A method and system for collaborative evaluation of e-commerce data assets

Abstract

The application discloses a kind of collaborative evaluation method and system of e-commerce data assets, the method includes: obtaining cross-border e-commerce data source, constructs global commodity-data semantic mapping atlas, the data source is classified, and applies natural language processing technology to parse trade contract text to extract the ownership information of data;According to the classification result and ownership information, an evaluation index system containing data quality dimension, application scenario dimension and market value dimension is constructed, and the initial evaluation value of data assets is calculated;Real-time current policy variable data and initial evaluation value are input into the dynamic valuation engine based on reinforcement learning, and the value adjustment coefficient is output and the initial evaluation value is corrected, to obtain dynamic valuation;A quantitative mapping model between digital trade rules and the value of data assets is established, and according to dynamic valuation and current policy variable data, the optimal cross-border data circulation path is output;Based on the optimal cross-border data circulation path and dynamic valuation, an evaluation report is generated.

Description

Technical Field

[0001] This invention relates to the field of data asset valuation technology, and in particular to a collaborative valuation method and system for e-commerce data assets. Background Technology

[0002] With the rapid development of the global digital economy, data has become a key production factor driving economic growth. In the field of cross-border e-commerce, the value of data assets is increasingly prominent, becoming an important component of a company's core competitiveness. However, existing technologies for data asset valuation and management in the cross-border e-commerce environment have the following main shortcomings: First, the assessment dimensions are too narrow and lack adaptability to specific scenarios. Traditional data asset assessment methods mainly focus on basic attributes such as data volume and quality, failing to fully consider the application value of data in specific cross-border trade scenarios. Although the national standard GB / T 37550-2019, "E-commerce Data Asset Evaluation Index System," has established a basic framework for e-commerce data asset assessment, its adaptability to cross-border trade scenarios remains insufficient. Existing industry standards (such as T / CANA 001-2020, "E-commerce Data Rights Evaluation Standard") provide data rights value models, but fail to incorporate dynamic factors such as the cross-border policy environment into the assessment system, leading to discrepancies between assessment results and actual application value.

[0003] Secondly, existing methods have technical shortcomings in data ownership identification and classification. Cross-border e-commerce involves multiple legal systems, diverse trading entities, and complex contractual relationships, making the determination of data asset ownership extremely difficult. Traditional technical means lack the ability to automatically parse ownership clauses in trade contracts, making it difficult to efficiently and accurately extract key ownership information such as data owners, controllers, and usage restrictions. This makes the data asset ownership confirmation process a bottleneck in the entire assessment chain.

[0004] Third, policy compliance assessments lag behind. Major economies worldwide have established their own data governance rules, which differ significantly, creating a complex compliance environment. Existing assessment tools struggle to respond in real-time to the rapidly changing international policy environment, leading to compliance risks for companies in cross-border data flows.

[0005] Fourth, the assessment and circulation processes are disconnected, lacking collaborative optimization capabilities. In existing technologies, the valuation of data assets and the planning of cross-border data circulation routes are usually two separate processes. The assessment results fail to effectively guide the selection of circulation routes, and the constraints of circulation routes (such as policy compliance and transmission costs) are not incorporated into the assessment model, making it difficult for data assets to achieve optimal value allocation globally.

[0006] Therefore, there is an urgent need for a collaborative assessment method that can integrate multi-dimensional value assessment, dynamic policy response, data ownership identification, and circulation path optimization to address the complex challenges in the cross-border e-commerce environment. Summary of the Invention

[0007] This invention aims to address the following technical problems in the existing cross-border e-commerce data asset assessment and management: the assessment dimensions are singular and lack adaptability to cross-border scenarios; the identification and classification of data ownership lack automated analysis methods, making it difficult to accurately extract ownership information; policy compliance assessment is lagging behind and cannot dynamically respond to changes in the international policy environment; and the assessment and circulation processes are disconnected, lacking collaborative optimization capabilities.

