Supply provider risk assessment model training method and device and computer equipment

Abstract

The invention relates to a supply provider risk assessment model training method and device and computer equipment. The method comprises the steps of obtaining data of a plurality of sample supply providers, obtaining a plurality of pieces of risk index feature data according to the data, inputting the risk index feature data into an index weight construction module in a to-be-trained risk assessment model to obtain a corresponding index weight, inputting each risk index feature data and each index weight into a logistic regression module in a to-be-trained risk assessment model to obtain logistic regression values of a plurality of to-be-assessed parameters, and inputting the risk index feature data and the logistic regression values into a risk prediction module in the to-be-trained risk assessment model to obtain corresponding prediction risk data; and constructing a corresponding loss function according to the predicted risk data and the logistic regression value, and adjusting a neuron weight, a threshold value and a to-be-estimated parameter in the risk prediction module until the loss function obtains a minimum value, thereby obtaining a risk assessment model. According to the method, the evaluation accuracy of the trained model is improved.

Description

Technical Field

[0001] The present application relates to the technical field of risk assessment, and in particular to a training method, apparatus, computer equipment, computer-readable storage medium and computer program product for a supply provider risk assessment model. Background Art

[0002] In the global economic environment, supply chain management has become an important part of the core competitiveness of enterprises. As a key link in the supply chain, the stability and reliability of logistics supply providers directly affect the efficiency and stability of the entire supply chain. Therefore, assessing and managing the risks of suppliers is an important part of supply chain management.

[0003] Traditional supply providers' performance risk analysis decisions rely on subjective experience and manual judgment, and business personnel assess risks based on limited historical data and personal experience.

[0004] However, the current risk assessment methods that rely on subjective experience and manual judgment have the problem of low assessment accuracy of the risk assessment model. Summary of the invention

[0005] Based on this, it is necessary to provide a training method, device, computer equipment, computer-readable storage medium and computer program product for a supply provider risk assessment model that can improve the assessment accuracy in response to the above technical problems.

[0006] In a first aspect, the present application provides a method for training a supply provider risk assessment model, comprising:

[0007] Obtain full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data;

[0008] Input each risk indicator characteristic data into the indicator weight construction module in the risk assessment model of the supply provider to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module;

[0009] Inputting the characteristic data of each risk indicator and the weight of each indicator into the logistic regression module in the risk assessment model of the supply provider to be trained, and obtaining the logistic regression value associated with the risk assessment model of the supply provider to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated;

[0010] Input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained based on the predicted risk data and the logistic regression value;

[0011] The pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module are adjusted until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

[0012] In one embodiment, the indicator weight corresponding to each risk indicator characteristic data is obtained through the indicator weight construction module, including:

[0013] The information entropy corresponding to each risk indicator characteristic data is obtained through the indicator weight construction module, and the indicator weight corresponding to each risk indicator characteristic data is obtained according to each information entropy.

[0014] In one embodiment, the full life cycle data includes actual risk data associated with the sample supply provider;

[0015] After the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model, the following steps are included:

[0016] Obtaining the accuracy of the model performance of the risk assessment model for evaluating the supply provider based on the predicted risk data and the actual risk data;

[0017] Based on the accuracy analysis results, the performance of the supply provider risk assessment model is optimized.

[0018] In one embodiment, based on the accuracy analysis results, the performance of the supply provider risk assessment model is optimized, including:

[0019] When the analysis result indicates that the prediction accuracy of the supply provider risk assessment model does not reach the set target value, obtaining a regularization parameter pre-set for the supply provider risk assessment model;

[0020] According to the regularization parameter and the parameter to be estimated, the loss function of the supply provider risk assessment model is updated.

[0021] In an exemplary embodiment, the regularization parameter includes at least one of an L1 regularization parameter and an L2 regularization parameter;

[0022] According to the regularization parameter and the parameter to be estimated, the loss function of the risk assessment model of the supply provider is updated, including:

[0023] Based on the L1 regularization parameter and the parameter to be estimated, the absolute value and term of the parameter are obtained, and the loss function is updated according to the absolute value and term of the parameter;

[0024] and / or

[0025] According to the L2 regularization parameter and the parameter to be estimated, the square sum of the parameters is obtained, and the loss function is updated based on the square sum of the parameters.

[0026] In one embodiment, full life cycle data of multiple sample supply providers is obtained, including:

[0027] Obtain original full life cycle data from multiple sample supply providers;

[0028] Clean the original full life cycle data, use data filtering algorithms to remove noise and abnormal data, and correct illegal time data to obtain the cleaned original full life cycle data;

[0029] Data is classified according to the data type and business link of the cleaned original life cycle data to obtain full life cycle data.

