Data parallel processing system for enterprise credit rating

Abstract

The invention relates to the field of data processing, in particular to a data parallel processing system for enterprise credit rating, and the system comprises a target analysis module which is used for determining the correlation analysis strategy of each target analysis material and the material type of each target analysis material; the node analysis module is used for responding to the key material proportion of the node analysis condition to determine whether each target analysis node performs task scheduling analysis or not; the material analysis module is used for responding to the strategy judgment condition to determine a material analysis strategy of each target analysis material; the first execution module is used for responding to the adjustment judgment condition to determine whether the index extraction coefficient of the target analysis material is adjusted or not; and the second execution module is used for responding to the combination judgment condition to determine an extraction combination mode of the target analysis material, and obtaining a target extraction index according to the combination extraction condition corresponding to each target extraction combination, and the data analysis efficiency in the enterprise credit rating process is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and particularly to a data parallel processing system for enterprise credit rating. Background Art

[0002] In the process of enterprise credit rating, it is necessary to evaluate the business situation of the enterprise. This process often requires analyzing a large number of business materials. Among them, the process of extracting abnormal indicators from the business materials has a great impact on the accuracy of the evaluation results of the business situation. In addition, the specific business situations of different enterprises are different. Using a single index evaluation standard to evaluate the abnormal degree of indicators is likely to interfere with the judgment of abnormal indicators. Therefore, how to further screen abnormal indicators according to relevant materials to improve the accuracy and processing efficiency of the credit evaluation process is an urgent problem to be solved by those skilled in the art.

[0003] Chinese Patent Application Publication No. CN116308810A discloses an enterprise credit rating method, device, electronic device and storage medium. The method includes: determining a credit rating model for an enterprise to be rated according to the industry type to which the enterprise to be rated belongs. Among them, the credit rating model has corresponding data usage base periods and data indicators. The data usage base period is the starting statistical date when the enterprise to be rated is credit rated. Obtaining the initial business data corresponding to the enterprise to be rated according to the data usage base period, and processing the initial business data according to the data indicators to obtain index description information, where the index description information is used to describe the data index score situation of the enterprise to be rated, and determining the enterprise credit grade of the enterprise to be rated according to the index description information. However, the above solution has the following problems: It fails to effectively screen abnormal indicators according to the correlation between the data to be analyzed according to actual needs, resulting in some abnormal indicators obtained not matching the actual business situation of the enterprise, and further leading to low data processing efficiency in the credit evaluation process. Summary of the Invention

[0004] Therefore, the present invention provides a data parallel processing system for enterprise credit rating to overcome the problem in the prior art that abnormal indicators cannot be effectively screened according to the correlation between the data to be analyzed according to actual needs, resulting in some abnormal indicators obtained not matching the actual business situation of the enterprise, and further leading to low data processing efficiency in the credit evaluation process.

[0005] To achieve the above object, the present invention provides a data parallel processing system for enterprise credit rating, which is characterized by including:

[0006] A target analysis module, which is used to respond to relevant analysis conditions to determine the relevant analysis strategies for each target analysis material, and respond to the key index ratio and material correlation coefficient of the target judgment condition to determine the material category of each target analysis material;

[0007] A node analysis module, which is connected to the target analysis module, and is used to respond to the key material ratio of the node analysis condition to determine whether to perform task scheduling analysis on each target analysis node;

[0008] A material analysis module, which is connected to the target analysis module, and is used to respond to the strategy determination condition to determine the material analysis strategy for each target analysis material. The material analysis strategy is to detect the difference value of the key index parameters for the target analysis material or determine the extraction combination method according to the relevant analysis strategy of the target analysis material;

[0009] A first execution module, which is connected to the material analysis module, and is used to respond to the adjustment determination condition to determine whether to adjust the index extraction coefficient for the target analysis material to obtain the target extraction index;

[0010] A second execution module, which is connected to the material analysis module, and is used to respond to the combination determination condition to determine the extraction combination method of the target analysis material, and obtain the target extraction index according to the combination extraction conditions corresponding to each target extraction combination. The extraction combination method includes determining the target extraction combination according to the coincidence reference coefficient, and determining the target extraction combination according to the reference period overlap degree.

[0011] Furthermore, the target analysis module responds to the target judgment condition to determine the material category of each target analysis material;

[0012] For a single target analysis material,

[0013] The target judgment condition responded by the target analysis module is that the key index ratio is greater than the preset key index ratio, and it is determined that the target analysis material is a type I analysis material;

[0014] The target judgment condition responded by the target analysis module is that the key index ratio is less than or equal to the preset key index ratio and the material correlation coefficient is greater than the preset material correlation coefficient, and it is determined that the target analysis material is a type II analysis material;

[0015] The material correlation coefficient is the number of relevant analysis materials for each target analysis material.

[0016] Furthermore, the target analysis module responds to the relevant analysis conditions to determine the relevant analysis strategies for each target analysis material;

[0017] For a single target analysis material,

[0018] The relevant analysis condition responded by the target analysis module is that the proportion of the coincidence index is greater than the preset proportion of the coincidence index, and it is determined that the relevant analysis materials of the target analysis material are determined according to the material coincidence coefficient.

[0019] The relevant analysis condition responded by the target analysis module is that the proportion of the coincidence index is less than or equal to the preset proportion of the coincidence index and the historical reference coefficient is greater than the preset historical reference coefficient, and it is determined that the relevant analysis materials of the target analysis material are determined according to the reference matching degree and the cycle overlap degree.

