CN-122022988-A - Credit review-oriented enterprise data asset value assessment method and system

CN122022988ACN 122022988 ACN122022988 ACN 122022988ACN-122022988-A

Abstract

The invention relates to the technical field of asset value evaluation, in particular to a credit-review-oriented enterprise data asset value evaluation method and system, comprising the following steps of acquiring enterprise data asset metadata and orthogonally decomposing and extracting dimension characteristics, and constructing a feature vector set, combining the historical cases to generate a target mapping point and a transaction point, calculating price traction force, constructing a tension balance equation to calculate a mechanical balance position, and recombining the dynamic balance position into a transaction price to obtain a virtual anchor point price. According to the invention, independent feature vectors under different dimensions are extracted through orthogonal decomposition operation, a multidimensional attribute space is constructed, the traction force is quantized and synthesized by means of a tension balance equation, the dynamic weight of each historical trading point on the target asset estimation is accurately deduced, a virtual anchor point price is constructed based on a weighted recombination mode, a value estimation path facing market behavior feedback is established, and the problems of static evaluation result, strong subjectivity and estimation departure from credit scenes are effectively overcome.

Inventors

SUN XIAOJI
WANG ZHIYONG
SUN WEN
CHENG XIAOMEI
QIN JING

Assignees

山东青年政治学院

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (9)

