Factoring fund information security management method and system

CN119067767APending Publication Date: 2024-12-03CICC TONGSHENG COMMERCIAL FACTORING CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411042780.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-12-03

Smart Images

  • Figure CN119067767A_ABST
    Figure CN119067767A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of financial data processing, in particular to a factoring fund information security management method and system, and the method comprises the following steps: collecting factoring staff data, inputting factoring fund information data matched with factoring staff, and carrying out the data processing and screening of the factoring fund information data, and extracting behavior data of the factoring employees, and distinguishing and marking sensitive information and corresponding sensitive data features. Based on evaluation and risk prediction results, factoring funds associated and matched with employees, credit limit limitation, guarantee requirement increase and risk control warning line setting are performed, risks are reduced, benefits of factoring providers are protected, the situation that the employees access sensitive information without authorization can be found in time, and the safety of the employees is improved. The behavior of performing fraudulent activities by using the authority is avoided, and the factoring fund information security management effect and security are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of financial data processing, and in particular to a factoring fund information security management method and system thereof. Background Art

[0002] Factoring funds refer to the financing provided by the factoring company, a financial institution that provides factoring services, to the seller after the seller transfers its current or future accounts receivable generated by the goods sales / service contracts with the buyer to the factoring company. This financial service not only covers financing, but also includes a series of services such as buyer credit assessment, sales account management, credit risk guarantee, and account collection.

[0003] During the period of factoring guarantee, factoring funds are the financial support provided by the factor to the seller after the seller transfers its accounts receivable to the factor. This will involve the issue of Poly Fund information security. However, in actual operations, there are not only network risks, but also internal operation risks. Employees may access sensitive information without authorization, or use their authority to commit fraudulent activities. In routine operations, the operating procedures may also have defects, resulting in data processing errors or capital loss, which will lead to a reduction in the effectiveness and quality of factoring fund information security management. Summary of the invention

[0004] The purpose of the present invention is to provide a factoring fund information security management method and system thereof, so as to solve the problem raised in the above-mentioned background technology that in actual operation, there are not only risks on the network, but also risks in internal operations. Employees may access sensitive information without authorization, or use their authority to commit fraudulent activities. In addition, during routine operations, the operating procedures may also have defects, resulting in data processing errors or capital loss, which will reduce the effectiveness and security of factoring fund information security management.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for secure management of factoring fund information, comprising the following steps:

[0006] Collecting factoring company employee data and inputting factoring fund information data matching the factoring company employee, and processing and screening the factoring company employee data and factoring fund information data, extracting factoring company employee behavior data and distinguishing and marking sensitive information and corresponding sensitive data features;

[0007] Responding to the sensitive data features of the factoring company's employee behavior data, extracting data associated with the sensitive data features through a machine learning algorithm and generating historical associated behavior data, analyzing the factoring company's employee behavior trends and directions based on the associated behavior data, and then predicting the factoring company's employee behavior trends to generate risk prediction results; wherein the behavioral trend prediction specifically, after classifying the factoring company's employee sensitive information and the matched historical associated behavior data based on the sensitive data features, using the time series of the long short-term memory network LSTM to generate the time series of the factoring company's employee personal behavior data and construct a factoring fund information risk value prediction model, and generating corresponding evaluation and risk prediction results based on the factoring fund information risk value prediction model;

[0008] After determining the sensitive data of the factoring company's employees and matching it with the factoring fund information data associated with the factoring company's employees, the factoring fund information risk value prediction model is used for analysis. According to the results of the evaluation and risk prediction, the adjustment plan strategy for the factoring funds is matched and determined, and corresponding adjustments are made according to the adjustment plan strategy for the factoring funds.

[0009] Preferably, the factoring company employee data and factoring fund information data are processed and screened, wherein the factoring fund information data is screened according to preset conditions or rules, specifically, each factoring company employee data is used as the starting node of the relationship map, the association relationship between each factoring company employee and the user is analyzed, and the user who has a direct association relationship with the factoring company employee is used as the child node of the starting node, and the user who has a direct association relationship with each child node user is used as the next node of the child node. This step is repeated to form a multi-level relationship map until the preset level is reached, thereby completing the establishment of the relationship map between each user and the employee.

[0010] Preferably, the factoring fund information data includes data reflecting fund flow, risk status and user repayment ability, and is divided into user basic information category, financial status category, business operation category, risk control category, and company structure category, and key data points in each category are extracted.

[0011] Preferably, the sensitive information of the factoring company's employees is classified with the matched historical associated behavior data, and specifically, the data is classified and analyzed by applying the K nearest neighbor classification algorithm through transaction features, system access features, communication features, and file transfer features, and the K training samples closest to the sample to be classified in the feature space are found according to the transaction features, system access features, communication features, and file transfer features, and the distance between the classified sample and the training sample is calculated using a distance metric, and a voting decision is made based on the category labels of the transaction features, system access features, communication features, and file transfer features, and their average or weighted average is calculated to predict the numerical output of the sample to be predicted, and the classification algorithm class result is obtained, and the distance metric between the classified sample and the training sample is calculated as follows:

[0012]

[0013] Dist(x i ,y j ) is the distance measure between the classification sample and the training sample, x ik is the kth classification sample x i ,y jk is the kth training sample y j .

