Full-process automatic decision-making intelligent processing method and system

Abstract

The invention relates to the technical field of data reasoning, and discloses a full-process automatic decision intelligent processing method and system, and the method comprises the steps: mapping heterogeneous fields in lease data to a data dictionary, and obtaining the application features of the lease data; generating a decision identifier of a lease application based on the application feature and a lease rule, and predicting an overdue confidence coefficient of the lease application; when the overdue confidence is in a preset threshold interval, triggering secondary verification of the lease rule on the decision identifier, and generating a collaborative decision result of the lease application; extracting common characteristics of the collaborative decision result, and caching the common characteristics to a shared memory pool; s2, updating the lease rule and the overdue confidence in S2 based on the shared memory pool, and generating an executable lease protocol of the target lease application based on the updated lease rule and the updated overdue confidence; according to the invention, the quality and speed of the whole-process automatic decision intelligent processing result can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data reasoning technology, and in particular to a full-process automated decision-making intelligent processing method and system. Background Art

[0002] In the field of data reasoning technology, fully automated decision-making and intelligent processing require efficient processing of information such as leasing data to generate reliable decision results. However, existing technologies lack standardized mapping for heterogeneous fields in leasing data, and their decision rule generation and overdue confidence prediction mechanisms are incomplete. This makes it difficult to accurately generate decision markers and assess risks, resulting in low-quality processing results.

[0003] When it comes to decision verification for lease applications, existing technologies lack an effective verification mechanism for overdue confidence levels within a threshold range. Furthermore, lease rules and overdue confidence levels cannot be dynamically updated based on historical processing results. The processing flow is not sufficiently automated or intelligent, resulting in a slow processing speed throughout the entire process, which cannot meet the requirements for decision-making efficiency in practical applications. Summary of the Invention

[0004] The present invention provides a full-process automated decision-making intelligent processing method and system, the main purpose of which is to solve the problems of low processing result quality and slow processing speed during full-process automated decision-making intelligent processing.

[0005] To achieve the above objectives, the present invention provides a full-process automated decision-making intelligent processing method, comprising: S1. Mapping heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data; S2. generating a decision identifier for the lease application based on the application characteristics and the lease rules, and predicting an overdue confidence level for the lease application; S3. When the overdue confidence level is within a preset threshold range, triggering the lease rule to perform a secondary verification of the decision identifier, and generating a collaborative decision result for the lease application; S4. Extracting common features of the collaborative decision-making results and caching them into a shared memory pool; S5. Update the lease rule and the overdue confidence level in S2 based on the shared memory pool, and generate an executable lease agreement for the target lease application based on the updated lease rule and the updated overdue confidence level.

[0006] In a preferred embodiment, mapping heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data includes: Identifying types of heterogeneous fields in the lease data; Match the standardized format corresponding to the type based on the data dictionary of the leasing field; The heterogeneous fields are converted into application features of the lease data according to the standardized format.

[0007] In a preferred embodiment, the data dictionary based on the leasing field matches the standardized format corresponding to the type, including: Extracting field constraints from the data dictionary; Verifying whether the heterogeneous fields comply with the field constraints; An exception identifier is added to the heterogeneous field that does not meet the constraint condition, and the heterogeneous field with the exception identifier is corrected according to the mapping between the heterogeneous field and the standardized format.

[0008] In a preferred embodiment, generating a decision identifier for a lease application based on the application features and lease rules, and predicting the overdue confidence level of the lease application, includes: Build a rule tree for the lease rule, taking the lease requirement in the lease rule as the parent node and the default condition as the child node; Matching the application features with the conditional branches of the rule tree, and using the state of the conditional branches that meet the requirements as the decision identifier of the lease application; The overdue confidence level of the rental application is generated according to the overdue risk in the application characteristics.

[0009] In a preferred embodiment, generating the overdue confidence level of the rental application based on the risk factors in the application characteristics includes: extracting risk features from the application features and encoding the risk features into a risk vector; Assigning weight factors to the risk characteristics based on the historical leasing records and historical leasing data of the leasing applicant; The overdue confidence level of the lease application is obtained by weighting and summing the risk vectors of different types based on the weight factors.

[0010] In a preferred embodiment, the weighting factor of the risk characteristic is assigned based on the historical rental records and historical rental data of the rental applicant, including: The weight factors include: the value fluctuation coefficient of the leased property and the credit change gradient of the lease applicant; When the value fluctuation coefficient exceeds the valuables threshold, the weight of the credit change gradient is increased; When the credit change gradient is lower than the credit critical point, the weight ratio of the value fluctuation coefficient is increased.