[0030] In a second aspect, the present application also provides a training device for a supply provider risk assessment model, comprising:

[0031] A data construction module is used to obtain the full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data;

[0032] A weight construction module is used to input each risk indicator characteristic data into the indicator weight construction module in the risk assessment model of the supply provider to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module;

[0033] A logistic regression value construction module is used to input the characteristic data of each risk indicator and the weight of each indicator into the logistic regression module in the risk assessment model of the supply provider to be trained, and obtain the logistic regression value associated with the risk assessment model of the supply provider to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated;

[0034] A loss function construction module is used to input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained according to the predicted risk data and the logistic regression value;

[0035] The parameter adjustment module is used to adjust the pre-set neuron weights, thresholds and estimated parameters in the risk prediction module until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

[0036] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0037] Obtain full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data;

[0038] Input each risk indicator characteristic data into the indicator weight construction module in the risk assessment model of the supply provider to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module;

[0039] Inputting the characteristic data of each risk indicator and the weight of each indicator into the logistic regression module in the risk assessment model of the supply provider to be trained, and obtaining the logistic regression value associated with the risk assessment model of the supply provider to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated;

[0040] Input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained based on the predicted risk data and the logistic regression value;

[0041] The pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module are adjusted until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

[0042] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:

[0043] Obtain full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data;

[0044] Input each risk indicator characteristic data into the indicator weight construction module in the risk assessment model of the supply provider to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module;

[0045] Inputting the characteristic data of each risk indicator and the weight of each indicator into the logistic regression module in the risk assessment model of the supply provider to be trained, and obtaining the logistic regression value associated with the risk assessment model of the supply provider to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated;

[0046] Input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained based on the predicted risk data and the logistic regression value;

[0047] The pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module are adjusted until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

[0048] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0049] Obtain full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data;

[0050] Input each risk indicator characteristic data into the indicator weight construction module in the risk assessment model of the supply provider to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module;

[0051] Inputting the characteristic data of each risk indicator and the weight of each indicator into the logistic regression module in the risk assessment model of the supply provider to be trained, and obtaining the logistic regression value associated with the risk assessment model of the supply provider to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated;

[0052] Input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained based on the predicted risk data and the logistic regression value;

[0053] The pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module are adjusted until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

[0054] The training method, apparatus, computer equipment, computer-readable storage medium and computer program product of the above-mentioned supply provider risk assessment model obtain the full life cycle data of multiple sample supply providers, obtain multiple risk indicator feature data of each sample supply provider based on each life cycle data, input each risk indicator feature data into the indicator weight construction module in the supply provider risk assessment model to be trained, obtain the indicator weight corresponding to each risk indicator feature, and then input each risk indicator feature data and each indicator weight into the logistic regression module in the supply provider risk assessment model to be trained, obtain the logistic regression value containing multiple parameters to be estimated associated with the supply provider risk assessment model to be trained, input the risk indicator feature data and the logistic regression value into the risk prediction module in the supply provider risk assessment model to be trained, obtain the corresponding predicted risk data, and construct a corresponding loss function based on the predicted risk data and the logistic regression value, adjust the pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module until the loss function reaches the minimum value, and use the supply provider risk assessment model to be trained as the trained supply provider risk assessment model. The reliability of model training is improved by comprehensively considering multi-dimensional risk indicator characteristic data. The interpretability of logistic regression and the powerful fitting ability of neural networks are utilized to improve the risk assessment accuracy of the trained model, while also improving the effectiveness and reliability of the model in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 A diagram of an application environment for a training method for a supply provider risk assessment model in one embodiment;

[0057] Figure 2 A flowchart of a method for training a supply provider risk assessment model in one embodiment;

[0058] Figure 3 A flowchart of a method for training a supply provider risk assessment model in another embodiment;

[0059] Figure 4 A structural block diagram of a training device for a supply provider risk assessment model in one embodiment;

[0060] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] The training method of the supply provider risk assessment model provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the terminal 102 communicates with the server 104 through a network. The data storage system can store data that the server 104 needs to process. The data storage system can be integrated on the server 104, or placed on the cloud or other network servers. The server 104 obtains the full life cycle data of multiple sample supply providers from the terminal, and obtains multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data, and inputs each risk indicator characteristic data into the indicator weight construction module in the risk assessment model of the supply provider to be trained, and obtains the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module, and then inputs each risk indicator characteristic data and each indicator weight into the logistic regression module in the risk assessment model of the supply provider to be trained, and obtains the logistic regression value containing multiple parameters to be estimated and associated with the risk assessment model of the supply provider to be trained through the logistic regression module, and the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and finally inputs the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, and obtains the predicted risk data corresponding to each sample supply provider through the risk prediction module, and constructs the loss function corresponding to the risk assessment model of the supply provider to be trained based on the predicted risk data and the logistic regression value, and finally adjusts the pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction model until the loss function obtains the minimum value, and the risk assessment model of the supply provider to be trained is used as the trained supply provider risk assessment model. The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. Head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0063] In an exemplary embodiment, Figure 2As shown, a training method for a supply provider risk assessment model is provided, and the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate the method, which includes the following steps S201 to S205. Among them:

[0064] Step S201, obtain the full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each full life cycle data.