[0014] Preferably, the method further includes ensuring the accuracy and consistency of extracting the behavioral data of the factoring company's employees and distinguishing and marking the sensitive information and the corresponding sensitive data feature data, comprising the following steps:

[0015] Check whether the data meets the requirements according to the reasonable value range and mutual relationship of each variable, and find and correct logically unreasonable or contradictory data;

[0016] Further processing of invalid and missing values: If the preprocessing stage does not fully resolve the problem, this stage further processes the problem through manual comparison.

[0017] Preferably, the step of constructing a factoring fund information risk value prediction model is:

[0018] Receive factoring fund information risk instructions;

[0019] Obtaining the time series of factoring fund information risk categories and factoring company employee personal behavior data from the factoring fund information risk instruction;

[0020] Searching for a preset risk prediction model corresponding to the risk category of the factoring fund information, and obtaining decision parameters in the preset risk prediction model;

[0021] Collecting real-time decision factor data corresponding to the decision parameters;

[0022] The risk classification threshold of the preset risk prediction model is adjusted according to the real-time decision factor data to obtain a real-time factoring fund information risk value prediction model, which is specifically as follows:

[0023]

[0024] in, is the predicted value of the risk value of the factoring fund information of the vth factoring company employee, is the risk assessment value of the tth collection time series of the uth factoring fund information of the vth factoring employee, is the risk assessment judgment threshold of the u-th factoring fund information of the v-th factoring company employee, T is the number of items in the t-th collected time series, and α is the optimization coefficient;

[0025] The time series of the factoring company's employee personal behavior data is input into the real-time factoring fund information risk value prediction model to obtain the assessed risk category and the corresponding risk value prediction value, and the corresponding assessment and risk prediction results are generated according to the assessed risk category and the corresponding risk value prediction value.

[0026] Preferably, the analysis and generation of the factoring fund information risk assessment value of the factoring company's employees is as follows:

[0027] Constructing a sample data set of factoring fund information risk assessment values ​​of factoring company employees according to the factoring fund information risk categories and decision parameters;

[0028] After dividing the sample data set into a training data set and a validation data set, determining, based on the training data set, a plurality of decision parameters of risk categories associated with factoring fund information risks of factoring company employees;

[0029] By using a multi-factor binary logistic regression analysis method, in combination with the training data set, the regression coefficients corresponding to the intercept term and the decision parameters of each risk category in the training data set are obtained, and according to the regression coefficients corresponding to the intercept term and each characteristic parameter; the intercept term and the decision parameters of each risk category in the training data set are simultaneously used as parameters to be verified, and a factoring fund information risk value prediction model is constructed by verifying all the parameters to be verified and the regression coefficients corresponding to all the verified parameters to be verified;

[0030] Based on the analysis of the factoring fund information risk value prediction model, a factoring fund information risk assessment value of the factoring company's employees is generated and output to the real-time factoring fund information risk value prediction model.

[0031] Preferably, the real-time decision factor data includes: calculating the factoring fund information risk volatility coefficient according to the decision parameters;

[0032] Finding a data sample period corresponding to the risk volatility coefficient of the factoring fund information;

[0033] Get the latest decision factor data updated in real time;

[0034] Real-time decision factor data is extracted from the latest decision factor data according to the data sample period.

[0035] Preferably, the adjustment plan strategy of the factoring funds is adjusted through business model adjustment, initiating entity adjustment, credit risk management, trade type adjustment, user strategy adjustment, financing timing adjustment and risk control strategy adjustment.

[0036] The system for factoring fund information security management includes data collection module, data processing module, data analysis module, behavior trend prediction module and program strategy adjustment module;

[0037] The data collection module is connected to the data processing module, and the data collection module is used to collect factoring company employee data and input factoring fund information data matching the factoring company employee. The data processing module is used to process and screen the factoring company employee data and factoring fund information data, extract the factoring company employee behavior data, and distinguish and mark sensitive information and corresponding sensitive data features;

[0038] The data analysis module is used to analyze the processing results of the data processing module, respond to the sensitive data features of the factoring company's employee behavior data, and extract data associated with the sensitive data features through a machine learning algorithm and generate historical associated behavior data;

[0039] The behavior trend prediction module is used to analyze the behavior trend and direction of the factoring company's employees according to the associated behavior data, and then predict the behavior trend of the factoring company's employees to generate a risk prediction result;

[0040] The data collection module includes database, API access, crawler and user input;

[0041] The data processing module includes a data acquisition module, a data transmission module, a data preprocessing module, a data storage module, a data visualization module, a metadata management module, a job management and scheduling module, and a resource allocation and scheduling module;

[0042] The data analysis module includes a click stream data analysis module, a visit time analysis module, a stay time analysis module and a related behavior data analysis module;

