risk paths.

[0028] Calculate abnormal scores for transaction amount, transaction frequency, account behavior, and fund flow based on transaction data, account behavior data, and fund flow data.

[0029] A weighted comprehensive scoring method is used to calculate the overall risk score of the path; a risk warning threshold is defined, and a warning mechanism is triggered based on the overall risk score and the risk warning threshold, including: low risk: continue monitoring, no immediate action required; medium risk: mark the account and include it in the list of key transactions; high risk: trigger abnormal transaction interception.

[0030] The beneficial effects of this invention are as follows: Compared with the existing technology that relies on static rule matching and manual review, especially in the decentralized trading environment where transaction data is cross-platform, account identities are hidden, and transaction paths are complex, the existing technology cannot accurately identify high-risk transactions and abnormal fund flows. This invention achieves efficient, real-time, and dynamic financial auditing by using real-time streaming data processing, multi-scale time series alignment, transaction path risk propagation analysis, and reinforcement learning to optimize risk scoring. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This invention provides a schematic diagram of a business audit integration system based on big data.

[0033] Figure 2 This is a schematic diagram of an equipment for a big data-based business audit integration system provided by the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Example 1: As Figure 1 The diagram shown is a schematic of a big data-based business audit integration system provided by the present invention, and an embodiment of the big data-based business audit integration system of the present invention is presented.

[0036] In this embodiment, the big data-based business audit integration system includes:

[0037] The data acquisition and preprocessing module is used to collect decentralized finance audit data. It uses Apache Kafka and Flink to process the real-time data stream of decentralized finance audit data and remove abnormal data to obtain preprocessed financial audit data.

[0038] It should be noted that in the data acquisition and preprocessing module, the step of processing the real-time data stream of decentralized finance (DeFi) audit data using Apache Kafka and Flink to remove outliers specifically includes: processing the real-time data stream of DeFi audit data using Apache Kafka and Flink to obtain the abnormal transaction feature vector Z of the DeFi audit data; and then, based on the abnormal transaction feature vector Z and the DeFi audit data X... t The variational autoencoder method is used to calculate the probability P of outlier data; an outlier data probability threshold ∈ is set. th If P(X) t )<∈ th Then remove X t .

[0039] It should be understood that by introducing Variational Autoencoders (VAEs), this method can accurately detect and remove outlier data in large-scale real-time data streams, ensuring the accuracy of subsequent data processing. This method calculates anomaly scores for each transaction and determines whether a transaction is abnormal based on these scores, effectively removing outlier data that could lead to errors. By employing Apache Kafka and Flink for real-time data stream processing, the system can guarantee real-time response to large-scale decentralized finance data, while the anomaly detection using VAEs ensures the high accuracy of the removed data, thus preventing erroneous data from affecting subsequent audit results.

[0040] For example, in actual operation, when certain transaction data, such as abnormal transaction amounts or excessively high transaction frequency, is detected, VAE calculates anomaly scores for those transactions and decides whether to remove them based on set thresholds. For instance, in one experiment, the system detected frequent large-amount transfer transactions in real time, whose anomaly scores far exceeded the threshold. Therefore, these transactions were marked as abnormal data and removed, thus ensuring the accuracy of the system's processing.

[0041] The standardized feature extraction module is used to process the preprocessed financial audit data using a multi-scale time series alignment method and perform double-standard normalization to obtain the financial audit feature matrix.

[0042] It should be noted that in the standardized feature extraction module, the step of processing the preprocessed financial audit data using the multi-scale time series alignment method employs the following formula:

[0043]

[0044] in, X is the smoothed financial audit data for the i-th transaction at time j; i,t *For the i-th transaction at time t, financial industry audit data; w t α is the trading volume weighting factor; α is the time decay coefficient, used to control the impact of forward transactions on current financial data; T is a preset time window; exp(·) is the exponential function.

[0045] Since financial transaction data may have different time distributions (such as seconds, minutes, hours), this formula can ensure that all transaction data is converted to the same time scale.

[0046] The steps of dual-standard normalization processing specifically include: Z-score standardization is applied to transaction data in the preprocessed financial audit data; and Min-Max normalization is applied to account behavior data in the preprocessed financial audit data.

