Dynamic evaluation model and implementation method of data element value

Abstract

The present invention relates to a dynamic evaluation model and implementation method for the value of data elements, which belongs to the field of data assets, including quality evaluation: evaluating the quality of data elements; value evaluation: evaluating the value generated by data elements in the market in actual applications; risk evaluation: evaluating the risks that may be generated by data during use; comprehensive evaluation: combining the evaluation results of the data element quality evaluation module, the data element market application evaluation module and the data element value risk evaluation module, performing a correlation analysis of the quality, market application and risk of the data elements through a neural network, assigning corresponding weights to them, and using a weighted average formula to obtain a comprehensive evaluation result. The present invention can comprehensively and dynamically reflect the value status of data elements, ensuring the comprehensiveness and accuracy of the evaluation results.

Description

Technical Field

[0001] The present invention relates to the field of data assets, and in particular to a dynamic evaluation model and implementation method for the value of data elements. Background Art

[0002] Data elements refer to data resources that exist in electronic form and participate in production and operation activities through computing and play an important role. These data resources can be collected, sorted, processed, and formed into information, data sets, and data products, thus playing an important role in the production process. Data elements include not only traditional structured data, such as corporate financial statements and transaction records, but also semi-structured and unstructured data, such as social media content, images, and videos.

[0003] Data plays a vital role in the digital economy. It can optimize resource allocation, improve production efficiency, and promote high-quality economic and social development. Through the effective application of data, new industries and models can be created and new economic growth points can be cultivated. The multiplier effect of data is reflected in the aspects from connection to coordination, from use to reuse, from superposition to integration, which further improves the total factor productivity.

[0004] The value of data elements can be assessed using the income approach and the cost approach:

[0005] 1. Income method:

[0006] Definition: Based on the theory of utility value, the expected return value of data is used as data valuation;

[0007] Advantages: Considering the data usage value and time value factors, it can reflect the real data value;

[0008] Disadvantages: Expected returns are difficult to predict and accuracy cannot be guaranteed; pricing is subjective and discount rates are difficult to determine; it only takes the interests of the seller into consideration without considering competitive factors and market demand.

[0009] 2. Cost method:

[0010] Definition: Based on the cost-value theory, the replacement cost or acquisition cost of data assets is used as the basis for evaluation;

[0011] Advantages: It can ensure that the costs incurred by the enterprise are compensated and that a certain profit can be obtained under normal circumstances;

[0012] Disadvantages: It ignores the changes in product demand elasticity and cannot adapt quickly to the changing market; the correspondence between costs and benefits is weak, and price estimates are low.

[0013] The above methods ignore the real reaction of the value of data elements in the market, so it is more authentic to evaluate directly based on the market. Similar transactions in the market are used as pricing references, which can better reflect the market supply and demand relationship, and the calculation is simple and easy. However, since the data market is not yet mature and the transaction scale is small, it is difficult to find comparable cases. At the same time, the evaluation of the value of data elements is prone to bias and cannot fully reflect the value status of data elements. At the same time, the value of data elements will change with the market, and the accuracy, completeness, timeliness, market application value, and risks of the data will have an impact on the value of data elements. In summary, the evaluation of the value of data elements lacks objective quantitative evaluation, has low timeliness, and low evaluation efficiency. Summary of the invention

[0014] In order to solve the above-mentioned problems in the prior art, the present invention provides a dynamic evaluation model and implementation method for the value of data elements.

[0015] In a first aspect, the data element value dynamic evaluation model and implementation method provided by the present invention include the following steps:

[0016] Quality assessment: a data element quality assessment module is set up to assess the quality of data elements; the quality of data elements includes the accuracy, completeness and timeliness of the data;

[0017] Value assessment: Set up a data element market application assessment module to assess the value of data elements in the market in actual applications;

[0018] Risk assessment: Evaluate the risks that may arise during the use of data through the data element value risk assessment module;

[0019] Comprehensive evaluation: Combining the evaluation results of the data element quality evaluation module, the data element market application evaluation module and the data element value risk evaluation module, a neural network is used to analyze the correlation between the quality, market application and risk of the data elements, and corresponding weights are assigned to them. The weighted average formula is used to obtain the comprehensive evaluation results.