[0043] The behavior trend prediction module includes feature selection, data set selection, division and training module, model construction, evaluation and optimization module, feedback and iteration module.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] In the present invention, by collecting user factoring fund information, valuable information is extracted and sensitive information is marked, the behavioral data of factoring company employees accessing the internal website is collected, and the factoring fund information data is classified and pre-processed to ensure the accuracy and consistency of the data, thereby avoiding data processing errors or fund losses caused by defects in the operating process, and by analyzing the behavioral data of factoring company employees accessing the internal work website, the personal credit status of factoring company employees is evaluated using models and algorithms, the default risk and fraud risk that the factoring company may face in the future are predicted, and the results of credit evaluation and risk prediction are output. Based on the results of credit evaluation and risk prediction, the factoring funds associated with the individual factoring company employees are restricted in credit line, collateral requirements are increased, and risk control warning lines are established, so as to reduce risks and protect the interests of the factoring company, and the situation of unauthorized access to sensitive information by factoring company employees can be discovered in time, so as to avoid the use of power to commit fraudulent activities, thereby improving the effect and security of factoring fund information security management. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flowchart of the present invention;

[0047] Figure 2 It is a schematic diagram of the system module structure of the present invention. DETAILED DESCRIPTION

[0048] 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.

[0049] See also Figure 1-Figure 2 , an embodiment provided by the present invention:

[0050] The method for managing factoring fund information security includes the following steps:

[0051] S1. Collect the factoring company employee data and input the factoring fund information data matching the factoring company employee, and process and screen the factoring company employee data and the factoring fund information data, extract the factoring company employee behavior data and distinguish and mark the sensitive information and corresponding sensitive data features;

[0052] Data collection: Users provide and enter factoring fund information data, and the factoring company can collect internal employees’ personal information and preferences through forms, questionnaires, registration, etc. Website tracking: The factoring company can use data tracking technology to collect employees’ behavioral data when visiting the website, such as click streams, visit time, dwell time, etc.

[0053] Data processing in this embodiment: cleaning and preprocessing of factoring fund information data, deleting duplicate records, processing missing values, removing outliers, unifying formats, etc., to ensure data accuracy and consistency; data conversion, converting data types and data formats, and standardizing data for subsequent analysis and mining; data screening, screening data according to preset conditions or rules to extract valuable information and distinguish and mark sensitive information;

[0054] In data processing, some data may be missing, which may lead to inaccurate analysis results. In this case, deletion, interpolation or filling methods are needed to deal with missing data. In addition, the same data may be entered multiple times, wasting storage space and affecting analysis results. This problem can be solved by deleting, merging or marking duplicate data. When outliers appear, they can be restored to normal by deleting, replacing or correcting them. When data formats differ due to different source channels, the data format is adjusted through conversion and normalization methods. After data cleaning, the data is verified and checked to ensure its accuracy and completeness. Appropriate security measures are taken to protect the security and integrity of the data, such as data encryption and access control.

[0055] The control and supervision methods of the factoring company and its internal employees in this embodiment include: establishing a sound internal audit supervision and risk assessment system, regularly evaluating the effectiveness of the system and policies, equipping full-time legal personnel, strengthening the review of the reasonableness and effectiveness of contract content, terms and performance, strengthening business training, establishing a professional talent team, improving the business skills of handling personnel, standardizing operational behaviors, adopting a risk-based information security management model, following the "CIA model", ensuring the security, integrity and availability of information, establishing an information security management system ISMS, and managing and controlling information security in accordance with international standards.

[0056] S2. Respond to the sensitive data features of the factoring company's employee behavior data, extract the data associated with the sensitive data features through a machine learning algorithm and generate historical associated behavior data, analyze the factoring company's employee behavior trends and directions based on the associated behavior data, and then predict the factoring company's employee behavior trends to generate risk prediction results; specifically, the behavior trend prediction classifies the factoring company's employee sensitive information and the matched historical associated behavior data based on the sensitive data features, generates the time series of the factoring company's employee personal behavior data using the long short-term memory network LSTM time series and constructs a factoring fund information risk value prediction model, and generates corresponding evaluation and risk prediction results based on the factoring fund information risk value prediction model;

[0057] Among them, the use of long short-term memory network LSTM in time series: before processing time series data, the memory cells and hidden states of the LSTM network need to be initialized. Usually, the memory cells are initialized to all zeros, and the hidden states can also be initialized to all zeros or specific initial values; LSTM will process the input data moment by moment in the order of the time series data. At each time step t, the LSTM cell will process the input x at the current moment. t and the hidden state h at the previous moment t-1 , calculate the values ​​of the forget gate, input gate, and output gate at the current moment, and update the cell state and hidden state; LSTM transmits information in the time series through the cell state and hidden state. The cell state is used to store long-term dependencies, while the hidden state contains the output information of the current time step and serves as one of the inputs of the next time step; after processing the entire sequence, the backpropagation algorithm is usually used to update the weight parameters of the LSTM. This process calculates the loss of each time step and minimizes the loss function through optimization algorithms such as gradient descent, so that the network can better adapt to the data.