[0008] To address the aforementioned problems, this invention provides a collaborative evaluation method for e-commerce data assets, comprising the following steps: The data sources of cross-border e-commerce are acquired, and a global commodity-data semantic mapping graph is constructed based on the commodity-data mapping standard to classify the data sources, generate classification results, and apply natural language processing technology to parse trade contract texts to extract the ownership information of the data, and identify the data sources as data assets. Based on the classification results and ownership information, an evaluation index system is constructed that includes data quality, application scenario, and market value dimensions, and the initial valuation value of the data assets is calculated. The real-time acquired current policy variable data and the initial assessment value are input into a reinforcement learning-based dynamic valuation engine, which outputs a value adjustment coefficient and corrects the initial assessment value to obtain a dynamic valuation. Establish a quantitative mapping model between digital trade rules and the value of the data assets, and output the optimal cross-border data flow path based on the dynamic valuation and the current policy variable data; Based on the optimal cross-border data flow path and the dynamic valuation, an assessment report is generated that includes the comprehensive value of the data assets and the risk warning level.

[0009] The product-data mapping standard is the UN / CEFACT standard.

[0010] After acquiring the cross-border e-commerce data source, the process further includes the following steps: preprocessing the data source, including: data cleaning, including missing value imputation, abnormal transaction filtering, and data deduplication; mapping product descriptions in different languages ​​to the UN / CEFACT standard terminology library using a pre-trained multilingual BERT model, and uniformly converting multiple currencies through a real-time exchange rate interface; implementing a dynamic desensitization engine to automatically perform differential privacy desensitization on data that fails compliance checks and replace it with simulated data that retains statistical characteristics, in accordance with the compliance requirements of multi-jurisdictional data security regulations; and using a stream processing framework to achieve real-time valuation updates of the data assets.

[0011] The process of constructing a global commodity-data semantic mapping graph based on the commodity-data mapping standard to classify the data sources and generate classification results further includes the following steps: extracting HS codes from customs declarations, associating them with the business requirement specification model of the commodity-data mapping standard, and mapping the physical attributes of commodities to data asset feature tags; mapping each key business node in the trade process to a data asset generation scenario based on the supply chain reference data model of the commodity-data mapping standard; dividing the data assets into multiple preset categories based on the mapping graph; and constructing a dynamic update mechanism using knowledge graph embedding technology, automatically adjusting the weight edges between graph nodes when the commodity-data standard is updated or new regional trade agreement clauses are added.

[0012] The method of using natural language processing technology to parse trade contract texts and extract ownership information further includes the following steps: using a layout recognition model to locate ownership-related sections in the trade contract text; performing named entity recognition in the ownership-related sections based on a legal knowledge graph to extract data owner entities, data controller entities, usage restriction entities, and profit distribution entities respectively; using a graph neural network to detect logical conflicts between the clauses containing each of the entities; and for ownership items in the clauses that are not involved in conflicts and are not explicitly stipulated, performing Bayesian probabilistic inference based on international rules and the laws of the target country to generate ownership conclusions with confidence levels.

[0013] The construction of an evaluation index system encompassing data quality, application scenario, and market value dimensions, and the calculation of the initial assessed value of the data assets, further includes the following steps: transforming the qualitative evaluation of each dimension into cloud digital features, including expectation, entropy, and hyperentropy; generating cloud droplet distributions corresponding to each dimension using a positive cloud generator; employing a game theory-based combined weighting method to fuse subjective AHP weights with objective entropy weights to calculate the weight vectors for each dimension; based on the cloud digital features of each dimension and the weight vectors, calculating the digital features of the comprehensive cloud using a weighted comprehensive cloud algorithm; and determining the initial assessed value range based on the certainty of the cloud droplet distributions of each dimension at a preset value level, thereby controlling the pricing error rate within a preset error threshold.

[0014] The reinforcement learning-based dynamic valuation engine employs a deep Q-network architecture and further includes the following steps: defining a state space, which includes internal and external states, wherein the internal state contains the current benchmark value of the data asset, data category identifier, and ownership clarity score, and the external state is the current policy variable data; discretizing the value adjustment coefficient into multiple discrete actions within a preset coefficient range and a preset step size to form an action space; constructing a reward function based on the difference between the predicted value output by the dynamic valuation engine and the market transaction price of similar data assets, as well as the compliance score; designing an experience replay pool to store the quadruples of state, action, reward, and next state within a preset time period, and prioritizing the sampling of samples with high temporal difference errors for training; when a sudden change in policy factors is detected, outputting a new value adjustment coefficient through the converged deep Q-network during a preset forward propagation time, so that the compliance assessment of the data asset matches the policies of major global economies to a preset matching threshold.