[0020] Furthermore, the weighted average formula is:

[0021] Z=w1·M+w2·J+w3·F, where Z is the comprehensive evaluation result, w1, w2, and w3 are the weight coefficients of the evaluation results of data element quality M, market value J, and risk F respectively.

[0022] Furthermore, before the neural network is trained and processed, preprocessing and data encoding are also included, specifically: standardization of quality, market application and risk related data, and unique hot encoding of quality, market application and risk data.

[0023] Furthermore, the neural network includes an embedding layer, an input layer, a hidden layer and an output layer; the embedding layer converts high-dimensional unique hot encoding into a low-dimensional dense vector; the input layer: contains three nodes corresponding to the embedding vectors of quality, market value and risk respectively; each neuron in the hidden layer is connected to the node of the input layer through weights and biases, and uses an activation function for nonlinear transformation; the output layer contains a neuron, which outputs the weight coefficients {w1, w2, w3} of the comprehensive evaluation results, w1 is the quality evaluation weight, w2 is the market value evaluation weight, and w3 is the risk evaluation weight;

[0024] Furthermore, the loss function of the neural network is defined as follows: the mean square error is used as the loss function to measure the difference between the weight of the neural network output and the weight of the desired target comprehensive evaluation result.

[0025] Furthermore, the training process of the neural network is as follows:

[0026] Data preparation: Collect the assessment values ​​of quality, market value and risk and the corresponding comprehensive assessment results as training data;

[0027] Initialize weights and biases: Randomly initialize the weights and biases of the neural network;

[0028] Forward propagation: forward propagation of input data through the neural network to calculate the weights w1, w2, w3 of the output layer;

[0029] Calculate loss: Calculate the loss value under the current weight according to the loss function;

[0030] Backpropagation: Use optimization algorithms such as gradient descent to calculate the gradient of the loss function with respect to the weights, and update the weights and biases of the neural network through backpropagation;

[0031] Iterative training: Repeat the process of forward propagation, loss calculation, and back propagation until the loss function converges or the preset number of training rounds is reached;

[0032] Validation and testing: Use validation and test sets to evaluate the performance of the neural network and ensure that the model performs well on unseen data.

[0033] Furthermore, in the data preparation stage, the assessment values ​​of collection quality, market value and risk and the corresponding comprehensive assessment results are verified according to expert knowledge, experimental results and actual observed shapes, the data samples are manually determined, and the data samples are amplified through small sample learning.

[0034] The beneficial effects of the present invention are reflected in that the present invention conducts a comprehensive evaluation from multiple dimensions such as the quality, market application, and value risk of data elements, and can comprehensively and dynamically reflect the value status of data elements; a correlation analysis of the quality, market application, and risk of data elements is performed through a neural network, and corresponding weights are assigned to them, and then a weighted average formula is used to obtain a comprehensive evaluation result, thereby ensuring the comprehensiveness and accuracy of the evaluation result. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A flow chart of the method provided by the present invention; DETAILED DESCRIPTION

[0036] 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.

[0037] Example 1: Reference Figure 1 As shown, the data element value dynamic assessment model and implementation method of the embodiment of the present invention include a data element quality assessment module, a data element market application assessment module, a data element value risk assessment module and a data element value comprehensive assessment model; the implementation method is as follows:

[0038] Quality assessment: Assess the quality of data elements;

[0039] Value assessment: assess the value of data elements in the market in actual applications;

[0040] Risk assessment: assess the risks that may arise during the use of data;

[0041] Comprehensive evaluation: Combining the evaluation results of the data element quality evaluation module, the data element market application evaluation module and the data element value risk evaluation module, a neural network is used to analyze the correlation between the quality, market application and risk of the data elements, and corresponding weights are assigned to them. The weighted average formula is used to obtain the comprehensive evaluation results.