1. A credit review oriented enterprise data asset value assessment method, comprising the steps of: s1, acquiring metadata of enterprise data assets to be evaluated, performing orthogonal decomposition, extracting identity recognition, operation behaviors and performance history dimension characteristics, and constructing an independent feature vector set; S2, calling a historical transaction case library, extracting time-efficiency attenuation rate, main body association degree and industry verticality parameters of cases, constructing a multidimensional attribute vector space, mapping the independent feature vector set and the historical transaction cases to the multidimensional attribute vector space, and generating a target mapping point and a discrete transaction point; s3, performing Euclidean distance operation and Mahalanobis distance operation on the target mapping points and the discrete transaction points, and calculating a price traction value according to the space geometric distance value and the discrete transaction point time long-distance parameter; S4, constructing a tension balance equation based on the price traction force value, solving the mechanical balance position of the target mapping point, and analyzing the dynamic estimation weight of the discrete transaction point according to the mechanical balance position; and S5, acquiring the historical transaction price of the discrete transaction point, and carrying out weighted linear recombination on the historical transaction price based on the dynamic estimation weight to obtain the virtual anchor point price.
2. The credit-oriented enterprise data asset value assessment method of claim 1, wherein the set of independent feature vectors comprises an identity attribute feature component, a business behavior feature component, and a historical performance feature component, the discrete transaction points comprise a timeliness dimension coordinate, a subject association dimension coordinate, and an industry verticality dimension coordinate, the price traction values comprise a distance decay intensity value, a time penalty correction value, and a sample attraction contribution value, the dynamic valuation weights comprise a spatial tension balance coefficient, a sample local distribution density, and a map point attribution probability, and the virtual anchor prices comprise a historical transaction benchmark price, a comprehensive feature correction value, and a market liquidity discount.
3. The credit-review-oriented enterprise data asset value assessment method of claim 1, wherein the step of obtaining the independent feature vector set specifically comprises: S111, accessing a bottom storage unit interface of enterprise data assets to be evaluated, scanning a field definition structure of a data table, extracting a creation time stamp, an owner permission identification and a storage capacity numerical value, analyzing a service description tag field in metadata, converting a text description tag into a high-dimension value vector, carrying out normalization splicing processing on the numerical value vector and a numerical value statistics field, and establishing a multidimensional metadata attribute set; S112, calling the multidimensional metadata attribute set, executing feature orthogonalization and disassembly according to credit business logic, mapping unified social credit codes in the attribute set into identity recognition dimensions, mapping quarternary credit fluctuation rate and supply chain interaction frequency into operation behavior dimensions, mapping past loan overdue days and guarantee performance records into performance history dimensions, respectively calculating the mean value and variance statistics of attribute data in each dimension, and generating orthogonal dimension feature components; S113, calculating a covariance matrix among each dimension according to the orthogonal dimension characteristic components, performing eigenvalue decomposition operation on the covariance matrix, extracting eigenvalues and corresponding eigenvectors, screening principal component directions with accumulated contribution rates larger than a preset contribution rate according to the magnitude of eigenvalue values, projecting the orthogonal dimension characteristic components into an orthogonal coordinate system formed by the principal component directions, and constructing an independent eigenvector set.
4. The credit-review-oriented enterprise data asset value assessment method of claim 3, wherein the discrete point of transaction obtaining step specifically comprises: S211, calling a historical transaction case library, traversing a transaction record index stored in the historical transaction case library, extracting a transaction completion time stamp, a main body stock right relation chain and an industry category classification code of each record, calculating data timeliness attenuation rate, main body association degree and industry verticality parameters, defining an orthogonal coordinate axis and a measurement scale range according to a numerical distribution interval of the three parameters, and establishing a multidimensional attribute vector space; s212, invoking the multidimensional attribute vector space and the independent feature vector set, analyzing a time dimension component, a main dimension component and an industry dimension component in the feature vector set, executing normalized mapping on the three components according to the measurement scale of the coordinate axis, and projecting the normalized feature vector to the geometric coordinate position of the coordinate system to generate a target mapping point; S213, for the target mapping points, retrieving parameter values corresponding to all cases in the historical transaction case library, converting each group of parameter values into space coordinate vectors based on the axial definition of the shared coordinate system, and projecting the historical transaction cases into vector spaces distributed around the target mapping points in a scattered point mode to generate discrete transaction points.
5. The credit-review-oriented enterprise data asset value assessment method of claim 4, wherein the step of obtaining the price traction value comprises the steps of: S311, invoking coordinate position data of the target mapping point and the discrete transaction point in a multidimensional space, executing vector difference value operation, obtaining coordinate deviation value under each dimension, executing standardized projection calculation on the coordinate deviation value by combining an inverse matrix of a covariance matrix, and carrying out weighted merging processing on a projection result and a coordinate absolute deviation model length to generate a space geometric distance value; s312, extracting a transaction delivery time stamp in the discrete transaction point metadata, calculating a time span of the time stamp and the current credit review reference day, and performing exponential normalization transformation on the time span based on a preset decay half-life constant to generate a time long parameter; S313, calling the space geometric distance value and the time long-distance parameter, introducing a data confidence score, a noise interference coefficient and a market association flux as price influence factors, and calculating to obtain a price traction value.
6. The credit-review-oriented enterprise data asset value assessment method of claim 5, wherein the formula for computing the value of the obtained price traction is specifically: ; Wherein, the Representing the value of the price traction force, Representing the normalized spatial distance coefficient, Representing the normalized time decay factor, Representing the confidence index of the normalized data, Representing the normalized noise-to-interference ratio, Representing normalized market correlation coefficients.
7. The method for credit-review-oriented enterprise data asset value assessment of claim 5, wherein the step of obtaining the dynamic valuation weights comprises: S411, defining scalar tension intensity acting on a target mapping point based on the price traction force value, defining and calculating a unit direction vector of the target mapping point to each discrete transaction point by combining a coordinate system of a multidimensional attribute vector space, performing product operation on the scalar tension intensity and the unit direction vector to obtain component force vectors, performing vector superposition on all the component force vectors to construct a resultant force function, performing equality constraint association on the resultant force function and a zero vector, and establishing a space tension balance equation set; S412, calling the space tension balance equation set, setting the original coordinates of the target mapping points as iteration initial values, calculating partial derivatives of resultant force functions about coordinate axes, obtaining force field gradient vectors, performing coordinate displacement updating, continuously monitoring the modular length of the resultant force vectors, judging whether the modular length falls into a preset convergence tolerance interval, stopping iteration and locking the current coordinate values when convergence conditions are met, and generating a mechanical balance position; S413, performing space distance measurement on the mechanical balance position and the coordinates of each discrete transaction point, calculating the reciprocal of the space distance value, quantifying the geometric affinity of each transaction case to the balance point, substituting the geometric affinity into a normalized calculation model to calculate the relative duty ratio value of the case in the whole sample set, mapping the relative duty ratio value into the price confidence level of the evaluation model to the case, and generating a dynamic evaluation weight.
8. The credit-review-oriented enterprise data asset value assessment method of claim 7, wherein the virtual anchor price obtaining step specifically comprises: S511, extracting settlement amount and payment currency codes according to transaction accounting records associated with the discrete transaction points, calling resident consumption price index sequences and exchange rate intermediate price historical data, and executing purchasing power flat price correction and currency unified conversion on the settlement amount according to time spans of evaluation reference days and transaction occurrence days to eliminate currency value deviation caused by currency expansion and exchange rate fluctuation and generate historical transaction reference price; S512, calling the dynamic estimation weight and the historical transaction reference price, performing scalar multiplication operation item by item on the reference price by taking the weight value as a projection coefficient, determining the value contribution amplitude of a single sample, acquiring the listing-up transaction conversion rate and the average holding period of a target data product in a transaction market, and calculating the transaction friction cost and the opportunity cost in the asset transition process according to the market depth to generate a market liquidity discount; And S513, performing linear weighted accumulation on the historical transaction reference price based on the dynamic estimation weight, obtaining a theoretical weighted average estimation, inputting the market mobility discount as a negative correction vector into price synthesis logic, and performing bidirectional amplitude smooth calibration on an operation result by combining with the scarcity coefficient of the industry to generate a virtual anchor point price.
9. A credit-review-oriented enterprise data asset value assessment system for implementing the credit-review-oriented enterprise data asset value assessment method of any of claims 1-8, the system comprising: The feature vector sorting module is used for acquiring metadata of enterprise data assets to be evaluated, performing orthogonal decomposition, extracting identity recognition, operation behaviors and performance history dimension features, and constructing an independent feature vector set; The transaction point identification module is used for calling a historical transaction case library, extracting the time efficiency attenuation rate, the main body association degree and the industry verticality parameters of cases, constructing a multidimensional attribute vector space, mapping the independent feature vector set and the historical transaction cases to the multidimensional attribute vector space, and generating a target mapping point and a discrete transaction point; The price influence analysis module is used for executing Euclidean distance operation and Mahalanobis distance operation on the target mapping points and the discrete transaction points, and calculating a price traction value according to the space geometric distance value and the discrete transaction point time long-distance parameter; The weight fluctuation analysis module is used for constructing a tension balance equation based on the price traction force value, solving the mechanical balance position of the target mapping point and analyzing the dynamic estimation weight of the discrete transaction point according to the mechanical balance position; and the value correction evaluation module is used for acquiring the historical transaction price of the discrete transaction point, and carrying out weighted linear recombination on the historical transaction price based on the dynamic estimation weight to obtain the virtual anchor point price.