[0015] The quantitative mapping model between the digital trade rules and the value of the data assets is a semi-parametric spatial econometric model. The construction process of the semi-parametric spatial econometric model includes the following steps: Establish a semi-parametric spatial econometric model between the value of the data assets and the policy intensity vector: , in, Let be the value of the i-th data asset in period t; W is the spatial weight matrix of cross-border data flow constructed based on trade flows; The policy intensity vector includes the data localization index, the digital services tax rate, and the cross-border data flow restriction index. The policy elasticity coefficient is obtained through GMM estimation; These are the spatial autoregressive coefficients; and For control variables; Individual effect; This is a time effect; This is the random error term; The digital trade rules and provisions are transformed into structured policy parameters using natural language processing technology and input into the semi-parametric spatial econometric model. Based on the model output, the discount rate of logistics data assets is adjusted to optimize the cross-border data flow path, so that the efficiency of the cross-border flow of the data assets reaches a preset efficiency improvement threshold.

[0016] The process of generating an assessment report that includes the comprehensive value of data assets and risk warning levels also includes the following steps: calling the data asset transaction ecosystem assessment matrix to comprehensively score the data assets from multiple ecosystem dimensions; inputting the historical volatility sequence of the data asset value, policy update frequency, and enterprise information change frequency into the Cox proportional risk regression model to predict the probability of default risk, and automatically lowering the pledge ratio when the probability of risk exceeds a preset threshold; and generating the assessment report through a two-way feedback mechanism of policy sandbox-technology platform, wherein the downlink is used to intercept non-compliant requests, and the uplink is used to aggregate data to optimize rule parameters.

[0017] In another aspect, the present invention provides a collaborative evaluation system for e-commerce data assets, employing the aforementioned collaborative evaluation method for e-commerce data assets, and comprising: The data acquisition and identification module is used to acquire cross-border e-commerce data sources, construct a global commodity-data semantic mapping graph based on commodity-data mapping standards to classify the data sources, generate classification results, and apply natural language processing technology to parse trade contract texts to extract data ownership information and identify the data sources as data assets. The multi-dimensional evaluation module is used to construct an evaluation index system that includes data quality dimension, application scenario dimension and market value dimension based on the classification results and ownership information, and to calculate the initial evaluation value of the data asset. The dynamic evaluation module is used to input the real-time acquired current policy variable data and the initial evaluation value into the reinforcement learning-based dynamic valuation engine, output the value adjustment coefficient and correct the initial evaluation value to obtain the dynamic valuation. The collaborative mapping module is used to establish a quantitative mapping model between digital trade rules and the value of the data assets, and output the optimal cross-border data flow path based on the dynamic valuation and the current policy variable data. The full-ecosystem assessment module is used to generate an assessment report that includes the comprehensive value of data assets and risk warning levels based on the optimal cross-border data flow path and the dynamic valuation.

[0018] Compared with the prior art, the present invention has the following significant advantages: (1) Comprehensiveness of assessment: By constructing a multi-dimensional value assessment system that includes data quality, application scenario and market value dimensions, it can comprehensively consider the economic, technical and social value of data assets in the cross-border e-commerce environment, overcoming the shortcomings of traditional methods with single assessment dimensions.

[0019] (2) Policy adaptability: By introducing real-time policy variable data and using a reinforcement learning-based dynamic valuation engine to dynamically correct the initial assessment value, a policy-sensitive dynamic assessment model was constructed. This model can respond to global policy changes in real time, significantly improving the accuracy and timeliness of data asset compliance assessment.

[0020] (3) High efficiency in circulation: By establishing a quantitative mapping model between digital trade rules and the value of data assets, and outputting the optimal cross-border data circulation path based on dynamic valuation and policy variable data, the collaborative mapping between assessment and circulation is achieved. This collaborative mapping algorithm can optimize the cross-border data circulation path and effectively improve the efficiency of cross-border data flow.

