WO2024108973A1 - Credit assessment method for construction enterprises - Google Patents

Abstract

A credit assessment method for construction enterprises, which credit assessment method relates to the field of informatization. The method is characterized by comprising the following steps: standard modeling, which involves: extracting data from indicators and analyzing same, so as to form a standard model library; data aggregation, which involves: detecting, in real time, the qualifications for aggregation regarding assessment objects and assessment users that have been uniformly authenticated, so as to form final assessment data, and performing data aggregation within the current assessment period; and special assessment, which involves: for assessment objects and assessment users that have been subjected to assessment limitation, calling data sources such as final assessment data and indicator models, taking the data sources as parameters, and importing the parameters into a computing engine to record process logs, acquire operation results and perform persistent storage. The present invention has the advantages of providing unified standard indicator analysis, by means of importing qualification credit assessment standards of different construction enterprise, and mainly performing analysis on types, tracing durations, data sources, collection modes, etc. A data model of indicator extraction data is formed by means of indicator data analysis, and taking information of a single type as an example, standard indicator information is formed.

Description

A credit evaluation method for construction enterprises Technical Field

The present invention relates to the field of informatization, and in particular to a method for providing unified standard indicator analysis by accessing different construction enterprise qualification credit evaluation standards, mainly analyzing from aspects such as type, tracing time, data source, and collection mode, to form a data model for indicator extraction data. Taking a type of information as an example, standard indicator information is formed, and general calculation rules are formed by performing logical extraction of calculations on the extracted indicator data; the calculation logics such as grouping, splitting, and restriction modularize and subdivide the calculation of the entire indicator to form a set of configurable and modular calculation rule systems, which is a credit evaluation method for construction enterprises.

Background technique

With the development of society, informatization has gradually entered all walks of life. At present, a relatively complete management method is needed for the credit evaluation system of each enterprise in the construction industry: by accessing the credit evaluation standards of different construction enterprises, a unified standard indicator analysis is provided, mainly from the aspects of type, traceability time, data source, collection mode, etc., to form a data model for indicator extraction data. Taking one type of information as an example, standard indicator information is formed, and the logic of calculation is extracted through the extracted indicator data to form a general calculation rule; the calculation logic such as grouping, splitting, and restriction modularizes and subdivides the calculation of the entire indicator to form a set of configurable and modular calculation rules system.

Summary of the invention

An embodiment of the present invention provides a credit evaluation method for construction companies. The method of the present invention provides a unified standard indicator analysis by accessing different construction company qualification credit evaluation standards, and mainly analyzes from aspects such as type, traceability period, data source, and collection mode. Through indicator data analysis, a data model for indicator extraction data is formed. Taking a type of information as an example, standard indicator information is formed. The extracted indicator data is used to perform logical extraction of calculations to form general calculation rules; the calculation of the entire indicator is modularized and subdivided through calculation logic such as grouping, splitting, and restriction to form a set of configurable, modular calculation rule systems. Based on the indicator model formed by the indicator rule analysis, a set of indicator models based on institutionalized data is created. Configuration; Based on modular programming technology, the configuration of the indicator model is divided into scoring rules, data collection forms, duplicate verification, and scoring algorithms, which realizes configurable programming. Subsequent standard changes only need to add corresponding modules; using deep learning technology, through modeling of existing indicator analysis and training its self-learning ability, the module of the indicator model is gradually realized to automatically optimize; based on the modular idea and the principle of single responsibility, the functions of dynamic generation of input forms, global verification of input data, general configuration of scoring rules, and fine-grained scoring algorithm are realized. Provide two data collection methods: data extraction and autonomous data aggregation. Among them, autonomous data aggregation verifies data validity through human-machine mutual review. Human review provides three levels of review (preliminary review, re-review, and confirmation). In the early stage or complex information, human intervention review is required. Machine review has a built-in machine learning module. Through continuous learning of human review data and data models and accumulation of sample data, it will gradually realize the method of machine review + manual review in the later stage. Based on Java pre-research and self-developed scoring calculation engine. Multi-threading technology enables real-time calculation of tens of millions of data, and the use of distributed computing greatly improves the calculation efficiency of scoring; the calculation process is fully asynchronous, avoiding thread blocking, including asynchronous tasks, asynchronous messages, asynchronous exception handling, etc.; the breakpoint-resume function is realized, and when the scoring is interrupted due to force majeure, a breakpoint-resume function similar to breakpoint-resume can be realized; the entire scoring process provides full-process data traceability, including real-time monitoring of scoring data and log recording of the entire evaluation process. In order to ensure the speed of data query, the system adopts a distributed big data storage system, and the use of unstructured data greatly improves the data query speed; some configuration data uses structured data storage to improve data readability.

