A computer management system based on electronic bills

Abstract

The present invention discloses a computer management system based on electronic bills, and relates to the technical field of electronic bills. The system collects bills and transaction data from multiple consumption channels by integrating a bill data collection module. The collected data is also preprocessed by a preprocessing unit to generate a high-quality standard data set. In the feature extraction module, the system extracts a key consumption feature set from the standard data set and stores it in a database. Through the collaborative work of an abnormal analysis module, a comprehensive analysis module and an evaluation and early warning module, the system can use a multidimensional algorithm model to calculate the consumption period deviation Tdev, the consumption chain complexity Trcomplexity, the emotional fluctuation index Eindex and the consumption context factor Ccontext, and comprehensively calculate and generate a comprehensive abnormal score Stotal. The evaluation and early warning module monitors the comparison between the comprehensive abnormal score Stotal and the preset score threshold S in real time, triggers the early warning mechanism in time, and significantly improves the security of user funds and the convenience of consumption management.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic bills, and in particular to a computer management system based on electronic bills. Background Art

[0002] With the popularity of electronic payment and digital consumption, more and more users generate electronic bills through banks, payment platforms and merchant applications. These electronic bills record the user's consumption behavior and transaction details and are an important part of personal financial management. In order to improve the intelligence and security of personal financial management, electronic bill management systems have gradually become an indispensable tool to help users better track and analyze consumption behavior.

[0003] Although electronic bill management systems have played a positive role in helping users manage their consumption records, traditional systems have significant deficiencies in identifying and alerting abnormal consumption behaviors. Many existing systems can only provide simple bill classification and aggregation functions, but cannot accurately identify abnormal consumption behaviors that may pose risks in real time. This limitation may result in users failing to promptly detect potential financial risks, such as unauthorized transactions, consumer fraud, or irrational consumption behaviors caused by emotional fluctuations. In addition, the alarm mechanism of existing systems is often not sensitive enough, and is prone to false alarms or missed alarms, causing inconvenience to users.

[0004] This situation and shortcomings are mainly due to the limited ability of traditional systems to process complex, multi-dimensional data. Consumption behavior in electronic bills is not only related to the amount of consumption and merchant information, but also affected by many factors such as transaction time, location, user sentiment, market fluctuations, etc. Traditional systems lack the ability to comprehensively analyze these multi-dimensional data, and it is not easy to effectively incorporate these factors into risk assessment models. As a result, users may ignore normal consumption due to false alarms from the system, or fail to receive timely alerts when abnormal consumption behavior actually occurs, leading to potential financial losses or credit risks. The occurrence of this situation not only reduces users' trust in the system, but also weakens the security and effectiveness of personal financial management to a certain extent. Summary of the invention

[0005] In view of the deficiencies of the prior art, the present invention provides a computer management system based on electronic bills, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: including a bill data collection module, a feature extraction module, an abnormality analysis module, a comprehensive analysis module and an evaluation and early warning module;

[0007] The data collection module is used to set the connection port to integrate the consumption channel, collect bill data and transaction data in real time, and pre-process the collected bill data and transaction data to obtain a standard data set;

[0008] The feature extraction module is used to extract features based on the preprocessed standard data set, obtain the consumption feature set, and build a database to store the standard data set and the consumption feature set in the database;

[0009] The abnormal analysis module is used to build relevant algorithm models and extract consumption feature sets from the database, which are input into the relevant algorithm models in real time for calculation and output, including consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext;

[0010] The comprehensive analysis module constructs a comprehensive anomaly evaluation formula, extracts the consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext output by the anomaly analysis module, inputs them into the comprehensive anomaly evaluation formula, calculates and outputs the comprehensive anomaly score Stotal;

[0011] The evaluation and early warning module is used to receive the output comprehensive abnormality score Stotal in real time, preset the score threshold S for comparative evaluation, generate relevant evaluation reports, and then generate an alarm mechanism based on the evaluation report.

[0012] Preferably, the bill data collection module includes an integrated collection unit and a data preprocessing unit;

[0013] The integrated collection unit is used to build multiple API application interfaces to integrate with banks, payment platforms, trading markets, merchant evaluation platforms, social media software and electronic invoice systems, and develop SDK to encapsulate the calling logic of different API application interfaces, collect users' billing data and transaction data in real time, and then ensure data security through OAuth2.0 authentication mechanism;

[0014] The bill data includes the consumption amount Mc, the historical average consumption amount Mm, the market volatility △Mma, the benchmark market volatility Mba, the merchant reputation Rme, the merchant score Sme, the transaction amount Tam, the pre-consumption emotion score Ebe and the post-consumption emotion score Eaf;

[0015] The consumption amount Mc and the historical average consumption amount Mm are obtained by extracting statistical data of electronic bills;

[0016] The market volatility △Mma and the benchmark market volatility Mba are connected to the external trading market by using the API application program interface, extracting market data in real time, and obtaining them by mathematical calculation and analysis using Python;

[0017] The merchant reputation Rme, merchant rating Sme and transaction amount Tam are connected to the merchant evaluation platform through an API application program interface, transaction history data is extracted, and then calculated using an aggregation function;