[0058] It should be noted that machine learning algorithms can use Bayesian networks to extract sensitive data. When extracting sensitive data, Bayesian networks mainly use their probabilistic graphical model capabilities to analyze and infer sensitive information that may be contained in the data. Bayesian networks can represent the dependencies between variables and use probability distributions to describe the relationship between these variables, which makes them very effective in processing complex data and extracting sensitive information. First, it is necessary to build a suitable Bayesian network model based on the characteristics and requirements of the data, and then input the preprocessed data into the Bayesian network model. By building a good Bayesian network model and estimating the parameters, the reasoning ability of the Bayesian network is used to infer the sensitive information in the data.

[0059] To further explain, data processing: is to clean and pre-process the factoring fund information data, delete duplicate records, process missing values, remove outliers, unify the format, etc., to ensure the accuracy and consistency of the data; data conversion, convert data types and data formats, and standardize the data for subsequent analysis and mining; data screening, screen the data according to preset conditions or rules to extract valuable information and distinguish and mark sensitive information;

[0060] In data processing, some data may be missing, which may lead to inaccurate analysis results. In this case, it is necessary to use deletion, interpolation or filling methods to deal with missing data; and the same data may be entered multiple times, wasting storage space and affecting the analysis results. This problem can be solved by using deletion, merging or marking methods to deal with duplicate data; and when outliers appear, they can be restored to normal by using deletion, replacement or correction methods to deal with the outliers. When there are differences in data formats due to different source channels, the data format is adjusted through conversion and normalization methods. After data cleaning, the data is verified and checked to ensure the accuracy and completeness of the data, and to ensure that the data cleaning and processing process complies with personal privacy protection laws and regulations, such as the "Personal Information Protection Law" and the "General Data Protection Regulation", formulate and follow a clear privacy policy, clarify the purpose, method and scope of data collection, use and storage, and adopt appropriate security measures to protect the security and integrity of data, such as data encryption and access control.

[0061] S3. Determine the sensitive data of the factoring company's employees and match it with the factoring fund information data associated with the factoring company's employees, then analyze it through the factoring fund information risk value prediction model, match and determine the factoring fund adjustment plan strategy based on the results of the assessment and risk prediction, and make corresponding adjustments based on the factoring fund adjustment plan strategy.

[0062] In this embodiment, preferably, the factoring fund information data is processed and screened, wherein the factoring fund information data is screened according to preset conditions or rules, specifically, each employee data is used as the starting node of the relationship map, the relationship between each employee and the user is analyzed, and the user who has a direct relationship with the employee is used as the child node of the starting node, and the user who has a direct relationship with each child node user is used as the next node of the child node, and this step is repeated to form a multi-level relationship map until the preset level is reached, and the establishment of the relationship map between each user and the employee is completed.

[0063] It should be noted that the above-mentioned models and algorithms are all available in the prior art. According to the nature of the problem and the data characteristics, select the appropriate prediction model, such as classification algorithm: such as logistic regression, support vector machine, random forest; regression algorithm: such as linear regression, decision tree regression; clustering algorithm: such as K-means or deep learning model: such as recurrent neural network RNN, long short-term memory network LSTM, etc. Use the test data set to evaluate the prediction performance of the model. Common evaluation indicators include accuracy, recall rate, F1 score, AUC-ROC curve, etc. Adjust and optimize the model according to the evaluation results, such as adjusting model parameters, changing model algorithms, adding features, etc.

[0064] In addition, factoring fund information data includes data reflecting fund flow, risk status and user repayment ability, and is divided into user basic information, financial status, business operation, risk control, and company structure, and key data points in each category are extracted.

[0065] In this embodiment, the sensitive information of the factoring company's employees is classified with the matched historical associated behavior data. Specifically, the data is classified and analyzed by applying the K nearest neighbor classification algorithm through transaction features, system access features, communication features, and file transfer features. The K training samples closest to the sample to be classified in the feature space are found according to the transaction features, system access features, communication features, and file transfer features. The distance between the classified sample and the training sample is calculated using a distance metric. At the same time, a voting decision is made based on the category labels of the transaction features, system access features, communication features, and file transfer features, and their average or weighted average is calculated to predict the numerical output of the sample to be predicted, and the classification algorithm class result is obtained. The distance metric between the classified sample and the training sample is calculated as follows:

[0066]

[0067] Dist(x i ,y j ) is the distance measure between the classification sample and the training sample, x ik is the kth classification sample x i ,y jk is the kth training sample y j ;

[0068] It should be noted that the method also includes ensuring the accuracy and consistency of extracting the behavioral data of the factoring company's employees and distinguishing and marking the sensitive information and the corresponding sensitive data feature data, including the following steps:

[0069] Check whether the data meets the requirements according to the reasonable value range and mutual relationship of each variable, and find and correct logically unreasonable or contradictory data;

[0070] Further processing of invalid and missing values: If the preprocessing stage does not fully resolve the problem, this stage further processes the problem through manual comparison.