[0047] Understandably, the introduction of multi-scale time series alignment ensures effective alignment of data across different time scales. This is particularly crucial in decentralized finance (DeFi) auditing, where different transaction types have varying data collection granularities. This method guarantees that all transaction data can be calculated and analyzed on the same timeline, avoiding feature distortion caused by time dimension mismatch. The adoption of a dual-standard normalization strategy allows transaction data and account behavior data to be reasonably standardized across different feature spaces, improving the consistency of data distribution across different feature types and thus enhancing the model's adaptability and generalization ability.

[0048] For example, in a certain experimental environment, the system processed the transaction data of a DeFi platform over the past 30 days. It was found that without using a multi-scale time series alignment method, the inconsistent time granularity of the data led to an accuracy rate of less than 70% for identifying some transaction patterns. However, after adopting this method, the accuracy rate for transaction pattern identification increased to 88%. Simultaneously, using Z-score normalization to process transaction amounts improved the detection capability of abnormal transaction amounts by 25%, while Min-Max normalization improved the accuracy rate of account behavior pattern identification by 18%. These experimental results demonstrate that this invention, through multi-level optimization of data via a standardized feature extraction module, improves the interpretability of financial industry audit data and the model learning capability, providing a solid foundation for subsequent transaction pattern analysis and risk assessment.

[0049] The transaction link analysis module is used to construct a multi-dimensional transaction network based on the financial industry audit feature matrix, calculate the similarity of transaction patterns based on the multi-dimensional transaction network and combine it with cross-platform pattern matching analysis to obtain risky transaction groups; and obtain transaction path risk data by performing path detection on risky transaction groups.

[0050] It should be noted that the steps in the transaction link analysis module to construct a multi-dimensional transaction network based on the financial industry audit feature matrix specifically include: introducing the financial industry audit feature matrix M = {W1, W2, ..., W...} generated by the standardized feature extraction module. n}, where W p Let n be the dynamic feature weight of the p-th transaction feature in the financial industry audit feature matrix, and n be the total number of transaction features. Construct a multi-dimensional transaction network G based on the financial industry audit feature matrix M: G = (V, E), E e,r =f(X) e ',X r ',M), where G is the multidimensional transaction network; V is the account node; E e,r X represents the transaction path between transaction e and transaction r; e ',X r 'Transaction data and account behavior data in financial audit data that have been normalized by the standardized feature extraction module.'

[0051] In the transaction link analysis module, the steps for calculating transaction pattern similarity based on the multi-dimensional transaction network and combining it with cross-platform pattern matching analysis to obtain risky trading groups specifically include:

[0052] The mean and variance of financial industry audit features are calculated based on the financial industry audit feature matrix M. The similarity of transaction patterns S is calculated based on the mean μ and variance σ of the financial industry audit features. i ; Among them, W p X' represents the dynamic feature weights corresponding to the pre-defined financial industry audit data. p This represents the financial industry audit feature value corresponding to the p-th transaction feature;

[0053] Using dynamic feature weights W p After weighting the financial industry audit feature matrix M, anomaly pattern scores are calculated using cross-platform transaction pattern matching: Where Q is the abnormal mode score, and A e,r Let be the path weight of the transaction path between the e-th transaction and the r-th transaction in the multidimensional transaction network G;

[0054] A preset abnormal pattern scoring threshold is set, and transactions with an abnormal pattern score greater than the abnormal pattern scoring threshold are classified as risky transactions.

[0055] The steps to obtain transaction path risk data by performing path detection on risky trading groups specifically include: introducing a constrained risk propagation coefficient exp(-βd) e,r ) Calculate suspicious fund flow paths to obtain transaction path risk data: R path =Σ e,r Ae,r ·exp(-βd e,r )·Q i , where R path For transaction path risk data, d e,r Let β be the length of the transaction path, β be the risk propagation coefficient weight control factor, and exp(·) be the exponential function.

[0056] Understandably, constructing a multi-dimensional transaction network can effectively uncover hidden transaction patterns, especially in cross-platform trading scenarios. Through weighted optimization of the financial industry's feature matrix, transaction patterns across different platforms can be uniformly modeled and compared, thereby improving the accuracy of cross-platform transaction pattern matching. Furthermore, calculating transaction pattern similarity can effectively filter out accounts with highly similar trading behaviors. Combined with transaction path risk detection, it can further track the transaction flow of suspicious accounts and identify potential risky trading groups.