[0018] The pre-consumption emotion score Ebe and the post-consumption emotion score Eaf are integrated with social media software, and the sentiment analysis tool NLP library is used to analyze the user's social media and communication content, extract the sentiment data and obtain it quantitatively;

[0019] The transaction data includes the transaction time point Tcurrent, the historical average consumption time point Tmean, the number of payment path nodes Lnodes and the number of currencies Lcurrencies;

[0020] The data preprocessing unit uses the Python library to clean the collected bill data and transaction data, remove redundant data in the bill data and transaction data, correct erroneous data, and process missing data. It then converts data formats from different sources into DataFrame format using a parsing library, and then converts numerical values ​​in different units into unified standard units to obtain a standard data set.

[0021] Preferably, the feature extraction module includes a feature extraction unit and a data storage unit;

[0022] The feature extraction unit is used to extract features based on the standard data set to obtain a consumption feature set, wherein the consumption feature set includes a consumption habit mutation rate Hchange, a multidimensional trust coefficient Mtrust, and a situation matching item score Ci;

[0023] The consumption habit mutation rate Hchange is obtained by extracting the consumption amount Mc and the historical average consumption amount Mm from the standard data set and performing aggregate calculation. The calculation formula is: ;

[0024] The multi-dimensional trust coefficient Mtrust is calculated by combining the merchant reputation Rme, merchant rating Sme, and transaction amount Tam in the standard data set. The calculation formula is: , where Rmean represents the historical merchant reputation mean, Smean represents the historical merchant rating mean, and Tan represents the historical transaction amount mean;

[0025] The context matching item score Ci is obtained by extracting the user's location matching Cocation, time matching Ctime, event matching Cevent, social interaction matching Csocial and historical behavior matching Chistory, and analyzing them to obtain a matching score. The context matching item score Ci={Cocation, Ctime, Cevent, Csocial and Chistory}, where Ci represents the score of the i-th context matching;

[0026] The location matching Cocation obtains and processes geographic location information through a geographic information system (GIS) tool, and uses the Haversine formula or other geographic distance algorithms to calculate the distance between the current consumption location and the historical consumption location. The closer the distance, the higher the score, and vice versa.

[0027] The time matching Ctime analyzes the user's consumption time series through Python's pandas library, extracts common time periods and calculates deviations, identifies the user's consumption time patterns through a clustering algorithm, and calculates the matching degree between the current consumption time and these patterns;

[0028] The event matching Cevent obtains the user's calendar events by using the calendar API, analyzes the time association between events and consumption, uses NLP technology to analyze social media content, identifies activities and points of interest related to consumption, matches consumption behavior with related events based on time and geographic location, and calculates the matching degree;

[0029] The social interaction matching Csocial analyzes the user's social media content and instant messaging records to identify whether the current consumption behavior is influenced by social factors. For example, if a user immediately purchases a brand after discussing it, the matching score is high. It analyzes the group behavior of users. If multiple people make similar purchases at the same time and place, it may be the result of collective activities, and the matching score is high. It measures the intensity of social interaction before and after consumption. If social interaction surges and the consumption behavior is related to the interactive content, the matching score is high.

[0030] The historical behavior matching Chistory uses the user's long-term consumption records, including consumption type, merchant, amount, etc., to analyze the user's past consumption records, the matching degree of current consumption with the user's historical consumption type, if the current consumption type is consistent with the user's historical preference, the score is higher, analyze the user's preference for specific merchants or brands, if the current consumption merchant is frequented by the user, the score is higher, the deviation between the current consumption amount and the user's historical consumption amount, if the amount is consistent with the user's historical consumption level, the score is higher;

[0031] The data storage unit is used to construct a relational database to store the standard data set and the consumption feature set.

[0032] Preferably, the abnormal analysis module includes a consumption deviation analysis unit, a transaction chain analysis unit, a mood fluctuation analysis unit and a consumption scenario analysis unit;

[0033] The consumption deviation analysis unit constructs a deviation algorithm formula, extracts the transaction time point Tcurrent, the historical average consumption time point Tmean and the consumption habit mutation rate Hchange in the standard data set and the consumption feature set from the database, inputs them into the deviation algorithm formula, and calculates the deviation of the acquisition fee period Tdev;

[0034] The consumption period deviation Tdev is calculated by the following deviation algorithm formula:

[0035] ;

[0036] In the formula, It represents the standard deviation of the user's historical consumption time, and log represents the logarithmic function.

[0037] Preferably, the transaction chain analysis unit constructs a transaction chain algorithm formula, extracts the number of payment path nodes Lnodes, the number of currencies Lcurrencies and the multi-dimensional trust coefficient Mtrust in the standard data set and the consumption feature set from the database, inputs them into the transaction chain algorithm formula, and outputs the consumption chain complexity Trcomplexity;

[0038] The consumption chain complexity Trcomplexity is calculated by the following transaction chain algorithm formula;

[0039] ;

[0040] Where Lsimple represents the complexity of the standard simple transaction chain.