[0065] Among them, the full life cycle data includes but is not limited to qualification registration data (company registration time, registered capital, business scope, etc.), bidding and procurement data (number of bids, winning rate, winning amount, etc.), contract performance data (contract signing time, delivery time, delivery quantity, product quality acceptance, etc.), quality control data (material sampling pass rate, problem records during supervision, etc.), operation and maintenance data (failure rate after equipment is put into operation, number of repairs, etc.). At the same time, collect external multi-source data, such as supplier rankings published by industry associations, punishment records of market supervision departments, user evaluations on the Internet, and feedback from other companies in the same industry on the cooperation of the supplier. Risk indicator feature data can be understood as key features related to the performance risk of the supplier, such as extracting features such as the average number of delivery delay days and the proportion of batches with quality problems from contract performance data, and extracting features such as the rate of positive reviews and keywords for negative reviews from external user reviews.

[0066] Optionally, the server 104 obtains the full life cycle data of multiple sample supply providers, and the full life cycle data represents multi-source data from different sources at different times. The server 104 obtains multiple risk indicator characteristic data of each sample supply provider based on the full life cycle data. By acquiring a large amount of multi-source data to construct risk indicator characteristic data, the data validity and data generality of the risk indicator characteristic data are improved, thereby improving the evaluation accuracy of the model.

[0067] Step S202, input each risk indicator characteristic data into the indicator weight construction module in the supply provider risk assessment model to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module.

[0068] Among them, the risk assessment model for the supply provider to be trained can be understood as a risk prediction model that includes a logistic regression model and a neural network model, and the indicator weight can be understood as the proportion of the current risk indicator feature data in the risk data of the supply provider, that is, the relative importance.

[0069] Exemplarily, the server 104 inputs each risk indicator feature data into the indicator weight construction module in the supply provider risk assessment model to be trained, and uses the entropy weight method to assign weights to each risk indicator feature data. The entropy weight method can automatically calculate the weight of each risk indicator based on the distribution characteristics of the indicator data. This method does not rely on subjective judgment, helps to eliminate the influence of human factors, and thus improves the objectivity and scientificity of feature weight assignment.

[0070] Step S203, input each risk indicator characteristic data and each indicator weight into the logistic regression module in the supply provider risk assessment model to be trained, and obtain the logistic regression value associated with the supply provider risk assessment model to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated.

[0071] Optionally, let the supplier's performance risk be the dependent variable y (with a value of 0 for low risk and 1 for high risk), and the extracted characteristic variable be The expression of the logistic regression model is: ,in is the parameter to be estimated.

[0072] Estimate the parameters using maximum likelihood estimation , the likelihood function is: .

[0073] Take the logarithm of the likelihood function and take its derivative to get the derivative of the log-likelihood function, and then use optimization methods such as the gradient descent algorithm to iteratively solve the parameters , so that the log-likelihood function is maximized. The logistic regression value obtained in the above way lays the foundation for the subsequent construction of the loss function. At the same time, the use of the logistic regression algorithm improves the prediction ability of the model.

[0074] Step S204, input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained based on the predicted risk data and the logistic regression value.

[0075] Step S205, adjust the pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module until the loss function reaches a minimum value, and use the supply provider risk assessment model to be trained as the trained supply provider risk assessment model.

[0076] Optionally, a neural network is constructed that includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is the number of feature variables m, and the number of nodes in the output layer is 1 (indicating supplier performance risk).

[0077] The number of hidden layers and nodes are adjusted according to the actual situation. One hidden layer is set with the number of nodes being k.

[0078] The activation function of the neural network can choose a commonly used function, such as the ReLU (Rectified Linear Unit) function .

[0079] During the training process, the back propagation algorithm is used to calculate the gradient of the loss function to the parameters of each layer, and the parameters are updated by the gradient descent algorithm. The loss function can be the cross entropy loss function:

[0080]

[0081] The preprocessed data is divided into a training set and a test set, and the ratio can be set to a:b or c:d, etc. The model is trained using the training set. During the training process, the algorithm parameters (coefficients in logistic regression) are adjusted. , neuron weights and thresholds in the neural network, etc.) to minimize the loss function value of the model on the training set to achieve a better fitting effect. At the same time, cross-validation techniques such as k-fold cross-validation (k can be n (n is an integer greater than 0)) are used to ensure the stability and generalization ability of the model. After multiple iterations of training, the accuracy of the model on the test set reaches the expected level, and the accuracy reaches more than X%. By constructing a neural network for supplier performance risk assessment, combined with the application of back propagation and cross entropy loss functions, the model's complex feature learning ability and generalization ability can be effectively improved, and the model training form of cross-validation combined with training sets and test sets is used to improve the assessment accuracy of risk assessment of the trained model.