[0071] It should be emphasized that the steps of constructing the factoring fund information risk value prediction model in this embodiment are:

[0072] Receive factoring fund information risk instructions;

[0073] Obtain the time series of factoring fund information risk categories and factoring company employee personal behavior data from the factoring fund information risk instructions;

[0074] Finding a preset risk prediction model corresponding to the risk category of factoring fund information, and obtaining decision parameters in the preset risk prediction model;

[0075] Collect real-time decision factor data corresponding to decision parameters;

[0076] According to the real-time decision factor data, the risk classification threshold of the preset risk prediction model is adjusted to obtain the real-time factoring fund information risk value prediction model, which is as follows:

[0077]

[0078] in, is the predicted value of the risk value of the factoring fund information of the vth factoring company employee, is the risk assessment value of the tth collection time series of the uth factoring fund information of the vth factoring employee, is the risk assessment judgment threshold of the u-th factoring fund information of the v-th factoring company employee, T is the number of items in the t-th collected time series, and α is the optimization coefficient;

[0079] This embodiment further illustrates that the analysis and generation of the factoring fund information risk assessment value of the factoring company employee is specifically as follows:

[0080] According to the factoring fund information risk categories and decision parameters, a sample data set of factoring fund information risk assessment values ​​of factoring company employees is constructed;

[0081] After dividing the sample data set into a training data set and a validation data set, determining decision parameters of multiple risk categories associated with factoring fund information risks of factoring company employees based on the training data set;

[0082] Through multi-factor binary logistic regression analysis, combined with the training data set, the intercept term and the regression coefficients corresponding to the decision parameters of each risk category in the training data set are obtained, and based on the regression coefficients corresponding to the intercept term and each characteristic parameter; the intercept term and the decision parameters of each risk category in the training data set are simultaneously used as parameters to be verified, and the factoring fund information risk value prediction model is constructed through all the verified parameters to be verified and the regression coefficients corresponding to all the verified parameters to be verified;

[0083] Based on the analysis of the factoring fund information risk value prediction model, the factoring fund information risk assessment value of the factoring company's employees is generated and output to the real-time factoring fund information risk value prediction model;

[0084] The time series of factoring company employees' personal behavior data is input into the real-time factoring fund information risk value prediction model to obtain the assessed risk category and the corresponding risk value prediction value, and the corresponding assessment and risk prediction results are generated according to the assessed risk category and the corresponding risk value prediction value.

[0085] Real-time decision factor data includes: calculation of factoring fund information risk volatility coefficient based on decision parameters;

[0086] Find the data sample period corresponding to the risk volatility coefficient of factoring fund information;

[0087] Get the latest decision factor data updated in real time;

[0088] Real-time decision factor data is extracted from the latest decision factor data according to the data sample period.

[0089] The adjustment plan strategy for factoring funds includes adjustments to business models, initiating entities, credit risk management, trade types, user strategies, financing timing and risk control strategies.

[0090] Based on the above method, this embodiment also provides a system for factoring fund information security management, including a data collection module, a data processing module, a data analysis module, a behavior trend prediction module and a solution strategy adjustment module;

[0091] The data collection module is connected to the data processing module, and the data collection module is used to collect the factoring company's employee data and input the factoring fund information data matching the factoring company's employees. The data processing module is used to process and screen the factoring company's employee data and the factoring fund information data, extract the factoring company's employee behavior data, and distinguish and mark the sensitive information and the corresponding sensitive data features;

[0092] The data analysis module is used to analyze the processing results of the data processing module, respond to the sensitive data characteristics of the factoring company's employee behavior data, and extract data associated with the sensitive data characteristics through a machine learning algorithm and generate historical associated behavior data;

[0093] The behavior trend prediction module is used to analyze the behavior trends and directions of factoring company employees based on the associated behavior data, and then predict the behavior trends of factoring company employees to generate risk prediction results;

[0094] The data collection module includes database, API access, crawler and user input;

[0095] The data processing module includes data acquisition module, data transmission module, data preprocessing module, data storage module, data visualization module, metadata management module, job management and scheduling module, and resource allocation and scheduling module;

[0096] The data analysis module includes a click stream data analysis module, a visit time analysis module, a stay time analysis module and a correlation behavior data analysis module;

[0097] The behavior trend prediction module includes feature selection module, data set selection and division module and training module, model building evaluation module and optimization module, and feedback and iteration module.

[0098] The clickstream data analysis module includes website log analysis, Referrers and Sessionid analysis, as follows:

[0099] Website log analysis:

[0100] Principle: By analyzing website logs, we can obtain records of user visited pages, including the jump relationship between pages, so as to understand the user's click stream data.

[0101] Implementation: Use log analysis tools, such as ELK Stack (Elasticsearch, Logstash, Kibana) or Splunk, to parse and visualize website logs.

[0102] Note: Ensure the completeness and accuracy of the logs to obtain accurate clickstream data.