[0011] In a preferred embodiment, when the overdue confidence level is within a preset threshold range, triggering the lease rule to perform a secondary verification of the decision identifier and generating a collaborative decision result for the lease application includes: When the overdue confidence level is within a preset threshold range, obtaining the value fluctuation coefficient and the credit change gradient in the overdue confidence level; matching conflicting condition branches in the leasing rules based on the value fluctuation coefficient and the credit change gradient; Calling a multi-node decision terminal to vote and verify the conflict condition branch; The voting result is integrated with the decision identifier to generate a collaborative decision result for the lease application.

[0012] In a preferred embodiment, extracting common features of the collaborative decision-making results and caching them in a shared memory pool includes: Extract high-frequency decision path labels from vectorized collaborative decision-making results as common features; Encoding the common features into feature vector blocks of the lease application in time series; The feature vector block is written into a specified storage partition of a shared memory pool.

[0013] In a preferred embodiment, the updating of the lease rule and the overdue confidence in S2 based on the shared memory pool, and generating an executable lease agreement for the target lease application based on the updated lease rule and the updated overdue confidence, includes: Extracting the update frequency of the common features from the shared memory pool; An update amount of the lease rule is calculated based on the update frequency, wherein the calculation formula of the update amount is:

[0014] Where, is the update amount of the lease rule, is the risk loss sensitivity coefficient, is the historical overdue loss, For the said lease rules, is the business cost sensitivity coefficient, is the rule execution cost, For the said lease rules, is the symbol of partial derivative; adjusting the lease rule and the overdue confidence level according to the update amount; Generate an executable lease agreement for the target lease application based on the adjusted rules and thresholds.

[0015] In order to solve the above problems, the present invention also provides a full-process automated decision-making intelligent processing system, which includes: An application feature acquisition module, used to map heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data; A collaborative data generation module, configured to generate a decision identifier for a lease application based on the application characteristics and lease rules, and to predict an overdue confidence level for the lease application; A collaborative decision-making module, configured to trigger the secondary verification of the decision identifier by the lease rule when the overdue confidence level is within a preset threshold range, and generate a collaborative decision result for the lease application; A feature sharing module is used to extract common features of the collaborative decision-making results and cache them in a shared memory pool; A lease agreement generation module is configured to update the lease rules and the overdue confidence level in S2 based on the shared memory pool, and generate an executable lease agreement for the target lease application based on the updated lease rules and the updated overdue confidence level.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention maps heterogeneous fields in lease data to a data dictionary to generate standardized application features, ensuring data consistency and standardization, providing an accurate data foundation for subsequent decision-making, and thus improving the quality of processing results. Furthermore, based on application features and lease rules, decision identifiers are generated and overdue confidence levels are predicted. Combined with a secondary verification mechanism, collaborative decision-making results are generated, further enhancing the accuracy and reliability of decisions.

[0017] 2. This invention extracts common features from collaborative decision-making results and caches them in a shared memory pool. Based on these features, it updates lease rules and overdue confidence levels, enabling the system to continuously learn and optimize, achieving dynamic adjustments. This mechanism not only improves the quality of processing results but also accelerates the decision-making process through automated processes and dynamically updated rules. This speeds up the automated decision-making process across the entire process, enabling more efficient generation of executable lease agreements for target lease applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flowchart of a full-process automated decision-making intelligent processing method provided by one embodiment of the present invention; Figure 2 A functional module diagram of a full-process automated decision-making intelligent processing system provided by one embodiment of the present invention; The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0019] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] The embodiment of the present application provides a full-process automated decision-making intelligent processing method. The execution subject of the full-process automated decision-making intelligent processing method includes but is not limited to at least one of the electronic devices such as the server, the terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the full-process automated decision-making intelligent processing method can be executed by software or hardware installed on the terminal device or the server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and basic cloud computing services such as big data and artificial intelligence platforms.

[0021] Reference Figure 1 FIG. 1 is a flow chart of a full-process automated decision-making intelligent processing method provided by an embodiment of the present invention. In this embodiment, the full-process automated decision-making intelligent processing method includes: S1. Mapping heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data; In an embodiment of the present invention, mapping heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data includes: Identifying types of heterogeneous fields in the lease data; Match the standardized format corresponding to the type based on the data dictionary of the leasing field; The heterogeneous fields are converted into application features of the lease data according to the standardized format.

[0022] The data dictionary based on the leasing field matches the standardized format corresponding to the type, including: Extracting field constraints from the data dictionary; Verifying whether the heterogeneous fields comply with the field constraints; An exception identifier is added to the heterogeneous field that does not meet the constraint condition, and the heterogeneous field with the exception identifier is corrected according to the mapping between the heterogeneous field and the standardized format.