[0041] Preferably, the emotion fluctuation analysis unit constructs an emotion fluctuation algorithm formula, extracts the pre-consumption emotion score Ebe and the post-consumption emotion score Eaf in the standard data set and the consumption feature set from the database, and combines the output consumption period deviation Tdev, inputs the emotion fluctuation algorithm formula together, and calculates and obtains the emotion fluctuation index Eindex;

[0042] The emotional fluctuation index Eindex is calculated by the following emotional fluctuation algorithm formula;

[0043] ;

[0044] Where Emax represents the peak value of emotion fluctuation.

[0045] Preferably, the consumption context analysis unit is used to construct a consumption context algorithm formula, then extract the context matching item score Ci in the consumption feature set from the database, input it into the consumption context algorithm formula, and output the consumption context factor Ccontext;

[0046] The consumption context factor Ccontext is obtained by calculation through a consumption context algorithm formula;

[0047] ;

[0048] Where n represents the total number of context matching items, and Wi represents the weight of the corresponding context matching item.

[0049] Preferably, the comprehensive analysis module constructs a comprehensive anomaly evaluation formula, inputs the obtained consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext into the comprehensive anomaly evaluation formula, and calculates and outputs a comprehensive anomaly score Stotal;

[0050] The comprehensive anomaly score Stotal is calculated and obtained by the following comprehensive anomaly evaluation formula;

[0051] ;

[0052] Where a1, a2, a3 and a4 represent the preset weight values ​​of consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotion fluctuation index Eindex and consumption context factor Ccontext respectively.

[0053] Preferably, the evaluation and early warning module includes a comprehensive evaluation unit and an early warning unit;

[0054] The comprehensive evaluation unit analyzes historical consumption data, calculates the average of historical scores, sets a score threshold S, and then extracts the real-time output comprehensive abnormal score Stotal and compares it with the score threshold S to determine the abnormality of the electronic bill. The specific evaluation content is as follows;

[0055] When the comprehensive abnormality score Stotal> the score threshold S, it is determined that the current electronic bill has abnormal consumption behavior, and the first evaluation information is generated;

[0056] When the comprehensive abnormality score Stotal=the score threshold S, it is determined that the current electronic bill has abnormal consumption behavior, and the second evaluation information is generated;

[0057] When the comprehensive abnormality score Stotal is less than the score threshold S, the current electronic bill consumption behavior is determined to be normal, and there is no need to generate evaluation information.

[0058] Preferably, the early warning unit is used to receive the generated evaluation information in real time, and trigger a corresponding early warning mechanism based on the received evaluation information. The specific triggering method is as follows;

[0059] When receiving the first evaluation information, the system executes the first early warning mechanism, by sending an abnormal consumption behavior alert to the user, immediately blocking the transaction behavior, and marking the current consumption behavior as "review" and submitting it for manual review and analysis;

[0060] When the second evaluation information is received, the system executes the second early warning mechanism by sending an abnormal consumption behavior alert to the user and performing multiple identity authentication. If the verification is successful, the transaction will continue; otherwise, the first early warning mechanism will be executed.

[0061] The present invention provides a computer management system based on electronic bills. It has the following beneficial effects:

[0062] (1) By introducing a bill data collection module, the system can be seamlessly integrated with multiple consumer channels such as banks, payment platforms, trading markets, merchant evaluation platforms, social media, and electronic invoice systems. This integration mechanism enables the system to automatically collect users' bill data and transaction data in real time, and perform data cleaning, format conversion, and unit standardization through the data preprocessing unit to generate a high-quality standard data set. This comprehensive and automated data collection and processing capability not only improves the accuracy and consistency of the data, but also provides a reliable foundation for subsequent consumer behavior analysis, thereby enhancing users' trust and reliance on the system.

[0063] (2) Through the feature extraction module, the system can extract the key consumption features of the user's consumption habit mutation rate Hchange, multidimensional trust coefficient Mtrust and context matching item score Ci from the preprocessed standard data set, and summarize them to generate a consumption feature set. The consumption feature set provides a rich information basis for the system to understand and analyze the user's consumption behavior. By establishing and storing the consumption feature set, the system can more accurately capture the changes in the user's consumption pattern and update the feature information in the database in real time. The anomaly analysis module constructs relevant algorithm formulas based on the acquired standard data set and consumption feature set, and calculates and outputs the consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotion fluctuation index Eindex and consumption context factor Ccontext, so that the system has higher analysis capabilities and stronger anomaly detection accuracy when dealing with complex consumption behaviors, helping users to better manage and monitor their consumption behaviors.