[0082] In the training method of the above-mentioned supply provider risk assessment model, the full life cycle data of multiple sample supply providers are obtained, and multiple risk indicator feature data of each sample supply provider are obtained based on each life cycle data. Each risk indicator feature data is input into the indicator weight construction module in the supply provider risk assessment model to be trained to obtain the indicator weight corresponding to each risk indicator feature, and then each risk indicator feature data and each indicator weight are input into the logistic regression module in the supply provider risk assessment model to be trained to obtain the logistic regression value containing multiple parameters to be estimated associated with the supply provider risk assessment model to be trained, and the risk indicator feature data and the logistic regression value are input into the risk prediction module in the supply provider risk assessment model to be trained to obtain the corresponding predicted risk data, and a corresponding loss function is constructed based on the predicted risk data and the logistic regression value, and the pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module are adjusted until the loss function obtains the minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model. The reliability of model training is improved by comprehensively considering multi-dimensional risk indicator characteristic data. The interpretability of logistic regression and the powerful fitting ability of neural networks are utilized to improve the risk assessment accuracy of the trained model, while also improving the effectiveness and reliability of the model in practical applications.

[0083] In one embodiment, the indicator weight corresponding to each risk indicator characteristic data is obtained through the indicator weight construction module, including: obtaining the information entropy corresponding to each risk indicator characteristic data through the indicator weight construction module, and obtaining the indicator weight corresponding to each risk indicator characteristic data according to each information entropy.

[0084] For example, the entropy weight method is used to determine the indicator weight: Assume that there are m risk indicator feature data of the supplier and the number of samples is n. First, the indicator data is standardized. Assume that the jth indicator value of the i-th sample is The standardized index value is For positive indicators (the larger the indicator value, the better), ; For negative indicators (the smaller the indicator value, the better), ,in and are the minimum and maximum values ​​of the jth indicator respectively. Calculate the information entropy of the jth indicator ,in .

[0085] Calculate the weight of the jth indicator .

[0086] The entropy weight method uses objective data to calculate weights, eliminates the influence of subjective factors, ensures that the weights of each indicator are more fair and reasonable, and can adapt to different types of indicators, including positive and negative indicators, and enhances versatility. The weights determined by the entropy weight method reflect the amount of information contained in each indicator, ensure the scientificity and effectiveness of the weights, and thus improve the scientificity of the loss function.

[0087] In one of the embodiments, the full life cycle data includes actual risk data associated with sample supply providers; after the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model, it includes: obtaining the accuracy of the model performance used to evaluate the supply provider risk assessment model based on the predicted risk data and the actual risk data; and optimizing the performance of the supply provider risk assessment model based on the analysis results of the accuracy.

[0088] Among them, the accuracy rate can be understood as the ratio of predicted risk data to actual risk data.

[0089] Optionally, the server 104 uses the same amount of data as the predicted risk data and the actual risk data, and compares it with the amount of actual risk data to obtain the accuracy of the model performance of the supply provider risk assessment model, and finally optimizes the performance of the supply provider risk assessment model based on the analysis results of the accuracy. By obtaining the indicator data for evaluating the performance of the model based on the predicted risk data and the actual risk data, the prediction effect of the supply provider risk assessment model can be more comprehensively understood, and the model can be adjusted according to the analysis results to further improve the prediction ability of the model.

[0090] In an exemplary embodiment, based on the accuracy analysis results, the performance of the supply provider risk assessment model is optimized, including: when the analysis results indicate that the prediction accuracy of the supply provider risk assessment model does not reach the set target value, obtaining the regularization parameters pre-set for the supply provider risk assessment model; and updating the loss function of the supply provider risk assessment model based on the regularization parameters and the parameters to be estimated.

[0091] Among them, the regularization parameter can be understood as a parameter used in regularization to prevent overfitting of the model, adding a penalty term to the loss function, and a prefix of the penalty term.

[0092] For example, when the analysis results indicate that the prediction accuracy of the supplier risk assessment model has not reached the set target value, for example, the accuracy on the training set is very high, but the accuracy on the test set is low, then regularization technology can be used or the amount of training data can be increased. The use of regularization technology is to construct a corresponding penalty term based on the regularization parameter and the parameter to be estimated, and add the penalty term to the loss function of the supplier risk assessment model to form a new loss function. By adding the penalty term to the loss function, the regularization technology controls the size of the model parameters, prevents the model from becoming too complex due to the noise in the training data, and thus reduces the risk of overfitting. After the introduction of the regularization term, the model is more able to adapt to new data and improve the prediction accuracy of the test set.