[0057] It should be understood that employing Cross-Platform Pattern Matching (CTPM) and Transaction Path Risk Analysis (CRPE) can identify abnormal fund flow patterns hidden within transaction paths, avoiding the static analysis of isolated transactions in traditional transaction auditing methods and ensuring dynamic monitoring capabilities of transaction risks. Furthermore, optimizing risk calculations within the transaction network through a dynamic feature weighting method allows the influence of different types of transactions in the model to adaptively adjust, ensuring the stability and interpretability of the analysis results.

[0058] For example, in a certain experimental environment, the system analyzed the transaction data of a DeFi platform over the past 60 days and found that the transaction patterns of a certain group of accounts were highly similar, and the flow of funds frequently jumped between multiple DeFi platforms. Through multi-dimensional transaction network analysis, the system calculated that the similarity score of these accounts' transaction patterns was as high as 92%, and combined with transaction path analysis, the risk score of their fund flow path was calculated to be 87%. Subsequently, the system used cross-platform pattern matching to find that this risky transaction group had an 85% similarity with a known money laundering account, and finally the system successfully identified the hidden risky transaction network. The experimental results show that the transaction link analysis method provided by this invention can accurately identify high-risk transaction groups, improve cross-platform risk detection capabilities, and enhance transaction path tracking effects, thereby effectively improving the accuracy of decentralized finance auditing.

[0059] The risk comprehensive score calculation module is used to obtain transaction data, account behavior data and fund flow data on the transaction path based on the transaction path risk data, and calculate the risk comprehensive score.

[0060] It should be noted that the risk comprehensive score calculation module involves the following steps: obtaining transaction data, account behavior data, and fund flow data along the transaction path based on the transaction path risk data, and calculating the risk comprehensive score.

[0061] High-risk paths are identified based on transaction path risk data.

[0062] Detailed transaction data, including transaction data, account behavior data, and fund flow data, is extracted from high-risk paths.

[0063] Calculate abnormal scores for transaction amount, transaction frequency, account behavior, and fund flow based on transaction data, account behavior data, and fund flow data.

[0064] A weighted comprehensive scoring method is used to calculate the overall risk score of the path; a risk warning threshold is defined, and a warning mechanism is triggered based on the overall risk score and the risk warning threshold, including: low risk: continue monitoring, no immediate action required; medium risk: mark the account and include it in the list of key transactions; high risk: trigger abnormal transaction interception.

[0065] It should be noted that high-risk paths are selected from transaction path risk data. The selection criteria are based on the risk propagation score of the path, which is calculated by weighting the risks of all transactions on the path. The selection process includes: calculating the average risk score and standard deviation of all transaction paths, and then setting a risk threshold. If the risk score of a transaction path exceeds the threshold, the path is considered a high-risk path and proceeds to the next step of in-depth analysis.

[0066] For the high-risk paths identified, extract the transaction data, account behavior data, and fund flow data involved, and calculate the following four types of anomaly scores respectively:

[0067] Transaction Amount Anomaly Score: This score determines whether the amount of a transaction significantly deviates from the historical average. If the amount is much higher than the average transaction amount, it is considered an abnormal transaction.

[0068] Abnormal Transaction Frequency Score: This score calculates the frequency of an account's transactions within a given time period. If an account engages in frequent transactions within a short period, it may indicate that the account is involved in arbitrage, money laundering, or other high-risk activities.

[0069] Account behavior anomaly scoring: Analyze account behavior patterns, including device switching frequency and IP address changes. If an account frequently switches login devices in a short period of time, it may indicate that it is trying to evade audit tracking.

[0070] Abnormal Fund Flow Scoring: Analyze the account's fund inflow and outflow patterns. If the account experiences a large inflow of funds in a short period of time, followed by immediate splitting and transfer out, it may be a typical characteristic of money laundering.

[0071] Understandably, the risk comprehensive scoring module ensures that risk assessment is not based on a single transaction characteristic, but rather comprehensively considers multiple dimensions such as transaction amount, transaction frequency, account behavior, and fund flow through multi-dimensional anomaly detection, thereby providing a more accurate risk score. Furthermore, the introduction of a dynamic risk threshold optimization mechanism allows the system to adaptively adjust to changes in the market environment, reducing false positive rates and improving the identification rate of high-risk transactions.

[0072] It should be understood that the transaction path-level risk calculation method, compared to the traditional approach based on single-transaction risk scoring, can more comprehensively analyze the transaction chain risks in decentralized finance (DeFi) audits. This is especially true when funds flow through multiple accounts, multiple trading platforms, or when smart contracts are used for automated execution; this method can effectively identify potential risky transactions in complex transaction chains. Furthermore, the weighted comprehensive scoring method combined with reinforcement learning to optimize thresholds makes risk assessment more accurate and adaptable to market changes.