[0021] (4) System integrity: It covers all aspects from data asset identification (ownership confirmation), multi-dimensional assessment, dynamic valuation, circulation path optimization to risk warning report generation, forming a complete ecosystem assessment system. This system can provide comprehensive technical support for the entire life cycle of data assets from ownership confirmation to transaction and financing, helping enterprises to realize data assetization in a complex international policy environment, and has significant advantages in system integrity. Attached Figure Description

[0022] Figure 1 A flowchart illustrating a collaborative evaluation method for e-commerce data assets according to an embodiment of the present invention; Figure 2 for Figure 1 Data preprocessing flowchart for step S10; Figure 3 for Figure 1 Flowchart of the classification process in step S10; Figure 4 for Figure 1 The flowchart of step S10, which uses natural language processing technology to parse trade contract text and extract data ownership information; Figure 5 for Figure 4 A microservice deployment architecture diagram for steps S120-S123; Figure 6 for Figure 1 Flowchart of step S20; Figure 7 for Figure 1 Flowchart of step S30; Figure 8 for Figure 1 Flowchart of step S40; Figure 9 for Figure 1 Flowchart of step S50; Figure 10 A schematic diagram of the structure of a collaborative evaluation system for e-commerce data assets provided in another embodiment of the present invention; In the attached figures, the following labels are used: S10~S50, S100~S123, S200~S202, S300~S304, S400~S401, S500~S502 - Steps; 2- Collaborative evaluation system for e-commerce data assets; 20-Data Acquisition and Recognition Module; 21-Multi-dimensional evaluation module; 22-Dynamic evaluation module; 23-Cooperative mapping module; 24 - Full Ecosystem Assessment Module. Detailed Implementation

[0023] # Check whether data assets comply with policy requirements compliance_score = self.check_compliance(data_asset,policy_requirements) # Generate a compliance gap analysis report gap_analysis = self.generate_gap_analysis(data_asset,policy_requirements) compliance_results[region] = { "compliance_score": compliance_score, "gap_analysis": gap_analysis } return compliance_results def dynamic_valuation(self, data_asset, policy_changes): # Dynamic valuation based on reinforcement learning valuation = self.dynamic_engine.calculate_value(data_asset,policy_changes) return valuation ``` In step S40: By analyzing the relationship between digital trade rules and the value of data assets, a quantitative mapping model between policy variables and data value is established. Based on a multi-dimensional value analysis framework for cross-border digital resource transactions, this model comprehensively considers multiple dimensions, including economic, technological, and social value, to systematically assess the impact of policy changes on the value of data assets. This model can optimize cross-border data flow paths, eliminate data flow barriers caused by policy differences, and improve cross-border data flow efficiency by more than 30%.

[0024] The quantitative mapping model between digital trade rules and the value of data assets is a semi-parametric space econometric model, such as... Figure 8 As shown, the construction process of the semi-parametric spatial econometric model includes the following steps: S400: Establishing a semi-parametric spatial econometric model between the value of data assets and the policy intensity vector: , in, Let be the value of the i-th data asset in period t, which can be a dynamic valuation; W is a cross-border data flow spatial weight matrix constructed based on trade flows; The policy intensity vector includes the data localization index, the digital services tax rate, and the cross-border data flow restriction index. The policy elasticity coefficient is obtained through GMM estimation; These are the spatial autoregressive coefficients; and For control variables; Individual effect; This is a time effect; This is the random error term; S401: Transform digital trade rules into structured policy parameters using natural language processing technology, inputting them into a semi-parametric spatial econometric model. Based on the model output, adjust the discount rate of logistics data assets, optimize cross-border data flow paths, and improve the efficiency of cross-border data asset flow to a preset efficiency improvement threshold.

[0025] Specifically, the logic for constructing collaborative mapping implementation is as follows: First, we completed the rule transformation and parsing, converting the RCEP, CPTPP, and GDPR clauses into structured policy parameters using NLP.

[0026] Then, a new efficiency improvement mechanism can be constructed. For example, based on the model output, in the scenario of overlapping "dual pilot" policies (cross-border e-commerce comprehensive pilot zone + free trade zone), the time for cross-border data compliance approval can be shortened by 40%. This parameter directly corrects the discount rate of logistics data assets, thereby increasing the value of data assets and ultimately achieving an efficiency improvement of ≥30% in cross-border data circulation.