[0064] (3) Through the introduction of the anomaly analysis module and the comprehensive analysis module, the system can apply the key consumption features output by the feature extraction module, the consumption period deviation Tdev, the consumption chain complexity Trcomplexity, the emotional fluctuation index Eindex and the consumption context factor Ccontext, to the comprehensive anomaly evaluation formula to calculate the comprehensive anomaly score Stotal. The evaluation and early warning module automatically generates an evaluation report and triggers the corresponding early warning mechanism by comparing and evaluating with the preset score threshold S. This process ensures that the system can identify and handle abnormal consumption behaviors in real time. For example, when an anomaly is identified, the system can immediately send an alert to the user, block suspicious transactions or require multiple identity verification, thereby effectively preventing potential financial risks. This efficient and accurate anomaly detection and early warning mechanism not only improves the security of the system, but also enhances the confidence of users when facing complex consumption scenarios, ensuring that their financial security is effectively protected. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 The present invention is a flowchart of a computer management system based on electronic bills. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Example 1

[0068] See also Figure 1 The present invention provides a computer management system based on electronic bills. To achieve the above purpose, the present invention is implemented through the following technical solutions: including bill data collection module, feature extraction module, abnormal analysis module, comprehensive analysis module and evaluation and early warning module;

[0069] The data collection module is used to set up connection ports to integrate consumption channels, collect bill data and transaction data in real time, and pre-process the collected bill data and transaction data to obtain standard data sets;

[0070] The feature extraction module is used to extract features based on the preprocessed standard data set, obtain the consumption feature set, and build a database to store the standard data set and the consumption feature set in the database;

[0071] The abnormal analysis module is used to build relevant algorithm models and extract consumption feature sets from the database, which are input into relevant algorithm models in real time for calculation and output, including consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext;

[0072] The comprehensive analysis module constructs a comprehensive anomaly evaluation formula, extracts the consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext output by the anomaly analysis module, inputs them into the comprehensive anomaly evaluation formula, and calculates and outputs the comprehensive anomaly score Stotal;

[0073] The evaluation and early warning module is used to receive the output comprehensive anomaly score Stotal in real time, preset the score threshold S for comparative evaluation, generate relevant evaluation reports, and then generate an alarm mechanism based on the evaluation report.

[0074] In this embodiment, the system integrates the bill data collection module, and the system can collect bills and transaction data from multiple consumption channels in real time and automatically. The data collection module not only ensures the comprehensiveness and real-time nature of the data, but also cleans, standardizes and converts the collected data through the preprocessing unit to generate a high-quality standard data set. This automated and accurate data processing process effectively improves the system's ability to integrate multi-source data, laying a solid foundation for subsequent feature extraction and analysis. In the feature extraction module, the system extracts key consumption features from the standard data set, such as consumption habit mutation rate, multidimensional trust coefficient and context matching item score, and constructs these features into a consumption feature set and stores them in the database. Through the collaborative work of the abnormal analysis module, the comprehensive analysis module and the evaluation and early warning module, the system can use the multidimensional algorithm model to calculate the consumption period deviation Tdev, the consumption chain complexity Trcomplexity, the emotional fluctuation index Eindex and the consumption context factor Ccontext, and comprehensively calculate and generate the comprehensive abnormal score Stotal. The evaluation and early warning module monitors the comparison between the comprehensive abnormal score Stotal and the preset score threshold S in real time, and triggers the early warning mechanism in time. Compared with the traditional simple anomaly detection system, this system not only greatly improves the sensitivity and accuracy of anomaly detection, but also effectively reduces the probability of false positives and false negatives. Through intelligent analysis and real-time early warning of complex consumption scenarios, the system significantly improves the security of user funds and the convenience of consumption management.

[0075] Example 2

[0076] This embodiment is explained in Example 1, please refer to Figure 1,Specifically: the bill data collection module includes an ,integrated collection unit and a data pre-processing unit;

[0077] The integrated collection unit is used to build multiple API application interfaces to integrate with banks, payment platforms, trading markets, merchant evaluation platforms, social media software and electronic invoice systems, and develop SDK to encapsulate the calling logic of different API application interfaces, collect users' billing data and transaction data in real time, and then ensure data security through the OAuth2.0 authentication mechanism;

[0078] The bill data includes the consumption amount Mc, the historical average consumption amount Mm, the market volatility △Mma, the benchmark market volatility Mba, the merchant reputation Rme, the merchant score Sme, the transaction amount Tam, the pre-consumption emotion score Ebe, and the post-consumption emotion score Eaf;

[0079] The consumption amount Mc and the historical average consumption amount Mm are obtained by extracting statistical data from electronic bills;

[0080] The market volatility △Mma and the benchmark market volatility Mba are obtained by connecting to the external trading market through the API application program interface, extracting market data in real time, and performing mathematical operations and analysis using Python;

[0081] Merchant reputation Rme, merchant rating Sme and transaction amount Tam are connected to the merchant evaluation platform through the API application program interface, transaction history data is extracted, and then calculated using the aggregation function;

[0082] The pre-consumption emotion score Ebe and the post-consumption emotion score Eaf are integrated with social media software, using the sentiment analysis tool NLP library to analyze user social media and communication content, extract sentiment data and obtain it quantitatively;

[0083] Transaction data includes transaction time point Tcurrent, historical average consumption time point Tmean, number of payment path nodes Lnodes and number of currencies Lcurrencies;

[0084] The data preprocessing unit uses the Python library to clean the collected bill data and transaction data, remove redundant data in the bill data and transaction data, correct erroneous data, and process missing data. It then converts data formats from different sources into DataFrame format using the parsing library, and then converts values ​​in different units into unified standard units to obtain a standard data set.