[0042] In this embodiment, 1. Data element quality assessment module: the assessment objectives mainly include the accuracy, completeness and timeliness of data elements. The details are as follows:

[0043] Accuracy: Assess the accuracy of the data and whether the data source is reliable;

[0044] Completeness: assess whether the data is complete and contains all the information required;

[0045] Timeliness: Check the data update frequency to ensure that the data reflects the actual situation in a timely manner, analyze the data timestamp and validity period, and ensure that the data is valid during the evaluation period. Use indicators such as data delay time to evaluate data timeliness.

[0046] 2. Data element market application evaluation module: used to evaluate the value of data elements in the market in actual applications, including the scarcity, relevance, market demand and transaction price of data elements;

[0047] Scarcity: Assess the uniqueness and difficulty of data, and whether there are other data resources that can replace the target data. The lower the substitutability, the higher the market value of the data;

[0048] Relevance: Evaluate whether the data is relevant to the business objectives and strategy of the enterprise;

[0049] Market demand: Assess the market demand for data, including the number of potential users, the breadth of industry applications, etc.

[0050] Transaction price: the transaction price of evaluation data, including various forms of prices such as purchase, subscription, and authorization.

[0051] 3. Data element value risk assessment module evaluates the risks that may arise during the use of data, including data leakage, compliance and reliability.

[0052] Data leakage: assess whether there is a risk of data leakage during use;

[0053] Compliance: Assess whether the data may lead to legal liability due to violation of regulations or contracts;

[0054] Reliability: Assess the stability and availability risks of data during use, such as whether the data is susceptible to damage, loss or tampering;

[0055] 4. In the comprehensive evaluation model of data element value, the comprehensive evaluation result Z is obtained through the weighted average formula, Z = w1·M+w2·J+w3·F, where w1, w2, and w3 are the weight coefficients of the evaluation results of data element quality M, market value J, and risk F respectively;

[0056] Before the neural network is trained and processed, preprocessing and data encoding are required. Specifically, standardization of quality, market application and risk-related data is performed, and one-hot encoding is performed on quality, market application and risk. The neural network structure includes an embedding layer, an input layer, a hidden layer and an output layer.

[0057] Embedding layer: converts high-dimensional one-hot encoding into low-dimensional dense vectors;

[0058] Input layer: contains three nodes, corresponding to the embedding vectors of quality, market value and risk respectively;

[0059] Hidden layer: Set up one or more hidden layers, each of which contains a certain number of neurons (nodes). Each neuron is connected to the nodes of the input layer through weights and biases, and uses an activation function (such as ReLU, Sigmoid, or Tanh) for nonlinear transformation;

[0060] Output layer: contains one neuron, which outputs the weight coefficients {w1, w2, w3} of the comprehensive evaluation results, where w1 is the quality evaluation weight, w2 is the market value evaluation weight, and w3 is the risk evaluation weight;

[0061] Definition of loss function: The mean square error (MSE) is used as the loss function to measure the difference between the weight of the neural network output and the weight of the expected target comprehensive evaluation result; that is, the sum of the squares of the differences between the comprehensive evaluation result Z obtained by calculating the weight combination of the neural network output and the true value or target value.

[0062] The training process is as follows:

[0063] 1. Data preparation: Collect the assessment values ​​of quality M, market value J and risk F and the corresponding comprehensive assessment results Z as training data;

[0064] 2. Initialize weights and biases: Randomly initialize the weights and biases of the neural network.

[0065] 3. Forward propagation: The input data is forward propagated through the neural network to calculate the weights w1, w2, and w3 of the output layer.

[0066] 4. Calculate loss: Calculate the loss value under the current weight according to the loss function.

[0067] 5. Backpropagation: Use optimization algorithms such as gradient descent to calculate the gradient of the loss function with respect to the weights, and update the weights and biases of the neural network through backpropagation.

[0068] 6. Iterative training: Repeat the process of forward propagation, loss calculation, and back propagation until the loss function converges or reaches the preset number of training rounds.

[0069] 7. Validation and testing: Use validation and test sets to evaluate the performance of the neural network to ensure that the model performs well on unseen data.