[0103] Referrers and Sessionid analysis:

[0104] Principle: The Referrers parameter is used to determine the order in which web pages are visited, and the Sessionid is used to uniquely identify each session, thereby depicting the user's click stream data trajectory curve.

[0105] Implementation: Record user access information on the server, including Referrers and Sessionid, and use data analysis tools for statistical analysis.

[0106] Note: Ensure that the server can accurately record user access information and pay attention to the protection of user privacy.

[0107] The access time analysis module includes:

[0108] Server time acquisition:

[0109] Method 1: Use NTP (Network Time Protocol) or SNTP (Simple Network Time Protocol) to communicate with the server and obtain server time data.

[0110] Method 2: Directly read the system time on the server, such as using the GetSystemTime function on Windows systems and the time command or time() function on Unix / Linux systems.

[0111] Implementation: Select an appropriate method to obtain the server time and correlate it with the user's access records for analysis.

[0112] Third-party API access acquisition:

[0113] Principle: Obtain server time data by sending HTTP requests to Internet service providers or open API access platforms.

[0114] Implementation: Select a reliable API access interface and call it to obtain time data when the user accesses the site, and perform correlation analysis with the access records.

[0115] The residence time analysis module includes:

[0116] Server log analysis:

[0117] Principle: By analyzing the user access records in the server log, combined with the page loading time and the timestamp when the user leaves the page, the user's stay time on the page is calculated.

[0118] Implementation: Use log analysis tools to parse server logs, extract user access records and timestamp information, and perform statistical analysis.

[0119] JavaScript Timer:

[0120] Principle: JavaScript code is embedded in the page to record the timestamp when the user opens the page and the timestamp when the user leaves the page. By comparing the two timestamps, the user's stay time on the page is calculated.

[0121] Implementation: Add JavaScript code to the page and ensure it executes correctly when a user visits the page. Collect timestamp data and perform statistical analysis.

[0122] Third-party analytics tools:

[0123] Principle: Use third-party analysis tools, such as Google Analytics, Facebook Pixel, etc., embed codes in the page to record users' access behavior and stay time.

[0124] Implementation: Choose a suitable third-party analysis tool and add the code snippet provided by it to the page. Use the statistical reports provided by the tool to analyze the user's stay time.

[0125] For clickstream data analysis, you can use methods such as website log analysis and Referrers, Sessionid analysis, etc.; for access time analysis, you can use methods such as server time acquisition and third-party API access acquisition; for residence time analysis, you can use methods such as server log analysis, JavaScript timer and third-party analysis tools.

[0126] The acquisition module and data transmission module are the starting points of the data processing process. They are responsible for capturing and receiving data from various data sources, such as sensors, log files, databases, etc. Once the data is received, it will undergo preliminary verification to ensure the integrity of the data and the correctness of the format, and then pass the data to the data preprocessing module, which receives the data from the collection module and cleans, transforms, and integrates it. The cleaning process includes removing duplicate data, handling missing values, correcting errors, etc. to ensure data quality. The transformation process may include standardization, normalization, discretization, etc. of the data to meet the needs of subsequent analysis or processing. The integration process involves merging data from different data sources into a unified data set for subsequent analysis.

[0127] The data storage module is responsible for storing the processed data in a database, data warehouse or other storage media. It is also responsible for data backup, recovery and archiving to ensure data security and reliability. The module also provides data retrieval and access functions so that other modules can easily obtain the required data;

[0128] The job management and scheduling module is responsible for the control and scheduling of the entire data processing process. It arranges the execution order and time of each sub-module according to preset rules and strategies to ensure the orderly progress of data processing tasks. This module is also responsible for monitoring the execution of tasks and providing error handling and recovery mechanisms to ensure the stability and reliability of the data processing process.

[0129] Metadata management: The metadata management module is responsible for managing and maintaining metadata information of data, including data definition, structure, source, quality, etc. It provides contextual information of data to other modules to help them better understand the meaning and purpose of data. Metadata management also involves data quality control and verification to ensure the accuracy and reliability of data.

[0130] Through the collaborative work of these sub-modules, the data processing module can achieve comprehensive and efficient data processing. Each sub-module plays an indispensable role and jointly promotes the progress of the data processing process.

[0131] The behavior trend prediction module includes feature selection and construction module, model selection and training module, model evaluation and optimization module, and feedback and iteration module.

[0132] Model selection and training module:

[0133] Model selection: According to the nature of the problem and the characteristics of the data, select a suitable prediction model, such as a classification algorithm, such as logistic regression, support vector machine, random forest; regression algorithm: such as linear regression, decision tree regression; clustering algorithm: such as K-means or deep learning models such as recurrent neural network RNN, long short-term memory network LSTM, etc. In this embodiment, the time series of the personal behavior data of the factoring company employees is generated by the time series of the long short-term memory network LSTM.

[0134] Model training: Use data associated with sensitive data features and generate historical associated behavior data to train the model and adjust model parameters to optimize prediction performance.