[0023] Specifically, the data samples in the field are analyzed to determine the inherent characteristics of the data. For numeric data, the determination is based on its numerical composition and operation properties; for text data, the determination is based on character combination and semantic characteristics; for date data, the identification is based on characteristics such as date format and range, thereby identifying the type of heterogeneous fields.

[0024] Furthermore, the identified field types are matched with a data dictionary in the leasing field, where standardized formats corresponding to various types are pre-stored.

[0025] Furthermore, the data dictionary is traversed, and the identified field types are compared one by one with the types recorded in the dictionary to find the type record that completely corresponds to it, thereby obtaining the standardized format corresponding to the type.

[0026] Furthermore, the heterogeneous fields are converted according to the obtained standardized format to generate application features of the leasing data.

[0027] Furthermore, according to the requirements of the standardized format, the data of heterogeneous fields are reorganized and formatted, such as converting date fields according to the specified date format, unifying the precision of numeric fields, etc., and finally forming the leasing data application characteristics that meet the requirements.

[0028] Specifically, records are read one by one from the data dictionary. Each record contains relevant information about the field. The relevant content of the field constraints is filtered out, and these constraints are extracted and sorted out to form a set containing all field constraints.

[0029] Furthermore, the heterogeneous fields are compared with the extracted field constraint condition set. For each heterogeneous field, its data content, data type, data length and other aspects are checked to see whether they meet the requirements specified in the corresponding field constraint condition, and whether each heterogeneous field meets the constraint condition is determined.

[0030] Furthermore, for heterogeneous fields that do not meet the constraints, specific exception identifiers are added to the field data to clearly distinguish that there is a problem with the field; then, based on the previously determined mapping relationship between the heterogeneous fields and the standardized format, and in accordance with the requirements of the standardized format, the data content, format, etc. of the heterogeneous fields with exception identifiers are modified and adjusted to make them meet the constraints.

[0031] In general, by mapping heterogeneous fields in leasing data to a data dictionary to obtain application features, the quality and speed of the entire process can be improved in terms of data standardization and exception handling.

[0032] In general, in terms of processing quality, the system first identifies heterogeneous field types, then matches the standardized format based on the leasing field data dictionary, and converts the fields into application features according to the specifications. This process extracts the field constraints in the data dictionary, verifies and corrects the fields that do not meet the constraints (adds exception identifiers and corrects them according to the mapping relationship), ensures the consistency and accuracy of the original data, and provides an unambiguous standardized data foundation for subsequent decision rule matching and risk assessment, avoiding decision deviations caused by confusing data formats, thereby improving the reliability of processing results.

[0033] In general, in terms of processing speed, the automated mapping mechanism of heterogeneous fields can quickly convert multi-source heterogeneous data into a unified format without the need for manual intervention in the data cleaning process.

[0034] In general, the predefined constraints and standardized mapping rules of the data dictionary support batch verification and conversion of fields, significantly shortening the time spent on data preprocessing.

[0035] In general, standardized application features can be directly used for subsequent rule tree matching and overdue confidence calculation, avoiding repeated parsing problems caused by inconsistent data formats, achieving efficient connection of the entire process from data access to feature generation, and improving the response speed of automated decision-making.

[0036] S2. generating a decision identifier for the lease application based on the application characteristics and the lease rules, and predicting an overdue confidence level for the lease application; In an embodiment of the present invention, generating a decision identifier for a lease application based on the application characteristics and lease rules, and predicting the overdue confidence level of the lease application, includes: Build a rule tree for the lease rule, taking the lease requirement in the lease rule as the parent node and the default condition as the child node; Matching the application features with the conditional branches of the rule tree, and using the state of the conditional branches that meet the requirements as the decision identifier of the lease application; The overdue confidence level of the rental application is generated according to the overdue risk in the application characteristics.

[0037] Generating the overdue confidence level of the lease application based on the risk factors in the application characteristics includes: extracting risk features from the application features and encoding the risk features into a risk vector; Assigning weight factors to the risk characteristics based on the historical leasing records and historical leasing data of the leasing applicant; The overdue confidence level of the lease application is obtained by weighting and summing the risk vectors of different types based on the weight factors.

[0038] Specifically, a tree-like data structure is created in a data processing software or programming environment, and the lease requirements in the lease rules are entered as the root node, ie, the parent node, of the rule tree.

[0039] Furthermore, the contents related to the breach of contract conditions are added as child nodes one by one under the parent node according to the hierarchical relationship. By setting the association relationship between the nodes, a complete lease rule tree is constructed, so that the entire lease rule is presented in the form of a tree structure.