[0073] For example, in a risk monitoring experiment on a DeFi trading platform, the system analyzed 500,000 transactions over the past 90 days. The experiment found that without transaction path-level risk scoring, the system only detected 75% of high-risk transactions, with a false positive rate of 15%. However, after applying the risk comprehensive scoring method of this invention, the system successfully identified 92% of high-risk transactions, and the false positive rate decreased to 5%. Further analysis revealed that the system could accurately identify the covert fund transfer behavior of a single account across multiple DeFi platforms and uncovered the underlying abnormal fund flow patterns through path risk propagation analysis. This demonstrates that the risk comprehensive scoring method of this invention can effectively improve the risk identification capability of decentralized finance audits, reduce false positive rates, and improve the overall audit efficiency of the system.

[0074] The intelligent audit report generation module is used to generate interpretable audit reports based on a comprehensive risk score combined with semantic analysis.

[0075] It should be noted that semantic analysis can be performed on the report content using natural language processing (NLP) to ensure that the generated report is concise and easy to understand, and can clearly explain the risk score and potential risky trading behavior.

[0076] It should be understood that this module is designed to improve the transparency and efficiency of audits, especially in the decentralized finance (DeFi) environment where massive amounts of data and complex transaction patterns render traditional auditing methods inadequate for real-time and efficient operations. By introducing semantic analysis, high-dimensional data can be transformed into concise text reports, enabling financial regulators to more quickly understand complex risk assessment results and make informed decisions. Furthermore, this module is adaptive, automatically optimizing report templates and content structure to adapt to evolving transaction patterns and new audit requirements.

[0077] For example, the intelligent audit report generation module uses Natural Language Processing (NLP) to process the report content, ensuring that the generated reports are concise and easy to understand, clearly explaining risk scores and potential risky trading behaviors. After NLP processing, the following types of report content are generated: Risk Level Description: Based on the risk score, it automatically categorizes risks into low, medium, and high risk, and provides a concise explanation for each category. For example, Key Abnormal Trading Behaviors: It lists and explains key abnormal factors such as abnormal fund flows, abnormal transaction amounts, and frequent trading behaviors; Risk Sources: It clearly identifies the main sources of risk (such as a specific account, counterparty, fund flow, etc.); Generating a Complete Audit Report: It organizes all analysis results into a structured audit report, including: Risk Score: It provides a comprehensive risk score for the path; Report Summary: It briefly summarizes the risk status of the current trading path; Detailed Analysis: For each high-risk path, it lists and explains the main factors affecting the risk score; Audit Recommendations: It provides further recommendations or warnings, such as whether manual review of certain transactions is needed, or whether the account should be included in monitoring.

[0078] Example 2: In addition, the present invention also provides a business audit integration device based on big data, please refer to... Figure 2A big data-based integrated auditing device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to execute the big data-based integrated auditing system described in Embodiment 1 above. The big data-based integrated auditing device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This big data-based integrated auditing device is merely an example and should not limit the functionality or scope of the embodiments of this invention. A big data-based integrated auditing device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the big data-based integrated auditing device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows a big data-based business audit convergence device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a big data-based business audit convergence device with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0079] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the system diagram can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined above in the system of the embodiments disclosed in this invention.