[0093] In one embodiment, the regularization parameter includes at least one of an L1 regularization parameter and an L2 regularization parameter; based on the regularization parameter and the parameter to be estimated, the loss function of the supply provider risk assessment model is updated, including: based on the L1 regularization parameter and the parameter to be estimated, obtaining the absolute value and term of the parameter, and updating the loss function based on the absolute value and term of the parameter; and / or, based on the L2 regularization parameter and the parameter to be estimated, obtaining the square sum term of the parameter, and updating the loss function based on the square sum term of the parameter.

[0094] Among them, the L1 regularization parameter can be understood as a parameter used in L1 regularization to prevent model overfitting, adding an absolute value penalty term to the loss function, and a parameter prefixed with the absolute value penalty term; the L2 regularization parameter is similar, which can be understood as a parameter used in L2 regularization to prevent model overfitting, adding a square sum penalty term to the loss function, and a parameter prefixed with the square sum penalty term.

[0095] Optionally, the regularization parameter includes at least one of an L1 regularization parameter and an L2 regularization parameter, and the server 104 calculates the L1 regularization parameter according to the L1 regularization parameter. and the parameters to be estimated , get the absolute value and term of the parameter , and will Add to the loss function to form a new loss function Or, the server 104 uses the L2 regularization parameter and the parameters to be estimated , get the sum of squares of the parameters ,Will Add to the loss function to form a new loss function: , when both L1 regularization parameter and L2 regularization parameter are included, and Add to the loss function to form a new loss function:

[0096]

[0097] Introducing L1 or L2 regularization penalty terms alone can speed up the update of the loss function and ensure the training speed of the model. Introducing both L1 and L2 regularization penalty terms can more effectively control the complexity of the model. L1 regularization can select features, while L2 regularization provides a smooth weight distribution, which promotes more robust model training.

[0098] In one of the embodiments, full life cycle data of multiple sample supply providers are obtained, including: obtaining the original full life cycle data of multiple sample supply providers; performing data cleaning on the original full life cycle data, using a data filtering algorithm to remove noise data and abnormal data therein, and correcting illegal time data therein to obtain the cleaned original full life cycle data; performing data classification according to the data type and business link of the cleaned original life cycle data to obtain the full life cycle data.

[0099] For example, the data of the supplier's entire life cycle is obtained from the unified data management platform, including but not limited to qualification registration data (company registration time, registered capital, business scope, etc.), bidding and procurement data (number of bids, winning rate, winning amount, etc.), contract performance data (contract signing time, delivery time, delivery quantity, product quality acceptance, etc.), quality control data (material sampling pass rate, problem records during supervision, etc.), operation and maintenance data (failure rate after equipment is put into operation, number of repairs, etc.). At the same time, external multi-source data is collected, such as supplier rankings published by industry associations, punishment records of market supervision departments, user reviews on the Internet, and feedback from other companies in the same industry on the cooperation with the supplier.

[0100] The collected data is preprocessed, including data cleaning, data classification and feature extraction. During the data cleaning process, data filtering algorithms are used to remove noise data and outliers, and data that obviously deviates from the normal delivery time range is corrected or marked. Data classification is divided into different categories according to data type and business link, and qualification registration data is classified into one category, contract performance data is classified into another category, etc. In the feature extraction stage, key features related to supplier performance risks are extracted from various types of data, and features such as the average number of delivery delay days and the proportion of batches with quality problems are extracted from contract performance data. Features such as the rate of positive reviews and negative review keywords are extracted from external user reviews.

[0101] By clearing and classifying the original full life cycle data, the full life cycle data obtained in this way has higher data validity and availability, thereby improving the training effect of model training.

[0102] In an exemplary embodiment, Figure 3As shown, a specific implementation method of a training method for a supply provider risk assessment model is provided (the specific numerical values ​​appearing below do not limit the application to be implemented only in this case, but are a specific example of implementing the solution of the application), as follows:

[0103] Construct a supplier risk assessment model (i.e. the aforementioned supplier risk assessment model):

[0104] 1. Data collection and preprocessing:

[0105] The unified data management platform is used to obtain data on the supplier's entire life cycle, including but not limited to qualification registration data (company registration time, registered capital, business scope, etc.), bidding and procurement data (number of bids, winning rate, winning amount, etc.), contract performance data (contract signing time, delivery time, delivery quantity, product quality acceptance, etc.), quality control data (material sampling pass rate, problem records during supervision, etc.), and operation and maintenance data (failure rate and number of repairs after equipment is put into operation). At the same time, external multi-source data is collected, such as supplier rankings published by industry associations, penalty records of market supervision departments, user reviews on the Internet, and feedback from other companies in the same industry on the cooperation with the supplier.