[0027] Therefore, based on the collaborative mapping algorithm and the value measurement framework for cross-border transactions of digital resources, a simulation from macro policies to micro behaviors is achieved. The simplified Python implementation code is as follows: ``` # Cooperative mapping algorithm class SynergisticMappingAlgorithm: def __init__(self): self.policy_value_mappings = {} def build_mapping_model(self, policy_variables, data_values): # Constructing a mapping model between policy variables and data value # Using machine learning methods to learn the relationship between the two model = self.train_mapping_model(policy_variables, data_values) return model def optimize_data_flow(self, data_asset, source_region, target_region): # Analysis of policy differences between the two regions policy_gap = self.analyze_policy_gap(source_region, target_region) # Calculate the optimal flow path optimal_path = self.calculate_optimal_path(data_asset, policy_gap) # Estimated improvement in circulation efficiency efficiency_improvement = self.estimate_efficiency_improvement(optimal_path) return { "optimal_path": optimal_path, "efficiency_improvement": efficiency_improvement } ``` In step S50: Establish a comprehensive data asset transaction ecosystem assessment system and develop a risk early warning model for trade data asset pledge financing. Design a two-way feedback mechanism of "policy sandbox-technology platform" and build a cross-border data asset assessment alliance chain to achieve data mutual verification among regulatory nodes in multiple countries. This system draws on the practical experience of e-commerce companies in data asset valuation, combining multi-period excess return method and analytic hierarchy process to determine data asset value, providing comprehensive support for enterprises to include data assets in their financial statements and reducing the cost and risk for enterprises participating in data valuation. This system supports the intelligentization of the data asset transaction chain, thereby reducing enterprise financing costs by 10%-15%.

[0028] like Figure 9 As shown, generating an assessment report that includes the comprehensive value of data assets and risk warning levels also includes the following steps: S500: Invokes the full-ecosystem evaluation matrix for data asset transactions to comprehensively score data assets from multiple ecosystem dimensions; S501: Input the historical volatility sequence of data asset value, policy update frequency and corporate information change frequency into the Cox proportional risk regression model to predict the probability of default risk. When the risk probability exceeds the preset threshold, the pledge ratio is automatically reduced. S502: An evaluation report is generated through a two-way feedback mechanism between the policy sandbox and the technology platform. The downlink is used to block non-compliant requests, while the uplink is used to aggregate data to optimize rule parameters.

[0029] Specifically, firstly, an evaluation matrix of indicators is constructed. The data asset transaction ecosystem evaluation matrix in this invention includes quantitative indicators in the following four quadrants: compliance ecosystem, with cross-border compliance score and GDPR / CCPA matching degree ≥95% as the main measurement values; circulation ecosystem, with data call latency, cross-border bandwidth cost, and interoperability index as the main measurement values; financial ecosystem, with collateralization ratio cap, liquidity coverage ratio, and volatility index as the main measurement values; and technology ecosystem, with encryption strength, data lineage integrity, and sandbox test pass rate as the main measurement values. Secondly, a financing risk early warning model based on time-series analysis for predicting default probability was developed. The model adopted is the Cox proportional hazards regression model; the input features are the historical volatility sequence of data asset values, the update frequency of the policy negative list, and the frequency of changes in enterprise business registration information; the output is the predicted probability of default risk within the next 6 months and the corresponding early warning level (red / yellow / green). If the predicted probability exceeds a preset threshold (e.g., 0.15), the collateral ratio is automatically lowered and a margin call notice is triggered.

[0030] Finally, a two-way feedback mechanism between the policy sandbox and the technology platform is designed. The interactive technology principle includes a downlink (sandbox constraining the platform) and an uplink (platform feeding back to the sandbox). The downlink mainly uses the policy sandbox to issue smart contract rules (such as "prohibiting data from leaving the country"), while the technology platform uses the OPA (Open Policy Agent) engine to intercept non-compliant data asset assessment requests or transaction orders in real time. The uplink mainly uses the technology platform to collect compliance cost and performance bottleneck data during data circulation, and uses a federated learning framework to aggregate the data to regulatory nodes without disclosing the original data, providing data-driven decision-making basis for optimizing the policy sandbox rules (such as adjusting the whitelist).