[0135] Model evaluation and optimization module:

[0136] Model evaluation uses data associated with sensitive data features and generates historical associated behavior data to evaluate the predictive performance of the model. Common evaluation indicators include accuracy, recall rate, F1 score, AUC-ROC curve, etc.

[0137] Model optimization: adjust and optimize the factoring fund information risk value prediction model based on the evaluation results.

[0138] Behavior trend prediction module:

[0139] Real-time prediction: deploy the trained factoring fund information risk value prediction model to the production environment, predict the real-time data, and generate prediction results of behavioral trends.

[0140] Prediction result visualization module: Visualize the risk prediction results in the form of charts, images, etc., to help users more intuitively understand the assessment risk category and the corresponding risk value prediction value to generate the corresponding assessment and risk prediction results.

[0141] Feedback and Iteration Module:

[0142] Collect feedback, collect users' feedback and opinions on the prediction results, as well as the actual operating data of the factoring funds information risk value prediction model.

[0143] Model iteration: Based on feedback and new data, the factoring fund information risk value prediction model is iterated and optimized to improve prediction accuracy and performance.

[0144] Working principle: Collect factoring company employee data and input factoring fund information data that matches factoring company employees, process and screen factoring fund information data, extract factoring company employee behavior data and distinguish and mark sensitive information and corresponding sensitive data features; respond to factoring company employees' sensitive data features, and extract data associated with sensitive data features through machine learning algorithms and generate historical associated behavior data, analyze employee behavior trends and directions based on associated behavior data, and predict factoring company employees' behavior trends; Specifically, behavioral trend prediction involves classifying factoring company employees' sensitive information with the matched historical associated behavior data, using the time series of the long short-term memory network LSTM to generate a time series of factoring company employees' personal behavior data and construct a factoring fund information risk value prediction model to generate corresponding evaluation and risk prediction results; match the factoring company employees' adjustment plan strategy with the factoring fund adjustment plan strategy based on the evaluation and risk prediction results, and make corresponding adjustments.

[0145] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for managing factoring fund information security, characterized in that: The following steps are involved: Collecting factoring company employee data and inputting factoring fund information data matching the factoring company employee, and processing and screening the factoring company employee data and factoring fund information data, extracting factoring company employee behavior data and distinguishing and marking sensitive information and corresponding sensitive data features; Responding to the sensitive data features of the factoring company's employee behavior data, extracting data associated with the sensitive data features through a machine learning algorithm and generating historical associated behavior data, analyzing the factoring company's employee behavior trends and directions based on the associated behavior data, and then predicting the factoring company's employee behavior trends to generate risk prediction results; wherein the behavioral trend prediction specifically, after classifying the factoring company's employee sensitive information and the matched historical associated behavior data based on the sensitive data features, using the time series of the long short-term memory network LSTM to generate the time series of the factoring company's employee personal behavior data and construct a factoring fund information risk value prediction model, and generating corresponding evaluation and risk prediction results based on the factoring fund information risk value prediction model; After determining the sensitive data of the factoring company's employees and matching it with the factoring fund information data associated with the factoring company's employees, the factoring fund information risk value prediction model is used for analysis. According to the results of the evaluation and risk prediction, the adjustment plan strategy for the factoring funds is matched and determined, and corresponding adjustments are made according to the adjustment plan strategy for the factoring funds.

2. The factoring fund information security management method according to claim 1, characterized in that: The factoring company employee data and factoring fund information data are processed and screened, wherein the factoring fund information data is screened according to preset conditions or rules. Specifically, each factoring company employee data is used as the starting node of the relationship map, and the relationship between each factoring company employee and the user is analyzed. The user who has a direct relationship with the factoring company employee is used as the child node of the starting node, and the user who has a direct relationship with each child node user is used as the next node of the child node. Repeat this step to form a multi-level relationship map until the preset level is reached, and the establishment of the relationship map between each user and the employee is completed.

3. The factoring fund information security management method according to claim 1 or 2, characterized in that: The factoring fund information data includes data reflecting fund flow, risk status and user repayment ability, and is divided into user basic information category, financial status category, business operation category, risk control category, and company structure category, and key data points in each category are extracted.

4. The factoring fund information security management method according to claim 1, characterized in that: The sensitive information of the factoring company employees is classified with the matched historical associated behavior data. Specifically, the data is classified and analyzed by applying the K nearest neighbor classification algorithm through transaction features, system access features, communication features, and file transfer features. The K training samples closest to the sample to be classified in the feature space are found according to the transaction features, system access features, communication features, and file transfer features. The distance between the classified sample and the training sample is calculated using a distance metric. At the same time, a voting decision is made based on the category labels of the transaction features, system access features, communication features, and file transfer features, and their average or weighted average is calculated to predict the numerical output of the sample to be predicted, and the classification algorithm class result is obtained. The distance metric between the classified sample and the training sample is calculated as follows: Dist(x i ,y j ) is the distance measure between the classification sample and the training sample, x ik is the kth classification sample x i ,y jk is the kth training sample y j .