[0020] Further, the node analysis module responds to the proportion of key materials in the node analysis condition to determine whether to perform task scheduling analysis on each target analysis node.

[0021] The node analysis condition responded by the node analysis module is that the proportion of key materials of a target analysis node is greater than the preset proportion of key materials, and it is determined that task scheduling analysis is performed on the target analysis node.

[0022] The key materials are a type of analysis materials and a second type of analysis materials.

[0023] Further, the node analysis module responds to the scheduling analysis condition to perform scheduling node matching on the materials to be allocated for the target analysis node.

[0024] The node analysis module determines the materials to be allocated for the target analysis node according to the relevant reference values.

[0025] The node analysis module determines the scheduling priority coefficient of each node to be matched according to the proportion of key materials and the proportion of relevant materials, and determines the scheduling node of each material to be allocated based on the scheduling priority coefficient.

[0026] The scheduling analysis condition is that the node analysis module determines to perform task scheduling analysis on a target analysis node.

[0027] Further, the material analysis module responds to the policy determination condition to determine the material analysis strategy of each target analysis material.

[0028] The policy determination condition responded by the material analysis module is that the target analysis material is a type of analysis materials, and it is determined to detect the difference value of the key index parameters of the target analysis material.

[0029] The policy determination condition responded by the material analysis module is that the target analysis material is a second type of analysis materials, and it is determined to determine the extraction combination method according to the relevant analysis strategy of the target analysis material.

[0030] Further, the first execution module responds to the first analysis condition and determines whether to adjust the index extraction coefficient of the target analysis material according to the difference value of the key index proportion.

[0031] The adjustment determination condition for which the first execution module responds is that the difference value of the key index ratio is greater than the preset difference value of the key index ratio, then it is determined to adjust the index extraction coefficient for the target analysis material according to the key index ratio of various categories of key indicators;

[0032] The first analysis condition is that the material analysis module determines to detect the difference value of the key index parameters for a target analysis material.

[0033] Further, the second execution module responds to the second analysis condition, and the second execution module responds to the combination determination condition to determine the extraction combination method of the target analysis material;

[0034] The combination determination condition for which the second execution module responds is to determine the relevant analysis materials of the target analysis material according to the index coincidence coefficient, then it is determined to determine the target extraction combination according to the coincidence reference coefficient;

[0035] The combination determination condition for which the second execution module responds is to determine the relevant analysis materials of the target analysis material according to the reference matching degree and the cycle overlap degree, then it is determined to determine the target extraction combination according to the reference cycle overlap degree;

[0036] The second analysis condition is that the material analysis module determines the extraction combination method of a target analysis material according to the relevant analysis strategy.

[0037] Further, the second execution module responds to the first combination extraction condition and determines the key analysis indicators of the target extraction combination according to the index radiation reference value;

[0038] Determine the abnormal extraction coefficient of each key analysis indicator according to the difference value of the reference abnormal coefficient of the associated analysis material, and record the key analysis indicators with the abnormal extraction coefficient greater than the preset abnormal extraction coefficient as the target extraction indicators;

[0039] The coincidence reference coefficient of each target analysis material within any target extraction combination determined according to the coincidence reference coefficient is greater than the preset coincidence coefficient;

[0040] The first combination extraction condition is that the target extraction combination is determined according to the coincidence reference coefficient.

[0041] Further, the second execution module responds to the second combination extraction condition and performs matching abnormal analysis on the target extraction combination;

[0042] Determine the target extraction indicators of the target extraction combination according to the difference value of the abnormal coefficient, and record the key indicators with the difference value of the abnormal coefficient greater than the preset difference value of the abnormal coefficient as the target extraction indicators;

[0043] The reference period overlap degree of each target analysis material within any target extraction combination determined according to the reference period overlap degree is greater than the preset reference period overlap degree;

[0044] The second combination extraction condition is determined by a target extraction combination according to the reference period overlap degree.

[0045] Compared with the prior art, the beneficial effect of the present invention is that in the technical solution of the present invention, the target analysis module determines the category of each target analysis material according to the key index ratio and the material correlation coefficient, and determines a targeted material analysis strategy based on the category of the target analysis material, so as to improve the effectiveness of the index extraction result, avoid the obtained index being affected by a single judgment condition, resulting in redundant data in the obtained index data, and further resulting in low efficiency of subsequent data analysis. The present invention improves the data analysis efficiency in the enterprise credit rating process.

[0046] Further, in the present invention, it is determined whether to schedule the target analysis materials to be processed corresponding to the corresponding target analysis node according to the key material ratio of each target analysis node, so as to avoid uneven distribution of the analysis tasks of the key indicators that each target analysis node needs to complete during the data analysis process, and determine the scheduling node based on the key material ratio and the relevant material ratio, ensuring the uniformity of the tasks assigned to each target analysis node and avoiding a large number of material scheduling requirements during the analysis process. The present invention improves the overall analysis efficiency of the target analysis node.

[0047] Further, in the present invention, it is determined whether to adjust the index extraction coefficient of the corresponding type of analysis material according to the difference value of the key index ratio of each type of analysis material. Since when analyzing each target analysis material, the uniformity of the index distribution is likely to affect the extraction process of the key indicators, resulting in key indicators with a relatively small proportion being easily ignored, the balance of the index extraction process is ensured by adjusting the index extraction coefficients corresponding to various key indicators, thereby improving the effectiveness of the index extraction result.