[0040] Furthermore, the previously generated rental data application features are compared one by one with each conditional branch in the constructed rule tree.

[0041] Furthermore, starting from the root node of the rule tree, along the conditional path of each branch, check whether the application characteristics meet the condition requirements of each node. When the application characteristics meet the conditions of all nodes on a certain conditional branch, the state corresponding to the conditional branch is determined as the decision identifier of the lease application, thereby judging the feasibility status of the lease application.

[0042] Furthermore, the data related to overdue risk in the application characteristics are analyzed, and the possibility of overdue in the lease application is comprehensively assessed based on the historical overdue situation, repayment ability indicators, credit records and other relevant information reflected in the overdue risk data.

[0043] Furthermore, this possibility is converted into a corresponding overdue confidence level from low to high, which is used to quantify the possibility of overdue payment of the lease application in the future and serves as an important reference for leasing decisions.

[0044] Specifically, among the application characteristics of the leasing data, risk-related characteristic information is screened out, such as credit score, debt status, historical default records, etc.

[0045] Furthermore, these screened risk features are arranged in a certain order, and a unique code is assigned to each risk feature. In this way, all risk features are combined into a risk vector, so that the risk features are presented in the form of a vector.

[0046] Furthermore, the historical leasing records of the leasing applicant are collected, including the number of leases, the number of on-time repayments, the number of overdue repayments, etc., as well as relevant historical leasing data.

[0047] Furthermore, based on this historical data, we analyze the impact of each risk characteristic on overdue payments in previous leasing operations. Risk characteristics with greater impact are assigned a larger weighting factor, while those with lesser impact are assigned a smaller weighting factor. This is how we assign a corresponding weighting factor to each risk characteristic.

[0048] Furthermore, each risk feature in the encoded risk vector is multiplied by the corresponding weight factor.

[0049] Furthermore, all the multiplied results are added together. Through this weighted summation method of different types of risk vectors based on weight factors, a numerical value is finally obtained. This numerical value is the overdue confidence level of the lease application, which is used to measure the possibility of the lease application being overdue.

[0050] In general, by generating decision identification based on application characteristics and leasing rules and predicting overdue confidence, the quality and speed of the entire process can be improved from two aspects: rule-structured modeling and risk quantification assessment.

[0051] In general, in terms of processing quality, the system constructs a rule tree with the leasing requirements in the leasing rules as the parent node and the default conditions as the child nodes, accurately matching the application characteristics with the conditional branches of the rule tree, so that the generation of decision identifiers follows a systematic logical verification process, avoiding decision deviations caused by incomplete rule coverage.

[0052] In general, by extracting risk features from application characteristics and encoding them into risk vectors, combined with the historical leasing records of the leasing applicant and data allocation weight factors (such as the volatility coefficient of the leased property value and the credit change gradient), quantitative modeling of overdue risk is achieved, making the prediction results of overdue confidence more in line with the risk distribution of actual business scenarios, thereby improving the accuracy and reliability of decision-making results.

[0053] In general, in terms of processing speed, the structured design of the rule tree supports the rapid matching of application features, and through the parallel retrieval mechanism of conditional branches, the generation of decision identifiers can be completed in a short time.

[0054] In general, the encoding and weighted summation process of the risk vector adopts a standardized mathematical model (such as the dynamic adjustment mechanism of the weight factor). When the value volatility coefficient exceeds the threshold, the weight ratio of the credit change gradient is automatically increased, avoiding complex manual parameter adjustments and realizing the automatic and rapid calculation of overdue confidence.

[0055] In general, this mechanism of converting rule logic into tree structures and quantitative models that can be quickly executed by computers greatly shortens the processing time from data input to risk assessment, and improves the response efficiency of automated decision-making throughout the entire process.

[0056] S3. When the overdue confidence level is within a preset threshold range, triggering the lease rule to perform a secondary verification of the decision identifier, and generating a collaborative decision result for the lease application; In an embodiment of the present invention, when the overdue confidence level is within a preset threshold range, triggering the lease rule to perform a secondary verification of the decision identifier and generating a collaborative decision result for the lease application includes: When the overdue confidence level is within a preset threshold range, obtaining the value fluctuation coefficient and the credit change gradient in the overdue confidence level; matching conflicting condition branches in the leasing rules based on the value fluctuation coefficient and the credit change gradient; Calling a multi-node decision terminal to vote and verify the conflict condition branch; The voting result is integrated with the decision identifier to generate a collaborative decision result for the lease application.