[0085] In this embodiment, the system realizes the comprehensive integration and automatic processing of multi-channel, multi-format bills and transaction data by introducing a bill data collection module and its integrated collection unit and data preprocessing unit. The integrated collection unit integrates banks, payment platforms, trading markets, merchant evaluation platforms, social media and electronic invoice systems by building multiple API interfaces to ensure real-time and reliable acquisition of multi-dimensional data including consumption amount, market volatility, merchant reputation and sentiment score. At the same time, the application of OAuth2.0 authentication mechanism further enhances the security of data transmission. In the data preprocessing unit, the system uses the Python library to perform data cleaning, format conversion and unit standardization, and converts raw data from different sources into high-quality standard data sets. This process significantly improves the system's ability to integrate and process multi-source data, making the collection and preprocessing of bill data more efficient and accurate.

[0086] Example 3

[0087] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the feature extraction module includes a feature extraction unit and a data storage unit;

[0088] The feature extraction unit is used to extract features based on the standard data set to obtain the consumption feature set, which includes the consumption habit mutation rate Hchange, the multidimensional trust coefficient Mtrust and the situation matching item score Ci;

[0089] The consumption habit mutation rate Hchange is obtained by extracting the consumption amount Mc and the historical average consumption amount Mm from the standard data set and performing aggregate calculation. The calculation formula is ;

[0090] The multi-dimensional trust coefficient Mtrust is calculated by combining the merchant reputation Rme, merchant rating Sme, and transaction amount Tam in the standard data set. The calculation formula is: , where Rmean represents the historical merchant reputation mean, Smean represents the historical merchant rating mean, and Tan represents the historical transaction amount mean;

[0091] The context matching item score Ci is obtained by extracting the user's location matching Cocation, time matching Ctime, event matching Cevent, social interaction matching Csocial and historical behavior matching Chistory, and analyzing them to obtain the matching score. The context matching item score Ci={Cocation, Ctime, Cevent, Csocial and Chistory}, where Ci represents the score of the i-th context matching;

[0092] Location matching Cocation uses geographic information system (GIS) tools to obtain and process geographic location information, and uses the Haversine formula or other geographic distance algorithms to calculate the distance between the current consumption location and the historical consumption location. The closer the distance, the higher the score, and vice versa.

[0093] Time matching Ctime uses Python's pandas library to analyze the user's consumption time series, extract common time periods and calculate deviations, identify the user's consumption time patterns through clustering algorithms, and calculate the matching degree between the current consumption time and these patterns;

[0094] Event matching Cevent uses the calendar API to obtain the user's calendar events, analyzes the time correlation between events and consumption, uses NLP technology to analyze social media content, identifies activities and points of interest related to consumption, matches consumption behavior with related events based on time and geographic location, and calculates the matching degree;

[0095] Social interaction matching Csocial analyzes users' social media content and instant messaging records to identify whether current consumption behavior is influenced by social interaction. For example, if a user immediately buys a brand after discussing it, the matching score is high. It analyzes users' group behavior. If multiple people make similar purchases at the same time and place, it may be the result of collective activities, and the matching score is high. It measures the intensity of users' social interactions before and after consumption. If social interactions surge and consumption behavior is related to the interactive content, the matching score is high.

[0096] Historical behavior matching Chistory uses the user's long-term consumption records, including consumption type, merchant, amount, etc., to analyze the user's past consumption records, the matching degree between the current consumption and the user's historical consumption type. If the current consumption type is consistent with the user's historical preference, the score is higher. It also analyzes the user's preference for specific merchants or brands. If the current consumption merchant is frequented by the user, the score is higher. The deviation between the current consumption amount and the user's historical consumption amount. If the amount is consistent with the user's historical consumption level, the score is higher.

[0097] The data storage unit is used to construct a relational database to store standard data sets and consumption feature sets.

[0098] In this embodiment, the system provides the system with the ability to deeply analyze user consumption behavior through its feature extraction unit and data storage unit through the feature extraction module. The feature extraction unit uses the standard data set to extract and calculate key consumption features such as consumption habit mutation rate Hchange, multidimensional trust coefficient Mtrust and situation matching item score Ci. These features fully reflect the various dimensions of user consumption behavior through complex algorithms and data analysis technologies, such as aggregation computing, geographic information system GIS, NLP and clustering algorithms. This module can dynamically and finely identify changes in user consumption patterns, especially in situational factors such as location, time, events and social interactions, showing higher accuracy and sensitivity. In addition, the data storage unit stores the standard data set and consumption feature set in a relational database to ensure the structured and efficient access of data. This system architecture not only improves the persistence and security of data, but also provides strong data support for subsequent anomaly detection and analysis. Overall, the feature extraction module significantly improves the system's ability to understand and predict user consumption behavior, so that the accuracy of anomaly detection and the overall intelligence level of the system have been significantly enhanced.