[0031] Meanwhile, drawing on the practical experience of e-commerce companies in data asset valuation, and combining the multi-period excess return method, we provide data asset valuation services for enterprises. The simplified Python implementation code is as follows: ``` # Comprehensive Ecosystem Assessment System class FullEcosystemEvaluator: def __init__(self): self.assessment_matrix = DataAssetAssessmentMatrix() self.risk_warning_model = RiskWarningModel() self.feedback_mechanism = FeedbackMechanism() def ecosystem_assessment(self, data_asset, transaction_type): # Comprehensive Ecosystem Assessment assessment_result = self.assessment_matrix.comprehensive_assessment(data_asset) # Risk Warning risk_level = self.risk_warning_model.predict_risk(data_asset,transaction_type) # Feedback Mechanism feedback = self.feedback_mechanism.collect_feedback(data_asset) return { "assessment_result": assessment_result, "risk_level": risk_level, "feedback": feedback } def calculate_financing_benefit(self, data_asset): # Calculate the effect of reducing financing costs base_cost = self.estimate_base_financing_cost(data_asset) optimized_cost = self.estimate_optimized_financing_cost(data_asset) cost_reduction = (base_cost - optimized_cost) / base_cost return cost_reduction ``` To facilitate understanding of the above embodiments, a specific application scenario of the above embodiments will be used as an example for illustration below: Application scenario simulation: Data asset pledge financing for a cross-border e-commerce company in Qingdao, Shandong.

[0032] The scenario is as follows: Company A has transaction data and user profile data of the ASEAN market over the past 3 years and intends to pledge them to a bank for financing.

[0033] In the traditional assessment method, the company's data storage volume (TB) and general data quality are used for scoring and evaluation. The assessed value is about 2 million yuan. However, due to concerns about cross-border compliance risks (such as changes in policies such as ASEAN data localization and digital service tax rates), the bank only approved a credit line of 600,000 yuan.

[0034] The method of this invention is applied as follows: First, preprocessing and ownership resolution (NLP) are used to analyze the contract between the enterprise and the ASEAN agent to clarify that the enterprise has complete ownership and revenue rights to the data; second, policy-sensitive valuation is performed, using a reinforcement learning engine to capture the improved data flow facilitation within the region after the RCEP takes effect, and the valuation is increased by 8%; third, collaborative mapping optimization is performed, using a mapping algorithm to identify that using the "China-ASEAN Information Harbor" channel can avoid some local storage costs and improve circulation efficiency; fourth, a full-ecosystem early warning architecture is constructed, and the simulation analysis of this early warning model shows that the enterprise's supply chain data is stable and the probability of default is low; finally, the assessment results are output to guide the review and approval. For example, if the comprehensive assessment value is 3.5 million yuan, the bank will ultimately approve a loan amount of 2.8 million yuan based on the compliance matching degree (98%) and the regulatory endorsement provided by the two-way feedback mechanism.

[0035] The quantitative comparison between the method of the present invention and the traditional method through the above implementation methods is shown in Table 3 below: Table 3

[0036] The following are system embodiments corresponding to the above method embodiments. This embodiment can be implemented in conjunction with the above embodiments. The relevant technical details mentioned in the above embodiments are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.

[0037] like Figure 10 As shown, another embodiment of the present invention provides a collaborative evaluation system 2 for e-commerce data assets, employing the collaborative evaluation method for e-commerce data assets described in the above embodiments (such as...). Figure 1 ), and includes: The data acquisition and identification module 20 is used to acquire cross-border e-commerce data sources, construct a global commodity-data semantic mapping graph based on commodity-data mapping standards to classify the data sources, generate classification results, and apply natural language processing technology to parse trade contract texts to extract data ownership information and identify the data sources as data assets. The multi-dimensional evaluation module 21, connected to the data acquisition and identification module 20, is used to construct an evaluation index system that includes data quality dimension, application scenario dimension and market value dimension based on the classification results and the ownership information, and to calculate the initial evaluation value of the data assets. The dynamic evaluation module 22, connected to the multi-dimensional evaluation module 21, is used to input the real-time acquired current policy variable data and the initial evaluation value into the reinforcement learning-based dynamic valuation engine, output the value adjustment coefficient and correct the initial evaluation value to obtain the dynamic valuation. The collaborative mapping module 23, connected to the dynamic evaluation module 22, is used to establish a quantitative mapping model between digital trade rules and the value of data assets, and outputs the optimal cross-border data flow path based on dynamic valuation and current policy variable data. The full-ecosystem assessment module 24 and the connection and collaboration mapping module 23 are used to generate an assessment report that includes the comprehensive value of data assets and risk warning level based on the optimal cross-border data flow path and dynamic valuation.