[0048] Further, in the present invention, different target extraction combination division methods and analysis methods for each target extraction combination are determined according to the determination method of the relevant analysis materials of each type II analysis material, so that the analysis method for obtaining the target extraction index is more in line with the actual material situation, ensuring the accuracy of the screening result. By combining the past business materials of the target analysis enterprise to re-screen the key indicators, the present invention avoids the low effectiveness of the obtained key indicators due to evaluating the indicators only based on a single evaluation criterion, and improves the data analysis efficiency in the enterprise credit rating process. Brief Description of the Drawings

[0049] Figure 1It is a module connection diagram of the data parallel processing system for enterprise credit rating of the present invention;

[0050] Figure 2 It is a flowchart of the present invention's target analysis module for determining the material categories of each target analysis material based on the proportion of key indicators and material correlation coefficients in response to the target judgment conditions;

[0051] Figure 3 It is a flowchart of the present invention's node analysis module for determining whether to perform task scheduling analysis on each target analysis node based on the proportion of key materials in response to the node analysis conditions;

[0052] Figure 4 It is a flowchart of the present invention's material analysis module for determining the material analysis strategy of each target analysis material in response to the policy determination conditions. Detailed implementation manners

[0053] In order to make the objectives and advantages of the present invention clearer, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0054] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0055] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0056] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0057] Please refer to Figures 1 to 4 As shown, the present invention provides a data parallel processing system for enterprise credit rating, including:

[0058] A target analysis module, which is used to respond to relevant analysis conditions to determine the relevant analysis strategies for each target analysis material, and respond to the key index ratio and material correlation coefficient of the target judgment conditions to determine the material categories of each target analysis material;

[0059] A node analysis module, which is connected to the target analysis module, and is used to respond to the key material ratio of the node analysis conditions to determine whether to perform task scheduling analysis for each target analysis node;

[0060] A material analysis module, which is connected to the target analysis module, and is used to respond to the strategy determination conditions to determine the material analysis strategies for each target analysis material. The material analysis strategy is to detect the difference value of the key index parameters for the target analysis material or determine the extraction combination method according to the relevant analysis strategy of the target analysis material;

[0061] A first execution module, which is connected to the material analysis module, and is used to respond to the adjustment determination conditions to determine whether to adjust the index extraction coefficient for the target analysis material to obtain the target extraction index;

[0062] A second execution module, which is connected to the material analysis module, and is used to respond to the combination determination conditions to determine the extraction combination method of the target analysis material, and obtain the target extraction index according to the combination extraction conditions corresponding to each target extraction combination. The extraction combination method includes determining the target extraction combination according to the coincidence reference coefficient, and determining the target extraction combination according to the reference period overlap degree.

[0063] Among them, the present invention is applied to the data analysis of enterprise credit rating to effectively screen the data in the relevant analysis materials to avoid the problem of low data analysis efficiency of the relevant analysis materials. The enterprise undergoing enterprise credit rating is recorded as the target analysis enterprise, and the materials on which the enterprise credit rating of the target analysis enterprise is based are recorded as the target analysis materials. Each target analysis material is correspondingly marked with the project execution time and the material generation time. The project execution time is the start time of the enterprise project corresponding to the target analysis material, and the material generation time is the completion time of the editing of the target analysis material;

[0064] Extract various evaluation indicators included in each target analysis material. In the present invention, a preset index range is correspondingly set for each type of evaluation indicator. If an evaluation indicator is not within its corresponding preset index range, then this evaluation indicator is recorded as a key indicator. The types of evaluation indicators in the present invention include, but are not limited to: operating income growth rate, return on net assets, cash flow, net profit growth rate, capital turnover rate, inventory turnover days, accounts receivable turnover days, gross profit margin, and return on shareholders' equity. How to extract the evaluation indicators included in the target analysis material and how to set the preset index ranges for various evaluation indicators are easily understood by those skilled in the art and will not be elaborated here. For each key indicator, an anomaly coefficient detection is performed. For a single key indicator, its anomaly coefficient has a positive correlation with the range difference parameter, and the range difference parameter is the minimum value of the absolute value of the difference between the value of the key indicator and any value within its corresponding preset index range, which has a positive correlation.

[0065] In the present invention, there are several historical analysis records. Any one historical analysis record records at least once the proportion of key indicators, material correlation coefficient, the number of target analysis materials where the key indicators exist, the execution interval duration corresponding to each reference analysis material, the proportion of overlapping indicators, historical reference coefficient, material overlap coefficient, reference correlation coefficient, proportion of key materials, difference value of key indicator proportion, difference value of indicator proportion, overlapping reference coefficient, reference period overlap degree, index radiation reference value, anomaly extraction coefficient, anomaly coefficient difference value, and relevant reference value during the data analysis process of the target analysis material of a target analysis enterprise. And each historical analysis record corresponds to a qualified mark, and the qualified mark records whether the data processing efficiency in the credit assessment process meets the user's requirements.

[0066] Specifically, the target analysis module responds to the target judgment condition to determine the material category of each target analysis material;

[0067] For a single target analysis material,

[0068] If the target judgment condition to which the target analysis module responds is that the proportion of key indicators is greater than the preset proportion of key indicators, it is determined that the target analysis material is a type of analysis material;

[0069] If the target judgment condition to which the target analysis module responds is that the proportion of key indicators is less than or equal to the preset proportion of key indicators and the material correlation coefficient is greater than the preset material correlation coefficient, it is determined that the target analysis material is a type of two analysis material;

[0070] The material correlation coefficient is the number of relevant analysis materials of each target analysis material.