[0099] Example 4

[0100] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the abnormal analysis module includes the consumption deviation ,analysis unit, transaction chain analysis unit, emotional fluctuation analysis unit and consumption ,scenario analysis unit;

[0101] The consumption deviation analysis unit constructs a deviation algorithm formula, extracts the transaction time point Tcurrent, the historical average consumption time point Tmean and the consumption habit mutation rate Hchange in the standard data set and the consumption feature set from the database, inputs them into the deviation algorithm formula, and calculates the deviation of the acquisition fee period Tdev;

[0102] The consumption period deviation Tdev is calculated by the following deviation algorithm formula;

[0103] ;

[0104] In the formula, It represents the standard deviation of the user's historical consumption time, and log represents the logarithmic function.

[0105] The transaction chain analysis unit constructs a transaction chain algorithm formula, extracts the number of payment path nodes Lnodes, the number of currencies Lcurrencies and the multi-dimensional trust coefficient Mtrust in the standard data set and the consumption feature set from the database, inputs them into the transaction chain algorithm formula, and outputs the consumption chain complexity Trcomplexity;

[0106] The consumption chain complexity Trcomplexity is calculated by the following transaction chain algorithm formula;

[0107] ;

[0108] Where Lsimple represents the complexity of the standard simple transaction chain.

[0109] The emotion fluctuation analysis unit constructs an emotion fluctuation algorithm formula, extracts the pre-consumption emotion score Ebe and the post-consumption emotion score Eaf from the standard data set and the consumption feature set from the database, and combines the output consumption period deviation Tdev with the emotion fluctuation algorithm formula to calculate and obtain the emotion fluctuation index Eindex;

[0110] The emotional fluctuation index Eindex is calculated by the following emotional fluctuation algorithm formula;

[0111] ;

[0112] Where Emax represents the peak value of emotion fluctuation.

[0113] The consumption context analysis unit is used to construct a consumption context algorithm formula, then extract the context matching item score Ci in the consumption feature set from the database, input it into the consumption context algorithm formula, and output the consumption context factor Ccontext;

[0114] The consumption context factor Ccontext is obtained by calculating the consumption context algorithm formula;

[0115] ;

[0116] Where n represents the total number of context matching items, and Wi represents the weight of the corresponding context matching item.

[0117] In this embodiment, the abnormal analysis module of the system provides the system with multi-dimensional abnormal behavior detection capabilities through its consumption deviation analysis, transaction chain analysis, emotional fluctuation analysis and consumption context analysis units. Each analysis unit uses a specific algorithm formula to extract relevant data from the database and calculate key abnormal indicators, such as consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext. Compared with traditional systems that rely only on simple rules for anomaly detection, this system can more accurately identify potential anomalies in consumer behavior through complex mathematical models and comprehensive analysis of multi-source data. For example, consumption deviation analysis not only considers the deviation between the current transaction time and the historical consumption time, but also introduces the calculation of consumption habit mutation rate Hchange and standard deviation, which greatly improves the detection accuracy of time anomalies; transaction chain analysis accurately measures transaction complexity by evaluating the number of payment path nodes Lnodes and the number of currencies Lcurrencies, combined with the multi-dimensional trust coefficient Mtrust, thereby identifying potential high-risk transactions; emotional fluctuation analysis reveals the impact of emotional fluctuations on consumer behavior by combining emotional scores and consumption time deviation; and consumption scenario analysis deeply explores the background rationality of consumer behavior through the comprehensive scoring of multi-scenario matching items. Overall, the anomaly analysis module significantly improves the system's detection capabilities and accuracy in complex consumption scenarios, enabling the system to more effectively prevent and respond to potential financial risks, and enhance overall security and user trust.

[0118] Example 5

[0119] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the comprehensive analysis module constructs a comprehensive anomaly evaluation formula, inputs the obtained consumption time period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext into the comprehensive anomaly evaluation formula, and calculates and outputs the comprehensive anomaly score Stotal;

[0120] The comprehensive anomaly score Stotal is calculated using the following comprehensive anomaly evaluation formula:

[0121] ;

[0122] Where a1, a2, a3 and a4 represent the preset weight values ​​of consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotion fluctuation index Eindex and consumption context factor Ccontext respectively.

[0123] In this embodiment, the comprehensive analysis module of the system constructs a comprehensive anomaly evaluation formula to comprehensively calculate multiple key anomaly indicators such as consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext, and finally outputs a unified comprehensive anomaly score Stotal. This multi-dimensional comprehensive scoring method can more accurately reflect the overall abnormality of consumer behavior. By assigning preset weight values ​​to each indicator, the system can flexibly adjust the evaluation focus according to different scenarios and user needs, further improving the accuracy and adaptability of anomaly detection.

[0124] Example 6

[0125] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the evaluation and early warning module includes a ,comprehensive evaluation unit and an early warning unit;

[0126] The comprehensive evaluation unit analyzes historical consumption data, calculates the average of historical scores, sets the score threshold S, and then extracts the real-time output comprehensive abnormal score Stotal and compares it with the score threshold S to determine the abnormality of the electronic bill. The specific evaluation content is as follows;

[0127] When the comprehensive abnormality score Stotal> the score threshold S, it is determined that the current electronic bill has abnormal consumption behavior, and the first evaluation information is generated;

[0128] When the comprehensive abnormality score Stotal=the score threshold S, it is determined that the current electronic bill has abnormal consumption behavior, and the second evaluation information is generated;

[0129] When the comprehensive abnormality score Stotal is less than the score threshold S, the current electronic bill consumption behavior is determined to be normal, and there is no need to generate evaluation information.