5. The factoring fund information security management method according to claim 1, characterized in that: The method also includes a method for ensuring the accuracy and consistency of extracting behavioral data of factoring company employees and distinguishing and marking sensitive information and corresponding sensitive data feature data, including the following steps: Check whether the data meets the requirements according to the reasonable value range and mutual relationship of each variable, and find and correct logically unreasonable or contradictory data; Further processing of invalid and missing values: If the preprocessing stage does not fully resolve the problem, this stage further processes the problem through manual comparison.

6. The factoring fund information security management method according to claim 1, characterized in that: The steps of constructing the factoring fund information risk value prediction model are: Receive factoring fund information risk instructions; Obtaining the time series of factoring fund information risk categories and factoring company employee personal behavior data from the factoring fund information risk instruction; Searching for a preset risk prediction model corresponding to the risk category of the factoring fund information, and obtaining decision parameters in the preset risk prediction model; Collecting real-time decision factor data corresponding to the decision parameters; The risk classification threshold of the preset risk prediction model is adjusted according to the real-time decision factor data to obtain a real-time factoring fund information risk value prediction model, which is specifically as follows: in, is the predicted value of the risk value of the factoring fund information of the vth factoring company employee, is the risk assessment value of the tth collection time series of the uth factoring fund information of the vth factoring employee, is the risk assessment judgment threshold of the u-th factoring fund information of the v-th factoring company employee, T is the number of items in the t-th collected time series, and α is the optimization coefficient; The time series of the factoring company's employee personal behavior data is input into the real-time factoring fund information risk value prediction model to obtain the assessed risk category and the corresponding risk value prediction value, and the corresponding assessment and risk prediction results are generated according to the assessed risk category and the corresponding risk value prediction value.

7. The factoring fund information security management method according to claim 6, characterized in that: The real-time decision factor data includes: calculating the factoring fund information risk volatility coefficient according to the decision parameters; Finding a data sample period corresponding to the risk volatility coefficient of the factoring fund information; Get the latest decision factor data updated in real time; Real-time decision factor data is extracted from the latest decision factor data according to the data sample period.

8. The factoring fund information security management method according to claim 6, characterized in that: The analysis and generation of the factoring fund information risk assessment value of the factoring company’s employees is as follows: Constructing a sample data set of factoring fund information risk assessment values ​​of factoring company employees according to the factoring fund information risk categories and decision parameters; After dividing the sample data set into a training data set and a validation data set, determining, based on the training data set, a plurality of decision parameters of risk categories associated with factoring fund information risks of factoring company employees; By using a multi-factor binary logistic regression analysis method, in combination with the training data set, the regression coefficients corresponding to the intercept term and the decision parameters of each risk category in the training data set are obtained, and according to the regression coefficients corresponding to the intercept term and each characteristic parameter; Synchronously taking the intercept term and the decision parameters of each risk category in the training data set as parameters to be verified, and constructing a factoring fund information risk value prediction model by verifying all the parameters to be verified and the regression coefficients corresponding to all the parameters to be verified; Based on the analysis of the factoring fund information risk value prediction model, a factoring fund information risk assessment value of the factoring company's employees is generated and output to the real-time factoring fund information risk value prediction model.

9. The factoring fund information security management method according to claim 1, characterized in that: The adjustment plan strategy of the factoring funds includes adjustment of business model, adjustment of initiating entity, credit risk management, adjustment of trade type, adjustment of user strategy, adjustment of financing timing and adjustment of risk control strategy.

10. A system for factoring fund information security management applied to any one of claims 1 to 9, characterized in that: It includes data collection module, data processing module, data analysis module, behavior trend prediction module and program strategy adjustment module; The data collection module is connected to the data processing module, and the data collection module is used to collect factoring company employee data and input factoring fund information data matching the factoring company employee. The data processing module is used to process and screen the factoring company employee data and factoring fund information data, extract the factoring company employee behavior data, and distinguish and mark sensitive information and corresponding sensitive data features; The data analysis module is used to analyze the processing results of the data processing module, respond to the sensitive data features of the factoring company's employee behavior data, and extract data associated with the sensitive data features through a machine learning algorithm and generate historical associated behavior data; The behavior trend prediction module is used to analyze the behavior trend and direction of the factoring company's employees according to the associated behavior data, and then predict the behavior trend of the factoring company's employees to generate a risk prediction result; The data collection module includes database, API access, crawler and user input; The data processing module includes a data acquisition module, a data transmission module, a data preprocessing module, a data storage module, a data visualization module, a metadata management module, a job management and scheduling module, and a resource allocation and scheduling module; The data analysis module includes a click stream data analysis module, a visit time analysis module, a stay time analysis module and a related behavior data analysis module; The behavior trend prediction module includes a feature selection module, a data set selection and division module and a training module, a model building evaluation module and an optimization module, and a feedback and iteration module.

Citation Information

Patent Citations

  • Full-period management method and system for factoring service

    CN117036009A

  • Bank anti-call fraud data model construction method based on multi-feature fusion

    CN117993919A

  • System and method for blockchain transaction risk management using machine learning

    US20220067752A1