[0038] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. With the deepening development of the digital economy, this invention has broad application prospects and promotional value. This invention is limited only by the claims and their full scope and equivalents.

Claims

1. A collaborative evaluation method for e-commerce data assets, characterized in that, Includes the following steps: The data sources of cross-border e-commerce are acquired, and a global commodity-data semantic mapping graph is constructed based on the commodity-data mapping standard to classify the data sources, generate classification results, and apply natural language processing technology to parse trade contract texts to extract the ownership information of the data, and identify the data sources as data assets. Based on the classification results and ownership information, an evaluation index system is constructed that includes data quality, application scenario, and market value dimensions, and the initial valuation value of the data assets is calculated. The real-time acquired current policy variable data and the initial assessment value are input into a reinforcement learning-based dynamic valuation engine, which outputs a value adjustment coefficient and corrects the initial assessment value to obtain a dynamic valuation. Establish a quantitative mapping model between digital trade rules and the value of the data assets, and output the optimal cross-border data flow path based on the dynamic valuation and the current policy variable data; Based on the optimal cross-border data flow path and the dynamic valuation, an assessment report is generated that includes the comprehensive value of the data assets and the risk warning level.

2. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The product-data mapping standard is the UN / CEFACT standard.

3. The collaborative evaluation method for e-commerce data assets according to claim 2, characterized in that, After obtaining the cross-border e-commerce data source, the method further includes the following steps: preprocessing the data source, including: Data cleaning is performed on the data source, including missing value imputation, abnormal transaction filtering, and data deduplication; The pre-trained multilingual BERT model maps product descriptions in different languages ​​to the UN / CEFACT standard terminology database and converts multiple currencies uniformly through a real-time exchange rate interface. To meet the compliance requirements of data security regulations in multiple jurisdictions, a dynamic desensitization engine is implemented to automatically perform differential privacy desensitization on data that fails the compliance check and replace it with simulated data that retains statistical characteristics. A stream processing framework is used to refresh the real-time valuation of the data assets.

4. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The process of constructing a global commodity-data semantic mapping graph based on the commodity-data mapping standard to classify the data source and generate classification results also includes the following steps: Extract the HS code from the customs declaration form, associate it with the business requirement specification model of the commodity-data mapping standard, and map the physical attributes of the commodity into data asset feature tags; Based on the supply chain reference data model of the commodity-data mapping standard, each key business node in the trade process is mapped to a data asset generation scenario; Based on the mapping map, the data assets are divided into multiple preset categories; A dynamic update mechanism is constructed using knowledge graph embedding technology. When the commodity-data standard is updated or new regional trade agreement clauses are added, the weight edges between graph nodes are automatically adjusted.

5. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The method of using natural language processing technology to parse trade contract text to extract ownership information also includes the following steps: A layout recognition model was used to locate the ownership-related sections in the trade contract text. Based on the legal knowledge graph, named entity recognition is performed in the ownership-related chapters to extract the data owner entity, data controller entity, use restriction entity, and revenue distribution entity respectively. Graph neural networks are used to detect logical conflicts between the clauses containing the entities described. For ownership items that are not involved in conflicts or are not explicitly stipulated in the aforementioned clauses, Bayesian probabilistic inference is performed in accordance with international rules and the laws of the target country to generate ownership conclusions with confidence levels.

6. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The construction of an evaluation index system that includes data quality, application scenario, and market value dimensions, and the calculation of the initial valuation value of the data assets, further includes the following steps: Qualitative evaluations of each dimension are transformed into cloud digital features, which include expectation, entropy, and hyperentropy. Then, a positive cloud generator is used to generate cloud droplet distributions corresponding to each dimension. The game theory-based combined weighting method is used to integrate subjective AHP weights with objective entropy weights to calculate the weight vectors for each dimension. Based on the cloud digital features of each dimension and the weight vector, the digital features of the integrated cloud are calculated by a weighted integrated cloud algorithm. Based on the certainty of the distribution of cloud droplets of each dimension at a preset value level, the initial evaluation value range is determined, so that the pricing error rate is controlled within a preset error threshold.

7. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The reinforcement learning-based dynamic evaluation engine employs a deep Q-network architecture and further includes the following steps: Define a state space, which includes an internal state and an external state, wherein the internal state contains the current benchmark value of the data asset, the data category identifier, and the ownership clarity score, and the external state is the current policy variable data; The value adjustment coefficient is discretized into multiple discrete actions within a preset coefficient range and a preset step size, forming an action space; A reward function is constructed based on the difference between the predicted value output by the dynamic valuation engine and the market transaction price of similar data assets, as well as the compliance score. The design incorporates an experience replay pool that stores a quadruple of state, action, reward, and next state within a preset time period, and prioritizes sampling samples with high temporal difference errors for training. When a sudden change in policy factors is detected, a new value adjustment coefficient is output within a preset forward propagation time by training a converged deep Q network, so that the compliance assessment of the data asset and the matching degree of the policies of major global economies reach a preset matching degree threshold.

8. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The quantitative mapping model between the digital trade rules and the value of the data assets is a semi-parametric spatial econometric model. The construction process of the semi-parametric spatial econometric model includes the following steps: Establish a semi-parametric spatial econometric model between the value of the data assets and the policy intensity vector: ， in, Let be the value of the i-th data asset in period t; W is the spatial weight matrix of cross-border data flow constructed based on trade flows; The policy intensity vector includes the data localization index, the digital services tax rate, and the cross-border data flow restriction index. The policy elasticity coefficient is obtained through GMM estimation; These are the spatial autoregressive coefficients; and For control variables; Individual effect; This is a time effect; This is the random error term; The digital trade rules and provisions are transformed into structured policy parameters using natural language processing technology and input into the semi-parametric spatial econometric model. Based on the model output, the discount rate of logistics data assets is adjusted to optimize the cross-border data flow path, so that the efficiency of cross-border flow of the data assets reaches a preset efficiency improvement threshold.

9. The collaborative evaluation method for e-commerce data assets according to claim 1, characterized in that, The process of generating an assessment report that includes the overall value of the data assets and the risk warning level also includes the following steps: The data asset transaction ecosystem assessment matrix is ​​invoked to comprehensively score the data assets from multiple ecosystem dimensions. The historical volatility sequence of data asset value, policy update frequency, and corporate information change frequency are input into the Cox proportional risk regression model to predict the probability of default risk. When the probability of risk exceeds a preset threshold, the pledge ratio is automatically reduced. The assessment report is generated through a two-way feedback mechanism between the policy sandbox and the technology platform, where the downlink is used to block non-compliant requests and the uplink is used to aggregate data to optimize rule parameters.

10. A collaborative evaluation system for e-commerce data assets, characterized in that, The collaborative evaluation method for e-commerce data assets according to any one of claims 1 to 9, and includes: The data acquisition and identification module is used to acquire cross-border e-commerce data sources, construct a global commodity-data semantic mapping graph based on commodity-data mapping standards to classify the data sources, generate classification results, and apply natural language processing technology to parse trade contract texts to extract data ownership information and identify the data sources as data assets. The multi-dimensional evaluation module is used to construct an evaluation index system that includes data quality dimension, application scenario dimension and market value dimension based on the classification results and ownership information, and to calculate the initial evaluation value of the data asset. The dynamic evaluation module is used to input the real-time acquired current policy variable data and the initial evaluation value into the reinforcement learning-based dynamic valuation engine, output the value adjustment coefficient and correct the initial evaluation value to obtain the dynamic valuation. The collaborative mapping module is used to establish a quantitative mapping model between digital trade rules and the value of the data assets, and output the optimal cross-border data flow path based on the dynamic valuation and the current policy variable data. The full-ecosystem assessment module is used to generate an assessment report that includes the comprehensive value of data assets and risk warning levels based on the optimal cross-border data flow path and the dynamic valuation.

Images