[0130] The early warning unit is used to receive the generated evaluation information in real time, and trigger the corresponding early warning mechanism based on the received evaluation information. The specific triggering method is as follows;

[0131] When receiving the first evaluation information, the system executes the first early warning mechanism, by sending an abnormal consumption behavior alert to the user, immediately blocking the transaction behavior, and marking the current consumption behavior as "review" and submitting it for manual review and analysis;

[0132] When the second evaluation information is received, the system executes the second early warning mechanism by sending an abnormal consumption behavior alert to the user and performing multiple identity authentication. If the verification is successful, the transaction continues; otherwise, the first early warning mechanism is executed.

[0133] In this embodiment, the evaluation and early warning module of the system realizes real-time monitoring and intelligent early warning of consumption behavior through the synergy of the comprehensive evaluation unit and the early warning unit. The comprehensive evaluation unit uses historical consumption data to calculate the scoring threshold S, and compares the comprehensive abnormal score Stotal generated in real time with the scoring threshold S to accurately judge the abnormal behavior in the electronic bill. When the comprehensive abnormal score Stotal exceeds the scoring threshold S, the system immediately generates corresponding evaluation information, and the early warning unit triggers the corresponding early warning mechanism, such as sending abnormal alarms, blocking transactions, or requiring multiple identity verification. Compared with the traditional static rule system, this module significantly improves the sensitivity and response speed of anomaly detection through dynamic evaluation and multi-level early warning mechanism, and reduces the possibility of false alarms and missed reports. Ultimately, this improvement not only improves the security of user funds, but also enhances the system's ability to prevent potential risks, allowing users to make daily consumption with greater peace of mind.

[0134] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A computer management system based on electronic bills, characterized by: It includes bill data collection module, feature extraction module, abnormal analysis module, comprehensive analysis module and evaluation and early warning module; The data collection module is used to set the connection port to integrate the consumption channel, collect bill data and transaction data in real time, and pre-process the collected bill data and transaction data to obtain a standard data set; The feature extraction module is used to extract features based on the preprocessed standard data set, obtain the consumption feature set, and build a database to store the standard data set and the consumption feature set in the database; The abnormal analysis module is used to build relevant algorithm models and extract consumption feature sets from the database, which are input into the relevant algorithm models in real time for calculation and output, including consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext; The comprehensive analysis module constructs a comprehensive anomaly evaluation formula, extracts the consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext output by the anomaly analysis module, inputs them into the comprehensive anomaly evaluation formula, calculates and outputs the comprehensive anomaly score Stotal; The evaluation and early warning module is used to receive the output comprehensive abnormality score Stotal in real time, preset the score threshold S for comparative evaluation, generate relevant evaluation reports, and then generate an alarm mechanism based on the evaluation report.

2. A computer management system based on electronic bills according to claim 1, characterized in that: The bill data collection module includes an integrated collection unit and a data preprocessing unit; The integrated collection unit is used to build multiple API application interfaces to integrate with banks, payment platforms, trading markets, merchant evaluation platforms, social media software and electronic invoice systems, and develop SDK to encapsulate the calling logic of different API application interfaces, collect users' billing data and transaction data in real time, and then ensure data security through OAuth2.0 authentication mechanism; The bill data includes the consumption amount Mc, the historical average consumption amount Mm, the market volatility △Mma, the benchmark market volatility Mba, the merchant reputation Rme, the merchant score Sme, the transaction amount Tam, the pre-consumption emotion score Ebe and the post-consumption emotion score Eaf; The transaction data includes the transaction time point Tcurrent, the historical average consumption time point Tmean, the number of payment path nodes Lnodes and the number of currencies Lcurrencies; The data preprocessing unit uses the Python library to clean the collected bill data and transaction data, remove redundant data in the bill data and transaction data, correct erroneous data, and process missing data. It then converts data formats from different sources into DataFrame format using a parsing library, and then converts numerical values ​​in different units into unified standard units to obtain a standard data set.

3. A computer management system based on electronic bills according to claim 2, characterized in that: The feature extraction module includes a feature extraction unit and a data storage unit; The feature extraction unit is used to extract features based on the standard data set to obtain a consumption feature set, wherein the consumption feature set includes a consumption habit mutation rate Hchange, a multidimensional trust coefficient Mtrust, and a situation matching item score Ci; The consumption habit mutation rate Hchange is obtained by extracting the consumption amount Mc and the historical average consumption amount Mm from the standard data set and performing aggregate calculation. The calculation formula is: ; The multi-dimensional trust coefficient Mtrust is calculated by combining the merchant reputation Rme, merchant rating Sme, and transaction amount Tam in the standard data set. The calculation formula is: , where Rmean represents the historical merchant reputation mean, Smean represents the historical merchant rating mean, and Tan represents the historical transaction amount mean; The context matching item score Ci is obtained by extracting the user's location matching Cocation, time matching Ctime, event matching Cevent, social interaction matching Csocial and historical behavior matching Chistory, and analyzing them to obtain a matching score. The context matching item score Ci={Cocation, Ctime, Cevent, Csocial and Chistory}, where Ci represents the score of the i-th context matching; The data storage unit is used to construct a relational database to store the standard data set and the consumption feature set.