[0071] Among them, the material categories in the present invention include first-class analysis materials, second-class analysis materials, and conventional analysis materials. The conventional analysis materials are target analysis materials with the proportion of key indicators less than or equal to the preset proportion of key indicators and the material correlation coefficient less than or equal to the preset material correlation coefficient. For a single target analysis material, the proportion of key indicators = the number of key indicators included in the target analysis material / the number of evaluation indicators included in the target analysis material. The values of the preset proportion of key indicators and the preset material correlation coefficient can be determined by the user according to the actual working scenario. For example, the user can set them according to historical analysis records. The higher the user's requirement for the data processing efficiency in the credit assessment process, the smaller the value of the preset proportion of key indicators and the smaller the value of the preset material correlation coefficient. A method for obtaining the value of the preset proportion of key indicators is provided. The minimum value of the proportion of key indicators of the first-class analysis materials in the historical analysis records that meet the user's requirement for the data processing efficiency in the credit assessment process is recorded as the preset proportion of key indicators. A value of the preset proportion of key indicators is provided, and the value of the preset proportion of key indicators is 0.6. A method for obtaining the value of the preset material correlation coefficient is provided. The minimum value of the material correlation coefficient of the second-class analysis materials in the historical analysis records that meet the user's requirement for the data processing efficiency in the credit assessment process is recorded as the preset material correlation coefficient.

[0072] Specifically, the target analysis module responds to relevant analysis conditions to determine the relevant analysis strategies for each target analysis material;

[0073] For a single target analysis material,

[0074] When the relevant analysis condition responded by the target analysis module is that the proportion of overlapping indicators is greater than the preset proportion of overlapping indicators, it is determined that the relevant analysis materials of the target analysis material are determined according to the material overlap coefficient;

[0075] When the relevant analysis condition responded by the target analysis module is that the proportion of overlapping indicators is less than or equal to the preset proportion of overlapping indicators and the historical reference coefficient is greater than the preset historical reference coefficient, it is determined that the relevant analysis materials of the target analysis material are determined according to the reference matching degree and the cycle overlap degree.

[0076] Among them, for a single target analysis material, the coincidence index ratio = the number of coincidence indicators included in the target analysis material / the number of evaluation indicators included in the target analysis material. For any category of key indicators, if the number of target analysis materials with key indicators of this category is greater than the preset material quantity, then the key indicators of this category are recorded as coincidence indicators. The value of the preset material quantity can be determined by the user according to the actual working scenario. For example, the user can set it according to historical analysis records. A method for obtaining the value of the preset material quantity is provided. The historical analysis records of the relevant analysis materials of the target analysis material determined according to the reference matching degree and the cycle overlap degree are recorded as relevant reference records. The average value of the quantities of the relevant analysis materials of each coincidence indicator in the relevant reference records that meet the user's data processing efficiency requirements for the credit assessment process is recorded as the preset material quantity;

[0077] For a single target analysis material, the historical reference coefficient is the number of historical associated materials of the target analysis material. Target analysis projects with an interval duration between the project execution time and the project execution time corresponding to the target analysis material greater than the preset execution interval duration are recorded as reference analysis materials. If a reference analysis material has the same category of key indicators as the target analysis material, then the reference analysis material is recorded as the historical associated material of the target analysis material. The value of the preset execution interval duration can be determined by the user according to the actual working scenario. For example, the user can set it according to historical analysis records. A method for obtaining the value of the preset execution interval duration is provided. The average value of the execution interval durations corresponding to each reference analysis material in the historical analysis records that meet the user's data processing efficiency requirements for the credit assessment process is recorded as the preset execution interval duration;

[0078] The values of the preset coincidence index ratio and the preset historical reference coefficient can be determined by the user according to the actual working scenario. For example, the user can set them according to historical analysis records. A method for obtaining the value of the preset coincidence index ratio is provided. The maximum value of the coincidence index ratios of each target analysis material in the relevant reference records that meet the user's data processing efficiency requirements for the credit assessment process is recorded as the preset coincidence index ratio. A method for obtaining the value of the preset historical reference coefficient is provided. The minimum value of the historical reference coefficients of each target analysis material in the relevant reference records that meet the user's data processing efficiency requirements for the credit assessment process is recorded as the preset historical reference coefficient;

[0079] For the target analysis materials with the proportion of coincidence index greater than the preset proportion of coincidence index, the relevant analysis materials are determined according to the material coincidence coefficient. For a single target analysis material, the target analysis materials with the material coincidence coefficient greater than the preset material coincidence coefficient with respect to this target analysis material are recorded as relevant analysis materials. The material coincidence coefficient between any two target analysis materials is the number of categories of key indicators that exist in both of the above two target analysis materials. The value of the preset material coincidence coefficient can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirement for the data processing efficiency in the credit assessment process, the larger the value of the preset material coincidence coefficient. A method for determining the value of the preset material coincidence coefficient is provided. The historical analysis records of determining the relevant analysis materials of the target analysis materials according to the material coincidence coefficient are recorded as the first relevant reference records, and the minimum value of the material coincidence coefficients of the relevant analysis materials in the first relevant reference records that meet the user's requirement for the data processing efficiency in the credit assessment process is recorded as the preset material coincidence coefficient;