[0070] In the description of the embodiments of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "center", "top", "bottom", "top", "bottom", "inside", "outside", "inner side", "outer side" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Among them, "inside" refers to an internal or enclosed area or space. "Periphery" refers to the area surrounding a specific component or a specific area.

[0071] In the description of the embodiments of the present invention, the terms "first", "second", "third", and "fourth" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", and "fourth" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0072] In the description of the embodiments of the present invention, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0073] In the description of the embodiments of the present invention, the term "and / or" herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " herein generally indicates that the associated objects before and after are in an "or" relationship.

[0074] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic evaluation model and implementation method for the value of data elements, characterized in that: The following steps are involved: Quality assessment: a data element quality assessment module is set up to assess the quality of data elements; the quality of data elements includes the accuracy, completeness and timeliness of the data; Value assessment: Set up a data element market application assessment module to assess the value of data elements in the market in actual applications; Risk assessment: Evaluate the risks that may arise during the use of data through the data element value risk assessment module; Comprehensive evaluation: Combining the evaluation results of the data element quality evaluation module, the data element market application evaluation module and the data element value risk evaluation module, a neural network is used to analyze the correlation between the quality, market application and risk of the data elements, and corresponding weights are assigned to them. The weighted average formula is used to obtain the comprehensive evaluation results.

2. The data element value dynamic evaluation model and implementation method according to claim 1 is characterized in that: The weighted average formula is: Z=w1·M+w2·J+w3·F, where Z is the comprehensive evaluation result, w1, w2, and w3 are the weight coefficients of the evaluation results of data element quality M, market value J, and risk F respectively.

3. The data element value dynamic evaluation model and implementation method according to claim 1 is characterized in that: Before the neural network is trained and processed, preprocessing and data encoding are also included. Specifically, the quality, market application and risk-related data are standardized, and the quality, market application and risk data are one-hot encoded.

4. The data element value dynamic evaluation model and implementation method according to claim 1 is characterized in that: The neural network includes an embedding layer, an input layer, a hidden layer and an output layer; the embedding layer converts high-dimensional one-hot encoding into a low-dimensional dense vector; the input layer: contains three nodes corresponding to the embedding vectors of quality, market value and risk respectively; each neuron in the hidden layer is connected to the node of the input layer through weights and biases, and uses an activation function for nonlinear transformation; The output layer includes a neuron, which outputs the weight coefficients {w1, w2, w3} of the comprehensive evaluation results, where w1 is the quality evaluation weight, w2 is the market value evaluation weight, and w3 is the risk evaluation weight.

5. The data element value dynamic evaluation model and implementation method according to claim 1 is characterized in that: Definition of loss function of neural network: The mean square error is used as the loss function to measure the difference between the weight of the neural network output and the weight of the expected target comprehensive evaluation result.

6. The data element value dynamic evaluation model and implementation method according to claim 1 is characterized in that: The training process of the neural network is as follows: Data preparation: Collect the assessment values ​​of quality, market value and risk and the corresponding comprehensive assessment results as training data; Initialize weights and biases: Randomly initialize the weights and biases of the neural network; Forward propagation: forward propagation of input data through the neural network to calculate the weights w1, w2, w3 of the output layer; Calculate loss: Calculate the loss value under the current weight according to the loss function; Backpropagation: Use optimization algorithms such as gradient descent to calculate the gradient of the loss function with respect to the weights, and update the weights and biases of the neural network through backpropagation; Iterative training: Repeat the process of forward propagation, loss calculation, and back propagation until the loss function converges or the preset number of training rounds is reached; Validation and testing: Use validation and test sets to evaluate the performance of the neural network and ensure that the model performs well on unseen data.

7. The data element value dynamic evaluation model and implementation method according to claim 6 is characterized in that: In the data preparation stage, the assessment values ​​of quality, market value and risk and the corresponding comprehensive assessment results are collected, and data verification is performed based on expert knowledge, experimental results and actual observed shapes. The data samples are manually determined and amplified through small sample learning.

Images