[0106] The collected data is preprocessed, including data cleaning, data classification and feature extraction. During the data cleaning process, data filtering algorithms are used to remove noise data and outliers, and data that obviously deviates from the normal delivery time range is corrected or marked. Data classification is divided into different categories according to data type and business link, and qualification registration data is classified into one category, contract performance data is classified into another category, etc. In the feature extraction stage, key features related to supplier performance risks are extracted from various types of data, and features such as the average number of delivery delay days and the proportion of batches with quality problems are extracted from contract performance data. Features such as the rate of positive reviews and negative review keywords are extracted from external user reviews.

[0107] 2. Model construction and training:

[0108] The supplier risk assessment model is constructed by combining logistic regression, neural network algorithm and entropy weight method. The specific algorithm is as follows:

[0109] Entropy weight method determines the indicator weight:

[0110] Assume that there are m extracted supplier risk assessment characteristic indicators and the number of samples is n. First, the indicator data is standardized. Assume that the jth indicator value of the i-th sample is The standardized index value is For positive indicators (the larger the indicator value, the better), ; For negative indicators (the smaller the indicator value, the better), ,in and are the minimum and maximum values ​​of the jth indicator respectively. Calculate the information entropy of the jth indicator ,in .

[0111] Calculate the weight of the jth indicator .

[0112] Logistic Regression Algorithm:

[0113] Assume that the supplier performance risk is the dependent variable y (the value is 0 for low risk and 1 for high risk), and the extracted characteristic variables are The expression of the logistic regression model is: ,in is the parameter to be estimated.

[0114] Estimate the parameters using maximum likelihood estimation , the likelihood function is: .

[0115] Take the logarithm of the likelihood function and take its derivative to get the derivative of the log-likelihood function, and then use optimization methods such as the gradient descent algorithm to iteratively solve the parameters , so that the log-likelihood function is maximized.

[0116] Neural Network Algorithms:

[0117] Construct a neural network consisting of an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is the number of feature variables m, and the number of nodes in the output layer is 1 (indicating supplier performance risk).

[0118] The number of hidden layers and nodes are adjusted according to the actual situation. One hidden layer is set with the number of nodes being k.

[0119] The activation function of the neural network can choose a commonly used function, such as the ReLU function .

[0120] During the training process, the back propagation algorithm is used to calculate the gradient of the loss function to the parameters of each layer, and the parameters are updated by the gradient descent algorithm. The loss function can be the cross entropy loss function:

[0121]

[0122] The preprocessed data is divided into a training set and a test set, and the ratio can be set to 7:3 or 8:2, etc. The model is trained using the training set. During the training process, the algorithm parameters (coefficients in logistic regression) are adjusted. , neuron weights and thresholds in the neural network, etc.) to minimize the loss function value of the model on the training set to achieve a better fitting effect. At the same time, cross-validation techniques such as k-fold cross-validation (k can be 5 or 10, etc.) are used to ensure the stability and generalization ability of the model. After multiple iterations of training, the accuracy of the model on the test set reaches the expected level, with an accuracy of more than 85%.

[0123] 3. Model evaluation and optimization:

[0124] The trained model is evaluated using a variety of evaluation indicators, including accuracy, recall, F1 value, mean square error, etc. Accuracy is used to measure the proportion of samples correctly predicted by the model to the total samples, recall reflects the proportion of actual defaulting suppliers correctly predicted by the model as defaulting, F1 value comprehensively considers accuracy and recall, and mean square error is used to evaluate the deviation between the model's predicted value and the actual value. By analyzing these evaluation indicators, we can fully understand the performance of the model.

[0125] Optimize the model based on the evaluation results. If the model has overfitting (e.g., the accuracy is high on the training set, but low on the test set), regularization techniques (L1 regularization, L2 regularization) or increasing the amount of training data can be used to solve this problem. L2 regularization adds the square sum of the parameters to the loss function, i.e. (where λ is the regularization parameter and p is the number of parameters) to prevent the model parameters from being too large, thereby avoiding overfitting.

[0126] 4. Real-time monitoring and early warning mechanism:

[0127] In the process of material supply, the supplier's behavior status is monitored in real time based on the unified data management platform and supplier risk assessment model. By setting key nodes and thresholds, production planning time, manufacturing progress, delivery notification confirmation time, etc., abnormal situations can be discovered in time.