[0080] The target analysis module determines relevant analysis materials based on the reference matching degree and the cycle overlap degree for target analysis materials where the coincidence index ratio is less than or equal to the preset coincidence index ratio and the historical reference coefficient is greater than the preset historical reference coefficient. For a single target analysis material, the reference correlation coefficient between each target analysis material and this target analysis material is determined based on the reference matching degree and the cycle overlap degree. The reference correlation coefficient is the sum of the products of the reference matching degree and the cycle overlap degree and their corresponding influence coefficients respectively. Target analysis materials with a reference correlation coefficient greater than the preset reference correlation coefficient are recorded as the relevant analysis materials of this target analysis material. For any two target analysis materials, the reference matching degree = the number of categories of key indicators jointly included in the above two target analysis materials / the number of categories of key indicators included in the above two target analysis materials, and the cycle overlap degree = the maximum value of the execution cycle durations of the above two target analysis materials / the difference value of the execution cycle durations of the above two target analysis materials. The execution cycle duration is the interval duration between the project execution time corresponding to the target analysis material and the material generation time. Users can set the values of the influence coefficients corresponding to the reference matching degree and the cycle overlap degree according to the actual work scenario. A set of values for the influence coefficients corresponding to the reference matching degree and the cycle overlap degree is provided. The value of the influence coefficient corresponding to the reference matching degree is 0.5, and the value of the influence coefficient corresponding to the cycle overlap degree is 0.5. The value of the preset reference correlation coefficient can be determined by users according to the actual work scenario. For example, users can set it according to historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the larger the value of the preset reference correlation coefficient. A method for determining the value of the preset reference correlation coefficient is provided. The historical analysis record of determining relevant analysis materials based on the reference matching degree and the cycle overlap degree is recorded as the second relevant reference record, and the minimum value of the reference correlation coefficients of the relevant analysis materials in the second relevant reference records that meet the user's requirement for the data processing efficiency of the credit assessment process is recorded as the preset reference correlation coefficient.

[0081] Specifically, the node analysis module responds to the proportion of key materials in the node analysis condition to determine whether task scheduling analysis is to be performed for each target analysis node;

[0082] The node analysis condition responded by the node analysis module is that the proportion of key materials of a target analysis node is greater than the preset proportion of key materials, and it is determined that task scheduling analysis is to be performed for this target analysis node;

[0083] The key materials are type-one analysis materials and type-two analysis materials.

[0084] Among them, there are several target analysis nodes in the present invention. During the process of conducting enterprise credit rating for the target analysis enterprise, the same number of target analysis materials are correspondingly set for each target analysis node, and each target analysis node processes the corresponding target analysis materials in parallel. For a single target analysis node, the proportion of key materials = the number of key materials that need to be processed by this target analysis node / the number of target analysis materials that need to be processed by this target analysis node. The value of the preset proportion of key materials can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the smaller the value of the preset proportion of key materials. A method for obtaining the value of the preset proportion of key materials is provided. The minimum value of the proportion of key materials of each target analysis node that conducts task scheduling analysis in the historical analysis records that meet the user's requirement for the data processing efficiency of the credit assessment process is recorded as the preset proportion of key materials. A value of the preset proportion of key materials is provided, and the value of the preset proportion of key materials is 0.4.

[0085] Specifically, the node analysis module responds to the scheduling analysis condition and performs scheduling node matching for the materials to be allocated for this target analysis node;

[0086] The node analysis module determines the materials to be allocated for this target analysis node according to relevant reference values;

[0087] The node analysis module determines the scheduling priority coefficients of each node to be matched according to the proportion of key materials and the proportion of relevant materials, and determines the scheduling nodes of each material to be allocated based on the scheduling priority coefficients;

[0088] The scheduling analysis condition is that the node analysis module determines to conduct task scheduling analysis for a target analysis node.

[0089] Among them, for a single target analysis material, the relevant reference value is the number of relevant analysis materials existing in the same target analysis node. For any target analysis node that needs to conduct task scheduling analysis, the target analysis materials with relevant reference values less than the preset relevant reference value are recorded as the materials to be allocated for this target analysis node. The value of the preset relevant reference value can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the larger the value of the preset relevant reference value. A method for obtaining the value of the preset relevant reference value is provided. The average value of the relevant reference values of each material to be allocated in the historical analysis records that meet the user's requirement for the data processing efficiency of the credit assessment process is recorded as the preset relevant reference value. A value of the preset relevant reference value is provided, and the value of the preset relevant reference value is 3;

[0090] For a single material to be allocated, determine the scheduling priority coefficient of each node to be matched according to the proportion of key materials and the proportion of related materials. The node to be matched is the target analysis node whose proportion of key materials is less than or equal to the preset proportion of key materials. The proportion of related materials = the number of related analysis materials of this material to be allocated that the node to be analyzed needs to process / the number of target analysis materials that the node to be analyzed needs to process. The scheduling priority coefficient = ln(proportion of related materials / proportion of key materials). Select the node to be matched with the smallest scheduling priority coefficient and record it as the scheduling node of this material to be allocated, that is, record this material to be allocated as the target analysis material that the scheduling node needs to process.

[0091] Specifically, the material analysis module responds to the policy determination condition to determine the material analysis strategy of each target analysis material.

[0092] The policy determination condition to which the material analysis module responds is that the target analysis material is a type of analysis material, then it is determined to detect the difference value of the key index parameters for this target analysis material.

[0093] The policy determination condition to which the material analysis module responds is that the target analysis material is a type of two analysis material, then it is determined to determine the extraction combination method according to the related analysis strategy of this target analysis material.

[0094] Specifically, the first execution module responds to the first analysis condition and determines whether to adjust the index extraction coefficient of the target analysis material according to the difference value of the key index proportion.

[0095] The adjustment determination condition to which the first execution module responds is that the difference value of the key index proportion is greater than the preset difference value of the key index proportion, then it is determined to adjust the index extraction coefficient of the target analysis material according to the key index proportion of each category of key indexes.

[0096] The first analysis condition is that the material analysis module determines to detect the difference value of the key index parameters for a target analysis material.