[0057] Specifically, it is determined whether the overdue confidence level is within a pre-set threshold range. If it is within the range, the value fluctuation coefficient and credit change gradient are extracted from the overdue confidence level related data.

[0058] Furthermore, the value fluctuation coefficient reflects the changes in the value of the leased property or leasing business, and the credit change gradient reflects the changing trend of the applicant's credit status. These two data are obtained through specific data reading and separation operations.

[0059] Furthermore, the extracted value fluctuation coefficient and credit change gradient are compared with all conflicting condition branches in the leasing rules.

[0060] Furthermore, the conditions set for each conflict condition branch are checked one by one to see whether the value fluctuation coefficient and credit change gradient meet these conditions, and the conflict condition branch that matches them is found to determine the conflict rules related to the current lease application.

[0061] Furthermore, a multi-node decision terminal is enabled, and the matched conflict condition branches are sent to each decision node of the terminal.

[0062] Furthermore, each decision node independently analyzes and judges the conflicting condition branch, gives a judgment result such as agreement or disagreement, and then feeds back its own judgment result, thereby completing the voting verification process for the conflicting condition branch.

[0063] Furthermore, the voting results of the multi-node decision terminals are integrated with the previously determined decision identifier of the lease application.

[0064] Furthermore, the conflict situation reflected by the voting results and the basic application feasibility represented by the decision mark are comprehensively considered and processed according to certain logic and rules, and finally a comprehensive lease application collaborative decision result is generated to clarify whether the lease application is approved and the corresponding processing method.

[0065] In general, by triggering the leasing rules to conduct secondary verification of the decision identification and generate collaborative decision-making results, the quality and speed of the entire process can be improved in terms of accurate risk identification and optimization of the decision-making mechanism.

[0066] In general, in terms of processing quality, when the overdue confidence level is within the preset threshold range, the system obtains the value fluctuation coefficient and credit change gradient, matches the conflicting condition branches in the leasing rules, and uses the voting verification mechanism of the multi-node decision-making terminal to perform multi-dimensional cross-validation of potential risk points.

[0067] In general, this mechanism avoids the limitations of a single decision-making model. By integrating multi-node decision results with initial decision identification, it can correct misjudgments caused by data bias or insufficient rule coverage, making collaborative decision results more in line with actual risk scenarios, thereby improving the accuracy and reliability of decisions.

[0068] In general, in terms of processing speed, the secondary verification process adopts an automated conflict condition matching and multi-node parallel voting mechanism, which can quickly complete decision review without human intervention.

[0069] In general, the system automatically triggers the verification process based on preset rules, and multi-node terminals perform voting verification simultaneously, significantly shortening the decision-making time in risk dispute scenarios.

[0070] In general, the generation process of collaborative decision-making results and the integration of initial decision identification are automatically completed by the system, avoiding the time loss of traditional serial approval processes and achieving efficient connection of the entire process from risk warning to decision confirmation, thereby improving the processing speed of automated decision-making throughout the entire process.

[0071] S4. Extracting common features of the collaborative decision-making results and caching them into a shared memory pool; In an embodiment of the present invention, extracting common features of the collaborative decision-making results and caching them in a shared memory pool includes: Extract high-frequency decision path labels from vectorized collaborative decision-making results as common features; Encoding the common features into feature vector blocks of the lease application in time series; The feature vector block is written into a specified storage partition of a shared memory pool.

[0072] Specifically, the vectorized collaborative decision-making results are analyzed, and the frequency of occurrence of each decision path label is counted.

[0073] Furthermore, we find out several decision path labels that appear most frequently, and extract these decision path labels that appear frequently as common features reflecting the decision-making situation of leasing applications. These common features can reflect the general rules and characteristics of the leasing application decision-making process.

[0074] Furthermore, the extracted common features are encoded in sequence according to the chronological order of the lease applications.

[0075] Furthermore, a unique coding identifier is assigned to each common feature, and then these codes are arranged and combined in sequence according to the time series to form an ordered set. This set is the feature vector block of the lease application, which contains the common decision-making feature information of the lease application at different time points.

[0076] Furthermore, a shared memory pool is found, and a designated storage partition for storing the rental application characteristic information is determined.

[0077] Furthermore, through the data writing operation, the generated feature vector block is completely stored in the designated storage partition, so that other subsequent systems or programs can quickly and conveniently access and call the feature information of these rental applications to achieve data sharing and utilization.

[0078] In general, by extracting the common features of collaborative decision-making results and caching them in a shared memory pool, the quality and speed of the entire process can be improved from the aspects of data reuse and system optimization.