[0080] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0081] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A business auditing system based on big data, characterized in that the system... include: The data acquisition and preprocessing module is used to collect decentralized finance audit data. It uses Apache Kafka and Flink to process the real-time data stream of decentralized finance audit data and remove abnormal data to obtain preprocessed financial audit data. The standardized feature extraction module is used to process the preprocessed financial audit data using a multi-scale time series alignment method and perform double-standard normalization to obtain the financial audit feature matrix. The transaction link analysis module is used to construct a multi-dimensional transaction network based on the financial industry audit feature matrix, calculate the similarity of transaction patterns based on the multi-dimensional transaction network and combine it with cross-platform pattern matching analysis to obtain risky transaction groups; and obtain transaction path risk data by performing path detection on risky transaction groups. The risk comprehensive score calculation module is used to obtain transaction data, account behavior data and fund flow data on the transaction path based on the transaction path risk data, and calculate the risk comprehensive score. The intelligent audit report generation module is used to generate interpretable audit reports based on a comprehensive risk score combined with semantic analysis. In the standardized feature extraction module, the step of processing the preprocessed financial audit data using a multi-scale time series alignment method employs the following formula: ;in, For the first The transaction at the time point Smoothed financial industry audit data; For the first The transaction in time Financial industry data auditing; Volume-weighted factor; This is the time decay coefficient, used to control the impact of forward transactions on current financial audit data; This is a preset time window; ( ) represents the exponential function; The steps of dual-standard normalization processing specifically include: applying Z-score standardization to transaction data in the preprocessed financial audit data; and applying Min-Max normalization to account behavior data in the preprocessed financial audit data. In the transaction link analysis module, the steps for constructing a multi-dimensional transaction network based on the financial industry audit feature matrix specifically include: introducing the financial industry audit feature matrix generated by the standardized feature extraction module. ,in, The first in the financial industry audit feature matrix The dynamic feature weights of each transaction feature, where n is the total number of transaction features; based on the financial industry audit feature matrix. Building a multi-dimensional trading network : ,in, For multi-dimensional transaction networks; For account nodes; This represents the transaction path between transaction e and transaction r. Transaction data and account behavior data in financial audit data that have been normalized by the standardized feature extraction module; In the transaction link analysis module, the steps for calculating transaction pattern similarity based on the multi-dimensional transaction network and combining it with cross-platform pattern matching analysis to obtain risky trading groups specifically include: Based on the financial industry audit feature matrix Calculate the mean and variance of financial industry audit characteristics, based on the mean of financial industry audit characteristics. and financial industry audit characteristics variance Calculate transaction pattern similarity ; ,in, The dynamic feature weights corresponding to the preset financial industry audit data. For the first Financial industry audit feature values ​​corresponding to each transaction feature; Using dynamic feature weights Financial industry audit feature matrix After weighting, the abnormal pattern score is calculated using cross-platform transaction pattern matching: ,in, Scoring for abnormal patterns For multi-dimensional transaction networks The path weight of the transaction path between transaction e and transaction r; A preset abnormal pattern scoring threshold is set, and transactions with an abnormal pattern score greater than the abnormal pattern scoring threshold are classified as risky transactions. The steps to obtain transaction path risk data by performing path detection on risky trading groups specifically include: introducing a constrained risk propagation coefficient. Calculate suspicious fund flow paths to obtain transaction path risk data: ,in, For transaction path risk data, The length of the transaction path. As a weighted control factor for the risk transmission coefficient, It is an exponential function.

2. The business audit integration system based on big data as described in claim 1, characterized in that, In the data collection and preprocessing module, decentralized finance (DeFi) audit data includes transaction data, account behavior data, and fund flow data. Transaction data includes transaction amount, transaction time, counterparty, and transaction type. Account behavior data includes user login frequency, device change, and IP change records. Fund flow data includes account fund inflows and outflows.

3. The business audit integration system based on big data as described in claim 1, characterized in that, In the data acquisition and preprocessing module, the steps of processing the real-time data stream of decentralized finance audit data using Apache Kafka and Flink and removing outlier data specifically include: The real-time data stream of decentralized finance (DeFi) audit data is processed using Apache Kafka and Flink to obtain anomaly transaction feature vectors from the DeFi audit data. ; Based on the abnormal transaction feature vector and decentralized finance industry data audit The variational autoencoder method is used to calculate the probability of outlier data. ; Set an abnormal data probability threshold ,like Then remove .

4. The business audit integration system based on big data as described in claim 1, characterized in that, The risk comprehensive score calculation module involves the following steps: First, it obtains transaction data, account behavior data, and fund flow data along the transaction path based on the transaction path risk data. These steps are then used to calculate the comprehensive risk score. High-risk paths are identified based on transaction path risk data. Detailed transaction data, including transaction data, account behavior data, and fund flow data, is extracted from high-risk paths. Calculate abnormal scores for transaction amount, transaction frequency, account behavior, and fund flow based on transaction data, account behavior data, and fund flow data. A weighted comprehensive scoring method is used to calculate the overall risk score of the path; a risk warning threshold is defined, and a warning mechanism is triggered based on the overall risk score and the risk warning threshold, including: low risk: continue monitoring, no immediate action required; medium risk: mark the account and include it in the list of key transactions; high risk: trigger abnormal transaction interception.

Images