[0097] Among them, for any target analysis material that needs to detect the difference value of the key index parameters, the difference value of the key index proportion Let \(n\) be the number of categories of key indicators contained in the target analysis material, \(D_i\) be the proportion of key indicators of the \(i\)-th category of key indicators in the target analysis material, and \(D_0\) be the reference proportion of indicators. The reference proportion of indicators is the average of the proportions of key indicators of each category of key indicators in the target analysis material. For a single category of key indicators, the proportion of key indicators = the number of key indicators of this category in the target analysis material / the number of key indicators in the target analysis material. The value of the preset difference value of the proportion of key indicators can be determined by the user according to the actual working scenario. For example, the user can set it according to historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the smaller the value of the preset difference value of the proportion of key indicators. A method for obtaining the value of the preset difference value of the proportion of key indicators is provided. The historical analysis record of adjusting the index extraction coefficient of the target analysis material according to the reference value of each key indicator is recorded as the difference reference record. The minimum value of the difference value of the proportion of key indicators of the target analysis material in the difference reference records that meet the user's requirement for the data processing efficiency of the credit assessment process is recorded as the preset difference value of the proportion of key indicators;

[0098] If the index extraction coefficient of a target analysis material needs to be adjusted, then the corresponding index extraction coefficient is adjusted according to the proportion of key indicators of each category. Among them, if the absolute value of the difference value of the proportion of key indicators of a category of key indicators is greater than the preset proportion difference value, then the key indicators of this category are recorded as adjustment indicators. The difference value of the proportion of key indicators of any category of key indicators = the proportion of key indicators of this category of key indicators - the reference proportion of indicators. If the difference value of the proportion of key indicators of the adjustment indicator is greater than 0, then a decreasing adjustment is made to the index extraction coefficient of the key indicators of this category. The decreasing value of the index extraction coefficient is positively correlated with the absolute value of the difference value of the proportion of key indicators. If the difference value of the proportion of key indicators of the adjustment indicator is less than 0, then an increasing adjustment is made to the index extraction coefficient of the key indicators of this category. The increasing value of the index extraction coefficient is positively correlated with the absolute value of the difference value of the proportion of key indicators. The value of the preset proportion difference value can be determined by the user according to the actual working scenario. For example, the user can set it according to historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the smaller the value of the preset proportion difference value. A method for obtaining the value of the preset proportion difference value is provided. The minimum value of the absolute value of the difference value of the proportion of key indicators of each adjustment indicator in the difference reference records that meet the user's requirement for the data processing efficiency of the credit assessment process is recorded as the preset proportion difference value;

[0099] The indicator extraction coefficient is the degree of attention paid by the indicator extraction model to the key indicators of each category during the key indicator extraction process for the target analysis material. The indicator extraction coefficients of the key indicators of each category are balanced to avoid poor recognition rate for the key indicator categories with a small proportion of key indicators. The indicator extraction model is a model trained for key indicator extraction. The key indicators extracted by the model after adjusting the indicator extraction coefficient are marked as target extraction indicators. How to train the indicator extraction model and how to adjust the indicator extraction coefficients of the key indicators of each category are easy for technicians in this field to understand and will not be elaborated here.

[0100] Specifically, the second execution module is responsive to the second analysis condition, and the second execution module is responsive to the combination determination condition to determine the extraction combination mode of the target analysis material;

[0101] The combination determination condition responded by the second execution module is to determine the relevant analysis materials of the target analysis material according to the index coincidence coefficient, and then determine the target extraction combination according to the coincidence reference coefficient;

[0102] The combination determination condition responded by the second execution module is to determine the relevant analysis materials of the target analysis material according to the reference matching degree and the period overlap, and then determine to determine the target extraction combination according to the reference period overlap;

[0103] The second analysis condition is that the material analysis module determines an extraction combination method of a target analysis material according to a relevant analysis strategy.

[0104] Among them, if the relevant analysis materials of a target analysis material are determined according to the indicator overlap coefficient, the target analysis material and its relevant analysis materials are divided according to the overlap reference coefficient to determine the target extraction combination, and the overlap reference coefficients of each target analysis material in any determined target extraction combination are greater than the preset overlap coefficient. For a single target analysis material, the overlap reference coefficient is the average value of the material overlap coefficients between the target analysis material and other target analysis materials in the target extraction combination to which it belongs. The value of the preset overlap coefficient can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirements for the data processing efficiency of the credit evaluation process, the larger the value of the preset overlap coefficient. A method for setting the value of the preset overlap coefficient is provided, and the historical analysis record of the target extraction combination determined according to the overlap reference coefficient is recorded as the first combination analysis record, and the minimum value of the overlap reference coefficient in the first combination analysis record that meets the user's requirements for the data processing efficiency of the credit evaluation process is recorded as the preset overlap coefficient;

[0105] If the relevant analysis materials of a target analysis material are determined according to the reference matching degree and the cycle overlap degree, then the target analysis material and its relevant analysis materials are divided according to the reference cycle overlap degree to determine the target extraction combination. The reference cycle overlap degree of each target analysis material within any determined target extraction combination is greater than the preset reference cycle overlap degree. For a single target analysis material, the reference cycle overlap degree is the average value of the cycle overlap degrees between this target analysis material and other target analysis materials within the target extraction combination to which it belongs. The value of the preset reference cycle overlap degree can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirement for the data processing efficiency in the credit assessment process, the larger the value of the preset reference cycle overlap degree. A method for determining the value of the preset reference cycle overlap degree is provided. The historical analysis records of determining the target extraction combination according to the reference cycle overlap degree are recorded as the second combined analysis records, and the minimum value of the reference cycle overlap degree in the second combined analysis records that meet the user's requirement for the data processing efficiency in the credit assessment process is recorded as the preset reference cycle overlap degree.