[0079] In general, in terms of processing quality, the system uses the high-frequency decision path labels in the vectorized collaborative decision-making results as common features, and stores them after encoding them into feature vector blocks through time series. This can extract the core patterns of historical high-quality decisions and provide reusable experience templates for subsequent leasing rule updates and overdue confidence predictions.

[0080] In general, this mechanism enables the decision-making process for new applications to be deduced based on the common logic of historical successful cases, reducing decision-making errors caused by rule lags or risk assessment deviations, thereby improving the accuracy and reliability of processing results.

[0081] In general, in terms of processing speed, the high-speed storage and retrieval characteristics of the shared memory pool can realize real-time calling of common features.

[0082] In general, after the feature vector block is written to the specified storage partition, the system does not need to repeatedly parse historical decision data, but directly reads high-frequency features from the memory for rule matching and risk calculation, avoiding the IO delay caused by traditional disk storage.

[0083] In general, the caching of common features provides real-time data support for the update of leasing rules and overdue confidence levels, which can quickly respond to changes in business scenarios, shorten the data processing time in the decision-making chain, achieve efficiency optimization of the entire process from feature extraction to agreement generation, and significantly improve the response speed of automated decision-making.

[0084] S5. Update the lease rule and the overdue confidence level in S2 based on the shared memory pool, and generate an executable lease agreement for the target lease application based on the updated lease rule and the updated overdue confidence level.

[0085] In an embodiment of the present invention, updating the lease rule and the overdue confidence in S2 based on the shared memory pool, and generating an executable lease agreement for the target lease application based on the updated lease rule and the updated overdue confidence, includes: Extracting the update frequency of the common features from the shared memory pool; An update amount of the lease rule is calculated based on the update frequency, wherein the calculation formula of the update amount is:

[0086] Where, is the update amount of the lease rule, is the risk loss sensitivity coefficient, is the historical overdue loss, For the said lease rules, is the business cost sensitivity coefficient, is the rule execution cost, For the said lease rules, is the symbol of partial derivative; adjusting the lease rule and the overdue confidence level according to the update amount; Generate an executable lease agreement for the target lease application based on the adjusted rules and thresholds.

[0087] Specifically, a designated storage partition of the shared memory pool is accessed to read the stored common feature related data.

[0088] Furthermore, by analyzing the number of changes or update records of these data within a certain period of time, the update frequency of the common features, that is, the number of times the common features are updated per unit time, is determined.

[0089] Furthermore, based on the determined update frequency of common features, the extent to which the lease rules need to be updated is evaluated.

[0090] Furthermore, a high update frequency means that the leasing business environment or application situation changes frequently, which correspondingly increases the update amount of the leasing rules; a low update frequency indicates that there are fewer changes, which reduces the update amount of the leasing rules, and the update amount of the leasing rules is calculated accordingly.

[0091] Furthermore, the existing lease rule content is modified according to the calculated lease rule update amount.

[0092] Furthermore, new terms are added, terms that are no longer applicable are deleted, or the content of existing terms is adjusted; at the same time, based on the adjustments to the leasing rules and actual business needs, the assessment methods and standards for overdue confidence levels are modified accordingly to make the assessment of overdue confidence levels more in line with the adjusted leasing rules.

[0093] Furthermore, the adjusted lease rules and overdue confidence threshold are applied to the target lease application.

[0094] Furthermore, based on the specific circumstances of the target lease application and combined with the adjusted rules and thresholds, an executable lease agreement is generated that includes the rights and obligations of both parties to the lease, the lease term, the rent payment method, etc., to ensure that the content of the agreement complies with the latest lease rules and risk assessment standards.

[0095] Specifically, the sources of the parameters in the calculation formula of the update amount are: the risk loss sensitivity coefficient is determined by the historical risk assessment data of the leasing business, and the specific value is set by business experts in combination with industry standards and corporate risk preferences by analyzing the impact of historical risk loss changes on the adjustment of leasing rules; historical overdue losses are taken from the loss data caused by overdue leasing transactions in the past recorded in the leasing system, and are statistically summarized by time period; the business cost sensitivity coefficient is calculated based on the company's operating cost data, measuring the sensitivity of changes in rule execution costs to leasing rule adjustments, and is jointly evaluated and determined by the finance department and the business department; the rule execution cost is the various costs incurred in the actual implementation of the leasing rules, including manpower, system maintenance, compliance review and other costs, which are obtained through the cost accounting process; the leasing rules are a set of currently effective leasing business rules, including various leasing requirements, default conditions and other contents, which are directly obtained from the leasing rules database.