[0128] When the monitoring system detects that the supplier's behavior deviates from the normal range or the risk assessment index exceeds the preset threshold, the early warning mechanism is automatically triggered. The early warning information is promptly pushed to relevant business personnel through various channels (such as system pop-up windows, SMS notifications, email reminders, etc.) so that they can take corresponding measures, such as communicating and coordinating with suppliers, adjusting material supply plans, etc.

[0129] Compared with the prior art, this application has the following advantages:

[0130] 1. After obtaining multi-source data and cleaning and classifying the data, the corresponding risk indicator feature data is extracted, which improves the effectiveness and availability of the risk indicator feature data, thereby improving the training progress of the model training and the evaluation accuracy of the model.

[0131] 2. By utilizing the interpretability of logistic regression and the powerful fitting ability of neural networks, the fitting ability of the trained model is improved, as well as the risk assessment accuracy of the trained model, thereby improving the effectiveness and reliability of the model in practical applications.

[0132] 3. Through optimization methods such as maximum likelihood estimation and gradient descent, the model can effectively adjust parameters to ensure a good fit on the training set, while ensuring the stability and generalization ability of the model through cross-validation.

[0133] 4. Based on a unified data management platform, suppliers' behaviors are monitored in real time, abnormal situations are discovered in time and early warning information is immediately generated, and the early warning information is sent to the user terminal for warning, so that users can make timely adjustments based on the early warning information and ensure the supply stability of the material supply chain.

[0134] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0135] Based on the same inventive concept, the embodiment of the present application also provides a training device for a supply provider risk assessment model for implementing the training method for the supply provider risk assessment model involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the training device for one or more supply provider risk assessment models provided below can refer to the limitations of the training method for the supply provider risk assessment model above, and will not be repeated here.

[0136] In an exemplary embodiment, Figure 4 As shown, a training device for a supply provider risk assessment model is provided, comprising: a data construction module 401, a weight construction module 402, a logistic regression value construction module 403, a loss function construction module 404 and a parameter adjustment module 405, wherein:

[0137] The data construction module 401 is used to obtain the full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider according to each full life cycle data;

[0138] A weight construction module 402 is used to input each risk indicator characteristic data into an indicator weight construction module in the supply provider risk assessment model to be trained, and obtain the indicator weight corresponding to each risk indicator characteristic data through the indicator weight construction module;

[0139] The logistic regression value construction module 403 is used to input each risk indicator characteristic data and each indicator weight into the logistic regression module in the supply provider risk assessment model to be trained, and obtain the logistic regression value associated with the supply provider risk assessment model to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated;

[0140] The loss function construction module 404 is used to input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each sample supply provider through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained according to the predicted risk data and the logistic regression value;

[0141] The parameter adjustment module 405 is used to adjust the pre-set neuron weights, thresholds and estimated parameters in the risk prediction module until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

[0142] In one embodiment, the weight construction module 402 is further used to obtain information entropy corresponding to each risk indicator feature data through an indicator weight construction module, and obtain indicator weight corresponding to each risk indicator feature data according to each information entropy.

[0143] In one embodiment, the full life cycle data includes actual risk data associated with sample supply providers; the training device of the supply provider risk assessment model also includes a model optimization module, which is used to obtain the accuracy of the model performance used to evaluate the supply provider risk assessment model based on the predicted risk data and the actual risk data; based on the analysis results of the accuracy, the performance of the supply provider risk assessment model is optimized.

[0144] In an exemplary embodiment, the model optimization module is also used to obtain the regularization parameters pre-set for the supply provider risk assessment model when the analysis results indicate that the prediction accuracy of the supply provider risk assessment model does not reach the set target value; and update the loss function of the supply provider risk assessment model based on the regularization parameters and the parameters to be estimated.

[0145] In one embodiment, the regularization parameter includes at least one of an L1 regularization parameter and an L2 regularization parameter; the model optimization module is further used to obtain the absolute value and term of the parameter based on the L1 regularization parameter and the parameter to be estimated, and update the loss function according to the absolute value and term of the parameter; and / or, according to the L2 regularization parameter and the parameter to be estimated, obtain the square sum term of the parameter, and update the loss function based on the square sum term of the parameter.

[0146] In one of the embodiments, the data construction module 401 is also used to obtain the original full life cycle data of multiple sample supply providers; perform data cleaning on the original full life cycle data, use a data filtering algorithm to remove noise data and abnormal data therein, and correct illegal time data therein to obtain the cleaned original full life cycle data; perform data classification according to the data type and business link of the cleaned original life cycle data to obtain the full life cycle data.

[0147] Each module in the training device for the supply provider risk assessment model can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0148] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store full life cycle data, risk indicator feature data, indicator weights, and logistic regression value data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a training method for a supply provider risk assessment model is implemented.

[0149] Those skilled in the art will understand that Figure 5The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0150] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the training method of the supply provider risk assessment model in the above embodiment is implemented.