4. A computer management system based on electronic bills according to claim 3, characterized in that: The abnormal analysis module includes a consumption deviation analysis unit, a transaction chain analysis unit, an emotion fluctuation analysis unit and a consumption scenario analysis unit; The consumption deviation analysis unit constructs a deviation algorithm formula, extracts the transaction time point Tcurrent, the historical average consumption time point Tmean and the consumption habit mutation rate Hchange in the standard data set and the consumption feature set from the database, inputs them into the deviation algorithm formula, and calculates the consumption period deviation Tdev; The consumption period deviation Tdev is calculated by the following deviation algorithm formula: ； In the formula, It represents the standard deviation of the user's historical consumption time, and log represents the logarithmic function.

5. A computer management system based on electronic bills according to claim 4, characterized in that: The transaction chain analysis unit constructs a transaction chain algorithm formula, extracts the number of payment path nodes Lnodes, the number of currencies Lcurrencies and the multi-dimensional trust coefficient Mtrust in the standard data set and the consumption feature set from the database, inputs them into the transaction chain algorithm formula, and outputs the consumption chain complexity Trcomplexity; The consumption chain complexity Trcomplexity is calculated by the following transaction chain algorithm formula; ； Where Lsimple represents the complexity of the standard simple transaction chain.

6. A computer management system based on electronic bills according to claim 4, characterized in that: The emotion fluctuation analysis unit constructs an emotion fluctuation algorithm formula, extracts the pre-consumption emotion score Ebe and the post-consumption emotion score Eaf in the standard data set and the consumption feature set from the database, and combines the output consumption period deviation Tdev with the emotion fluctuation algorithm formula to calculate and obtain the emotion fluctuation index Eindex; The emotional fluctuation index Eindex is calculated by the following emotional fluctuation algorithm formula; ； Where Emax represents the peak value of emotion fluctuation.

7. A computer management system based on electronic bills according to claim 4, characterized in that: The consumption context analysis unit is used to construct a consumption context algorithm formula, then extract the context matching item score Ci in the consumption feature set from the database, input it into the consumption context algorithm formula, and output the consumption context factor Ccontext; The consumption context factor Ccontext is obtained by calculating through a consumption context algorithm formula; ； Where n represents the total number of context matching items, and Wi represents the weight of the corresponding context matching item.

8. A computer management system based on electronic bills according to claim 4, characterized in that: The comprehensive analysis module constructs a comprehensive anomaly evaluation formula, inputs the obtained consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotional fluctuation index Eindex and consumption context factor Ccontext into the comprehensive anomaly evaluation formula, calculates and outputs the comprehensive anomaly score Stotal; The comprehensive anomaly score Stotal is calculated and obtained by the following comprehensive anomaly evaluation formula; ； Where a1, a2, a3 and a4 represent the preset weight values ​​of consumption period deviation Tdev, consumption chain complexity Trcomplexity, emotion fluctuation index Eindex and consumption context factor Ccontext respectively.

9. A computer management system based on electronic bills according to claim 8, characterized in that: The evaluation and early warning module includes a comprehensive evaluation unit and an early warning unit; The comprehensive evaluation unit analyzes historical consumption data, calculates the average of historical scores, sets a score threshold S, and then extracts the real-time output comprehensive abnormal score Stotal and compares it with the score threshold S to determine the abnormality of the electronic bill. The specific evaluation content is as follows; When the comprehensive abnormality score Stotal> the score threshold S, it is determined that the current electronic bill has abnormal consumption behavior, and the first evaluation information is generated; When the comprehensive abnormality score Stotal=the score threshold S, it is determined that the current electronic bill has abnormal consumption behavior, and the second evaluation information is generated; When the comprehensive abnormality score Stotal is less than the score threshold S, the current electronic bill consumption behavior is determined to be normal, and there is no need to generate evaluation information.

10. A computer management system based on electronic bills according to claim 9, characterized in that: The early warning unit is used to receive the generated evaluation information in real time, and trigger the corresponding early warning mechanism based on the received evaluation information. The specific triggering method is as follows; When receiving the first evaluation information, the system executes the first early warning mechanism, by sending an abnormal consumption behavior alert to the user, immediately blocking the transaction behavior, and marking the current consumption behavior as "review" and submitting it for manual review and analysis; When the second evaluation information is received, the system executes the second early warning mechanism by sending an abnormal consumption behavior alert to the user and performing multiple identity authentication. If the verification is successful, the transaction continues; otherwise, the first early warning mechanism is executed.

Images