Description

Credit review-oriented enterprise data asset value assessment method and system Technical Field The invention relates to the technical field of asset value evaluation, in particular to a credit review-oriented enterprise data asset value evaluation method and system. Background The technical field of asset value assessment relates to systematic value judgment and quantification of tangible or intangible assets owned by enterprises, institutions or individuals, and the core matters comprise asset element identification, asset attribute modeling, value metering method construction and assessment result expression. In an enterprise data asset assessment scene, the field covers the fields of identification classification of data assets, establishment of a data quality index system, analysis of data usability, construction of a data asset pricing model, association modeling between data assets and business performance or financial value, and the like, and a technical path generally comprises structural management of data resources, mining of data value contribution factors and value calculation according to the data asset contribution factors, and is widely applied to the fields of financial audit, purchase assessment, risk control, credit giving and the like. The traditional enterprise data asset value evaluation method facing credit evaluation refers to taking data resources accumulated by enterprises in daily operations as evaluation objects, evaluating the scale, type, integrity, update frequency, use frequency and other attributes of data, and judging the variability of the data assets and the supporting degree of credit repayment capacity according to historical financing conditions, service use scenes, data production flow stability and other indexes. The traditional method generally adopts a static index analysis or expert experience scoring mode to assign values to the data assets, lacks of refinement judgment of dynamic circulation and application value of the data assets in an actual credit scene, generally relies on manual setting of a weight model in an evaluation flow, and carries out value calculation by combining data asset detail list, business system record, financial statement notes and other data. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a credit review-oriented enterprise data asset value assessment method and system. In order to achieve the above purpose, the invention adopts the following technical scheme that the method for evaluating the asset value of the enterprise data facing credit review comprises the following steps: s1, acquiring metadata of enterprise data assets to be evaluated, performing orthogonal decomposition, extracting identity recognition, operation behaviors and performance history dimension characteristics, and constructing an independent feature vector set; S2, calling a historical transaction case library, extracting time-efficiency attenuation rate, main body association degree and industry verticality parameters of cases, constructing a multidimensional attribute vector space, mapping the independent feature vector set and the historical transaction cases to the multidimensional attribute vector space, and generating a target mapping point and a discrete transaction point; s3, performing Euclidean distance operation and Mahalanobis distance operation on the target mapping points and the discrete transaction points, and calculating a price traction value according to the space geometric distance value and the discrete transaction point time long-distance parameter; S4, constructing a tension balance equation based on the price traction force value, solving the mechanical balance position of the target mapping point, and analyzing the dynamic estimation weight of the discrete transaction point according to the mechanical balance position; and S5, acquiring the historical transaction price of the discrete transaction point, and carrying out weighted linear recombination on the historical transaction price based on the dynamic estimation weight to obtain the virtual anchor point price. The invention is improved in that the independent feature vector set comprises an identity attribute feature component, an operation behavior feature component and a historical performance feature component, the discrete transaction points comprise an aging dimension coordinate, a main body association dimension coordinate and an industry verticality dimension coordinate, the price traction value comprises a distance attenuation intensity value, a time penalty correction value and a sample attraction contribution value, the dynamic estimation weight comprises a space tension balance coefficient, a sample local distribution density and a mapping point attribution probability, and the virtual anchor point price comprises a historical transaction reference price, a comprehensive feature correction amount and a market mobility discount. T