[0151] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the training method of the supply provider risk assessment model in the above embodiment is implemented.

[0152] In one embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the method for training the supply provider risk assessment model in the above embodiment.

[0153] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0154] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0155] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0156] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A training method for a supply provider risk assessment model, characterized in that: The method comprises: Acquire full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider based on each of the full life cycle data; Inputting each of the risk indicator characteristic data into an indicator weight construction module in the supply provider risk assessment model to be trained, and obtaining the indicator weight corresponding to each of the risk indicator characteristic data through the indicator weight construction module; Input each of the risk indicator characteristic data and each of the indicator weights into the logistic regression module in the supply provider risk assessment model to be trained, and obtain the logistic regression value associated with the supply provider risk assessment model to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each of the risk indicator characteristic data, and the logistic regression value contains multiple parameters to be estimated; Input the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtain the predicted risk data corresponding to each of the sample supply providers through the risk prediction module, and construct the loss function corresponding to the risk assessment model of the supply provider to be trained according to the predicted risk data and the logistic regression value; The pre-set neuron weights, thresholds and parameters to be estimated in the risk prediction module are adjusted until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

2. The method according to claim 1, characterized in that The indicator weight corresponding to each of the risk indicator characteristic data is obtained through the indicator weight construction module, including: The information entropy corresponding to each of the risk indicator characteristic data is obtained through the indicator weight construction module, and the indicator weight corresponding to each of the risk indicator characteristic data is obtained according to each of the information entropies.

3. The method according to claim 1, characterized in that The full life cycle data includes actual risk data associated with the sample supply provider; After the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model, the method includes: Obtaining, based on the predicted risk data and the actual risk data, an accuracy rate for evaluating the model performance of the supply provider risk assessment model; Based on the analysis results of the accuracy rate, the performance of the supply provider risk assessment model is optimized.

4. The method according to claim 3, characterized in that The performance optimization of the supply provider risk assessment model based on the analysis result of the accuracy rate includes: In the case where the analysis result indicates that the prediction accuracy of the supply provider risk assessment model does not reach a set target value, obtaining a regularization parameter pre-set for the supply provider risk assessment model; The loss function of the supply provider risk assessment model is updated according to the regularization parameter and the parameter to be estimated.

5. The method according to claim 4, characterized in that The regularization parameter includes at least one of an L1 regularization parameter and an L2 regularization parameter; The updating of the loss function of the supply provider risk assessment model according to the regularization parameter and the parameter to be estimated includes: Based on the L1 regularization parameter and the parameter to be estimated, obtaining the absolute value and term of the parameter, and updating the loss function according to the absolute value and term of the parameter; and / or According to the L2 regularization parameter and the parameter to be estimated, a square sum term of the parameter is obtained, and the loss function is updated based on the square sum term of the parameter.

6. The method according to any one of claims 1 to 5, characterized in that: The obtaining of full life cycle data of multiple sample supply providers includes: Obtain original full life cycle data from multiple sample supply providers; Performing data cleaning on the original full life cycle data, using a data filtering algorithm to remove noise data and abnormal data therein, and correcting illegal time data therein, to obtain the cleaned original full life cycle data; Data classification is performed according to the data type and business link of the cleaned original life cycle data to obtain the full life cycle data.

7. A training device for a supply provider risk assessment model, characterized in that: The device comprises: A data construction module, used to obtain full life cycle data of multiple sample supply providers, and obtain multiple risk indicator characteristic data of each sample supply provider according to each of the full life cycle data; A weight construction module, used for inputting each of the risk indicator characteristic data into an indicator weight construction module in the supply provider risk assessment model to be trained, and obtaining the indicator weight corresponding to each of the risk indicator characteristic data through the indicator weight construction module; A logistic regression value construction module, used for inputting each of the risk indicator characteristic data and each of the indicator weights into the logistic regression module in the supply provider risk assessment model to be trained, and obtaining the logistic regression value associated with the supply provider risk assessment model to be trained through the logistic regression module; the logistic regression value is used to characterize the conditional distribution probability of each of the risk indicator characteristic data, and the logistic regression value contains a plurality of parameters to be estimated; A loss function construction module, used for inputting the risk indicator characteristic data and the logistic regression value into the risk prediction module in the risk assessment model of the supply provider to be trained, obtaining the predicted risk data corresponding to each of the sample supply providers through the risk prediction module, and constructing the loss function corresponding to the risk assessment model of the supply provider to be trained according to the predicted risk data and the logistic regression value; A parameter adjustment module is used to adjust the pre-set neuron weights, thresholds and the parameters to be estimated in the risk prediction module until the loss function reaches a minimum value, and the supply provider risk assessment model to be trained is used as the trained supply provider risk assessment model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images