[0106] Specifically, the second execution module responds to the first combined extraction condition and determines the key analysis indicators of the target extraction combination according to the index radiation reference value;

[0107] Determine the abnormal extraction coefficients of each key analysis indicator according to the reference abnormal coefficient difference value of the associated analysis materials, and record the key analysis indicators with the abnormal extraction coefficients greater than the preset abnormal extraction coefficient as the target extraction indicators;

[0108] The coincidence reference coefficients of each target analysis material within any target extraction combination determined according to the coincidence reference coefficient are greater than the preset coincidence coefficient;

[0109] The first combined extraction condition is determined by the target extraction combination according to the coincidence reference coefficient.

[0110] Among them, the combined extraction conditions include the first combined extraction condition and the second combined extraction condition. For a target extraction combination determined according to the coincidence reference coefficient, the index radiation reference value of the key indicators of each category is the number of target analysis materials containing the corresponding key indicators within this target extraction combination. Record the key indicators with the index radiation reference value greater than the preset index radiation reference value as the key analysis indicators. The value of the preset index radiation reference value can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirement for the data processing efficiency in the credit assessment process, the larger the value of the preset index radiation reference value. A method for determining the value of the preset index radiation reference value is provided. The minimum value of the index radiation reference value of the key analysis indicators in the first combined analysis records that meet the user's requirement for the data processing efficiency in the credit assessment process is recorded as the preset index radiation reference value;

[0111] For a single key analysis indicator, the target analysis material containing this key analysis indicator is denoted as the associated analysis material. For a single associated analysis material, the average value of the anomaly coefficients of the key indicators contained therein is denoted as the reference anomaly coefficient of this associated analysis material. The difference value between the anomaly extraction coefficient and the reference anomaly coefficient of this key analysis indicator has a positive correlation. The difference value of the reference anomaly coefficient m is the number of associated analysis materials of this key analysis indicator, yj is the reference anomaly coefficient of the j-th associated analysis material of this key analysis indicator, y0 is the average value of the reference anomaly coefficients of the associated analysis materials containing this key analysis indicator. The value of the preset anomaly extraction coefficient can be determined by the user according to the actual working scenario. For example, the user can set it according to historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the smaller the value of the preset anomaly extraction coefficient. A method for obtaining the value of the preset anomaly extraction coefficient is provided. The minimum value of the anomaly extraction coefficients of the target extraction indicators in the first combined analysis records that meet the user's requirement for the data processing efficiency of the credit assessment process is denoted as the preset anomaly extraction coefficient.

[0112] Specifically, the second execution module responds to the second combined extraction condition and performs matching anomaly analysis on the target extraction combination;

[0113] Determine the target extraction indicators of the target extraction combination according to the difference value of the anomaly coefficients, and denote the key indicators with the difference value of the anomaly coefficients greater than the preset difference value of the anomaly coefficients as the target extraction indicators;

[0114] The reference period overlap degree of each target analysis material within any target extraction combination determined according to the reference period overlap degree is greater than the preset reference period overlap degree;

[0115] The second combined extraction condition is determined by a target extraction combination according to the reference period overlap degree.

[0116] Among them, for a target extraction combination determined according to the reference period overlap degree, detect the difference value of the anomaly coefficients of each key indicator. For the key indicators of a single category, the difference value of the anomaly coefficients Let \(s\) be the number of target analysis materials containing the key indicators of this category within the target extraction combination, \(xv\) be the abnormal average coefficient of the \(v\)-th target analysis material containing this key indicator within the target extraction combination, and \(x0\) be the average value of the abnormal average coefficients of each target analysis material containing this key indicator within the target extraction combination. The abnormal average coefficient is the average value of the abnormal coefficients of this key indicator contained in the target analysis material. The value of the preset abnormal coefficient difference can be determined by the user according to the actual working scenario. For example, the user can set it according to the historical analysis records. The higher the user's requirement for the data processing efficiency of the credit assessment process, the smaller the value of the preset abnormal coefficient difference. A method for obtaining the value of the preset abnormal coefficient difference is provided, and the minimum value of the abnormal coefficient difference of the target extraction index in the second combined analysis record that meets the user's requirement for the data processing efficiency of the credit assessment process is recorded as the preset abnormal coefficient difference.

[0117] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0118] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data parallel processing system for enterprise credit rating, characterized in that Including: A target analysis module, which is used to respond to relevant analysis conditions to determine the relevant analysis strategies for each target analysis material, and respond to the key index ratio of the target judgment condition and the material correlation coefficient to determine the material category of each target analysis material; A node analysis module, which is connected to the target analysis module, and is used to respond to the key material ratio of the node analysis condition to determine whether to perform task scheduling analysis for each target analysis node; A material analysis module, which is connected to the target analysis module, and is used to respond to the strategy determination condition to determine the material analysis strategy for each target analysis material. The material analysis strategy is to detect the difference value of the key index parameters for the target analysis material or determine the extraction combination method according to the relevant analysis strategy of the target analysis material; A first execution module, which is connected to the material analysis module, and is used to respond to the adjustment determination condition to determine whether to adjust the index extraction coefficient for the target analysis material to obtain the target extraction index; A second execution module, which is connected to the material analysis module, and is used to respond to the combination determination condition to determine the extraction combination method of the target analysis material, and obtain the target extraction index according to the combination extraction conditions corresponding to each target extraction combination. The extraction combination method includes determining the target extraction combination according to the coincidence reference coefficient, and determining the target extraction combination according to the reference period overlap degree.