[0096] Furthermore, the formula means that the amount of lease rule updates consists of two parts. One part is the product of the risk loss sensitivity coefficient and the rate of change of historical overdue losses on lease rules, which reflects the driving effect of risk loss changes on rule updates. The other part is the product of the business cost sensitivity coefficient and the rate of change of rule execution costs on lease rules, which reflects the impact of business cost changes on rule updates. By adding these two parts, the degree to which lease rules need to be updated can be comprehensively quantified, providing a quantitative basis for rule adjustments.

[0097] Furthermore, the trend of the formula is that when the rate of change of historical overdue losses to lease rules increases, under the condition that the risk loss sensitivity coefficient remains unchanged, the corresponding product term increases, which will increase the amount of lease rule updates, indicating that the greater the change in risk losses, the more the rules need to be updated; if the rate of change of rule execution costs to lease rules increases, and the business cost sensitivity coefficient remains unchanged, the other product term increases, which will also lead to an increase in the amount of updates, indicating that the more significant the cost changes, the higher the demand for rule updates; conversely, if these two change rates decrease, the amount of updates will also decrease accordingly. When the risk losses and business costs change less, the amount of lease rule updates will also be smaller.

[0098] In general, the process of updating leasing rules and overdue confidence levels based on a shared memory pool and generating executable agreements can improve processing quality and speed from the perspectives of data-driven optimization and automated processes.

[0099] In general, by extracting the common characteristics of collaborative decision-making results in the shared memory pool, the system can dynamically adjust leasing rules based on historical high-quality decision-making models, and combine the formulaic calculation of risk loss sensitivity coefficient and business cost sensitivity coefficient to make rule updates more accurately match actual business scenarios, thereby reducing overdue risk assessment deviations and improving the accuracy and reliability of decision-making results.

[0100] In general, the shared memory pool's caching mechanism can synchronize the update frequency of common features in real time, automatically complete the iterative optimization of rules and thresholds, and avoid process delays caused by manual intervention. At the same time, the updated rules can be directly used to generate target lease agreements, shortening the decision-making chain, and achieving efficiency improvements in the entire process from data processing to agreement generation, significantly accelerating the response speed of automated decision-making.

[0101] like Figure 2 , which is a functional module diagram of a full-process automated decision-making intelligent processing system provided by one embodiment of the present invention.

[0102] The fully automated decision-making intelligent processing system 100 described in the present invention can be installed in an electronic device. Depending on the functionality implemented, the fully automated decision-making intelligent processing system 100 may include an application feature acquisition module 101, a collaborative data generation module 102, a collaborative decision module 103, a feature sharing module 104, and a lease agreement generation module 105. A module, also referred to as a unit, is a series of computer program segments that can be executed by an electronic device processor and perform a fixed function, and is stored in the electronic device's memory.

[0103] In this embodiment, the functions of each module / unit are as follows: The application feature acquisition module 101 is used to map heterogeneous fields in the lease data to a data dictionary to obtain the application features of the lease data; The collaborative data generation module 102 is configured to generate a decision identifier for a lease application based on the application characteristics and lease rules, and to predict the overdue confidence level of the lease application; The collaborative decision module 103 is configured to trigger the secondary verification of the decision identifier by the lease rule when the overdue confidence level is within a preset threshold range, and generate a collaborative decision result for the lease application; The feature sharing module 104 is used to extract common features of the collaborative decision-making results and cache them in a shared memory pool; The lease agreement generation module 105 is configured to update the lease rules and the overdue confidence level in S2 based on the shared memory pool, and generate an executable lease agreement for the target lease application based on the updated lease rules and the updated overdue confidence level.

[0104] In the several embodiments provided by the present invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the module division is merely a logical function division, and other division methods may be used in actual implementation.

[0105] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0106] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0107] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0108] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A full-process automated decision-making intelligent processing method, characterized in that: The method comprises: S1. Mapping heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data; S2. generating a decision identifier for the lease application based on the application characteristics and the lease rules, and predicting an overdue confidence level for the lease application; S3. When the overdue confidence level is within a preset threshold range, triggering the lease rule to perform a secondary verification of the decision identifier, and generating a collaborative decision result for the lease application; S4. Extracting common features of the collaborative decision-making results and caching them into a shared memory pool; S5. Update the lease rule and the overdue confidence level in S2 based on the shared memory pool, and generate an executable lease agreement for the target lease application based on the updated lease rule and the updated overdue confidence level.

2. The full-process automated decision-making intelligent processing method according to claim 1, characterized in that: Mapping heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data includes: Identifying types of heterogeneous fields in the lease data; Match the standardized format corresponding to the type based on the data dictionary of the leasing field; The heterogeneous fields are converted into application features of the lease data according to the standardized format.