2. The data parallel processing system for enterprise credit rating according to claim 1, wherein The target analysis module responds to the target judgment condition to determine the material category of each target analysis material; For a single target analysis material, The target judgment condition responded by the target analysis module is that the key index ratio is greater than the preset key index ratio, and it is determined that the target analysis material is a first-class analysis material; The target judgment condition responded by the target analysis module is that the key index ratio is less than or equal to the preset key index ratio and the material correlation coefficient is greater than the preset material correlation coefficient, and it is determined that the target analysis material is a second-class analysis material; The material correlation coefficient is the number of relevant analysis materials of each target analysis material.

3. The data parallel processing system for enterprise credit rating according to claim 2, wherein The target analysis module responds to the relevant analysis conditions to determine the relevant analysis strategies for each target analysis material; For a single target analysis material, The relevant analysis condition responded by the target analysis module is that the coincidence index ratio is greater than the preset coincidence index ratio, and it is determined to determine the relevant analysis material of the target analysis material according to the material coincidence coefficient; The relevant analysis condition responded by the target analysis module is that the coincidence index ratio is less than or equal to the preset coincidence index ratio and the historical reference coefficient is greater than the preset historical reference coefficient, and it is determined to determine the relevant analysis material of the target analysis material according to the reference matching degree and the period overlap degree.

4. The data parallel processing system for enterprise credit rating according to claim 3, wherein The node analysis module responds to the key material ratio of the node analysis condition to determine whether to perform task scheduling analysis for each target analysis node; The node analysis condition responded by the node analysis module is that the key material ratio of a target analysis node is greater than the preset key material ratio, and it is determined to perform task scheduling analysis for the target analysis node; The key materials are first-class analysis materials and second-class analysis materials.

5. The data parallel processing system for enterprise credit rating according to claim 4, wherein The node analysis module responds to the scheduling analysis condition to perform scheduling node matching for the materials to be allocated for the target analysis node; The node analysis module determines the material to be allocated for the target analysis node according to relevant reference values; The node analysis module determines the scheduling priority coefficients of each node to be matched according to the proportion of key materials and the proportion of relevant materials, and determines the scheduling nodes of each material to be allocated based on the scheduling priority coefficients; The scheduling analysis condition is that the node analysis module determines to perform task scheduling analysis for a target analysis node.

6. The data parallel processing system for enterprise credit rating according to claim 5, wherein The material analysis module responds to the policy determination condition to determine the material analysis strategy of each target analysis material; The policy determination condition for which the material analysis module responds is that the target analysis material is a type of analysis material, then it is determined to detect the difference value of the key index parameters for the target analysis material; The policy determination condition for which the material analysis module responds is that the target analysis material is a type of two analysis materials, then it is determined to determine the extraction combination method according to the relevant analysis strategy of the target analysis material.

7. The data parallel processing system for enterprise credit rating according to claim 6, wherein The first execution module responds to the first analysis condition and determines whether to adjust the index extraction coefficient of the target analysis material according to the difference value of the key index proportion; The adjustment determination condition for which the first execution module responds is that the difference value of the key index proportion is greater than the preset difference value of the key index proportion, then it is determined to adjust the index extraction coefficient of the target analysis material according to the key index proportion of each category of key indexes; The first analysis condition is that the material analysis module determines to detect the difference value of the key index parameters for a target analysis material.

8. The data parallel processing system for enterprise credit rating according to claim 7, wherein The second execution module responds to the second analysis condition, and the second execution module responds to the combination determination condition to determine the extraction combination method of the target analysis material; The combination determination condition for which the second execution module responds is that the relevant analysis material of the target analysis material is determined according to the index coincidence coefficient, then it is determined to determine the target extraction combination according to the coincidence reference coefficient; The combination determination condition for which the second execution module responds is that the relevant analysis material of the target analysis material is determined according to the reference matching degree and the cycle overlap degree, then it is determined to determine the target extraction combination according to the reference cycle overlap degree; The second analysis condition is that the material analysis module determines to determine the extraction combination method of a target analysis material according to the relevant analysis strategy.

9. The data parallel processing system for enterprise credit rating according to claim 8, wherein The second execution module responds to the first combination extraction condition and determines the key analysis indexes of the target extraction combination according to the index radiation reference value; Determine the abnormal extraction coefficients of each key analysis index according to the difference value of the reference abnormal coefficient of the associated analysis material, and record the key analysis indexes with the abnormal extraction coefficient greater than the preset abnormal extraction coefficient as the target extraction indexes; The coincidence reference coefficients of each target analysis material within any target extraction combination determined according to the coincidence reference coefficient are all greater than the preset coincidence coefficient; The first combination extraction condition is that the target extraction combination is determined according to the coincidence reference coefficient.

10. The data parallel processing system for enterprise credit rating according to claim 9, wherein The second execution module responds to the second combination extraction condition and performs matching abnormality analysis on the target extraction combination; Determine the target extraction indexes of the target extraction combination according to the difference value of the abnormal coefficient, and record the key indexes with the difference value of the abnormal coefficient greater than the preset difference value of the abnormal coefficient as the target extraction indexes; The reference period overlap degree of each target analysis material within any target extraction combination determined according to the reference period overlap degree is greater than the preset reference period overlap degree; The second combination extraction condition is determined by a target extraction combination according to the reference period overlap degree.

Images