3. The full-process automated decision-making intelligent processing method according to claim 2, characterized in that: The data dictionary based on the leasing field matches the standardized format corresponding to the type, including: Extracting field constraints from the data dictionary; Verifying whether the heterogeneous fields comply with the field constraints; An exception identifier is added to the heterogeneous field that does not meet the constraint condition, and the heterogeneous field with the exception identifier is corrected according to the mapping between the heterogeneous field and the standardized format.

4. The full-process automated decision-making intelligent processing method according to claim 1, characterized in that: Generating a decision identifier for the lease application based on the application characteristics and the lease rules, and predicting the overdue confidence level of the lease application, includes: Build a rule tree for the lease rule, taking the lease requirement in the lease rule as the parent node and the default condition as the child node; Matching the application features with the conditional branches of the rule tree, and using the state of the conditional branches that meet the requirements as the decision identifier of the lease application; The overdue confidence level of the rental application is generated according to the overdue risk in the application characteristics.

5. The full-process automated decision-making intelligent processing method according to claim 4, characterized in that: Generating the overdue confidence level of the lease application based on the risk factors in the application characteristics includes: extracting risk features from the application features and encoding the risk features into a risk vector; Assigning weight factors to the risk characteristics based on the historical leasing records and historical leasing data of the leasing applicant; The overdue confidence level of the lease application is obtained by weighting and summing the risk vectors of different types based on the weight factors.

6. The full-process automated decision-making intelligent processing method according to claim 5, characterized in that: The weight factor of the risk characteristic is assigned based on the historical lease record and historical lease data of the lease applicant, including: The weight factors include: the value fluctuation coefficient of the leased property and the credit change gradient of the lease applicant; When the value fluctuation coefficient exceeds the valuables threshold, the weight of the credit change gradient is increased; When the credit change gradient is lower than the credit critical point, the weight ratio of the value fluctuation coefficient is increased.

7. The full-process automated decision-making intelligent processing method according to claim 6, characterized in that: When the overdue confidence level is within a preset threshold range, triggering the lease rule to perform a secondary verification on the decision identifier and generating a collaborative decision result for the lease application includes: When the overdue confidence level is within a preset threshold range, obtaining the value fluctuation coefficient and the credit change gradient in the overdue confidence level; matching conflicting condition branches in the leasing rules based on the value fluctuation coefficient and the credit change gradient; Calling a multi-node decision terminal to vote and verify the conflict condition branch; The voting result is integrated with the decision identifier to generate a collaborative decision result for the lease application.

8. The full-process automated decision-making intelligent processing method according to claim 1, characterized in that: The extracting common features of the collaborative decision-making results and caching them in a shared memory pool includes: Extract high-frequency decision path labels from vectorized collaborative decision-making results as common features; Encoding the common features into feature vector blocks of the lease application in time series; The feature vector block is written into a specified storage partition of a shared memory pool.

9. The full-process automated decision-making intelligent processing method according to claim 1, characterized in that: The updating of the lease rule and the overdue confidence in S2 based on the shared memory pool, and generating an executable lease agreement for the target lease application based on the updated lease rule and the updated overdue confidence, includes: Extracting the update frequency of the common features from the shared memory pool; An update amount of the lease rule is calculated based on the update frequency, wherein the calculation formula of the update amount is: ， Where, is the update amount of the lease rule, is the risk loss sensitivity coefficient, is the historical overdue loss, For the said lease rules, is the business cost sensitivity coefficient, is the rule execution cost, For the said lease rules, is the symbol of partial derivative; adjusting the lease rule and the overdue confidence level according to the update amount; Generate an executable lease agreement for the target lease application based on the adjusted rules and thresholds.

10. A full-process automated decision-making intelligent processing system, characterized in that: The system comprises: An application feature acquisition module, used to map heterogeneous fields in the lease data to a data dictionary to obtain application features of the lease data; A collaborative data generation module, configured to generate a decision identifier for a lease application based on the application characteristics and lease rules, and to predict an overdue confidence level for the lease application; A collaborative decision-making module, configured to trigger the secondary verification of the decision identifier by the lease rule when the overdue confidence level is within a preset threshold range, and generate a collaborative decision result for the lease application; A feature sharing module is used to extract common features of the collaborative decision-making results and cache them in a shared memory pool; A lease agreement generation module is configured to update the lease rules and the overdue confidence level in S2 based on the shared memory pool, and generate an executable lease agreement for the target lease application based on the updated lease rules and the updated overdue confidence level.

Images