The present invention provides a credit evaluation method for construction enterprises, the method comprising the following steps:

Standard modeling: extract the data type, duration, source, and collection mode characteristics of the indicators, complete the indicator data extraction and analysis, refine the calculation rules of specific indicators based on the indicator characteristics, and establish indicator nodes, observation nodes, scoring rules, aggregation forms, repeated verification, and scoring algorithm configuration in the indicator model to form a standard model library;

Data aggregation: Conduct real-time detection of aggregation qualifications for evaluation objects and evaluation users that have passed unified authentication. After passing the detection, they will automatically be qualified for aggregation and automatically extract and interactively aggregate the evaluation indicator model to complete the construction of the evaluation data source and form the final evaluation data; perform data aggregation for this evaluation cycle;

Special evaluation: For the evaluation objects and evaluation users defined by the evaluation, the final evaluation data, indicator model and other data sources are retrieved as parameters and brought into the calculation engine for process logging, calculation results and persistent storage.

A credit evaluation method for construction enterprises, wherein the standard modeling comprises the following steps:

Standard indicator extraction and analysis: sort out and group the evaluation criteria to form data types, analyze the validity period keywords in the data types, analyze the accuracy of the data sources to form rating data sources, and classify the rating data sources according to the enterprise declaration data, shared data, and data crawling types. Data aggregation forms early reserves;

Indicator data extraction and analysis: Analyze the types in the indicator data to obtain the specified indicator data items, and then analyze the specified indicator data according to real-time data and specified timeliness through tracing time analysis; determine the data source of the specified indicator data items through data source analysis, and complete the collection of the specified indicator data items through automatic docking, data sharing, and human-machine integration; form indicator data extraction rules;

Indicator calculation rule analysis: After obtaining the logic extraction of the extracted specified indicator data through the calculation formula to form a general calculation rule, the calculation of the entire indicator is modularized and subdivided through calculation logic such as grouping, splitting, and restriction to form a set of configurable scoring calculation models;

Indicator model configuration: Create indicator nodes according to the calculation formula and final calculation results in the scoring calculation model, configure one or more observation nodes for the created indicator nodes, define the information items in the existing observation point nodes to complete the scoring rules configuration, form the name in the scoring rules, and configure the specific scoring rules to the minimum score, and define the data drawing form to form the scoring data items, and verify the data content and repetition rate under the same observation point submitted by the same evaluation object; integrate and calculate the scoring rules and data content of the observation point nodes through the association algorithm group to obtain the evaluation results of a single observation node;

Standard model library management: The scoring criteria configuration, generated forms, evaluation execution rules, and duplicate verification rules constitute an indicator model and store it in the database; when a new indicator model is added, the data of the new indicator model is automatically expanded and integrated with the indicator model in the database.

A credit evaluation method for construction enterprises, wherein the standard index extraction and analysis comprises the following steps:

Analyze data to obtain data types: Sort out the evaluation content and scoring criteria in the evaluation criteria to obtain sorted information types, and then group the relevant information by type to obtain four data types: basic information, business performance, good information, and bad behavior information;

Retrospective duration analysis: Analyze the validity period of the time-sensitive data in each data type based on keywords;

Data source analysis: Analyze the accuracy of the sources of various types of data in each evaluation indicator, obtain the evaluation data source, and make preliminary preparations for data collection;

Data collection pattern analysis: The evaluation data source formed by data source analysis is classified and collected through enterprise declaration data, shared data, and data crawling, and the data after format unification is cached to make preliminary reserves for data collection.

A credit evaluation method for construction enterprises, wherein the index calculation rule analysis includes the following steps:

Same group calculation logic: According to the indicator data extraction rules, the specified indicator data is divided into the same group, and then the calculation logic of the data in the group is calculated; the formula is:

Where GROUP(i) represents the grouping calculation rules for evaluation data. Indicates the specific calculation rules between groups

Indicator data extraction rules: extract the specified indicator data in the group according to the actual needs after the calculation logic of the same group. One or more data extraction rules in row assignment, splitting, classification and timeout are used to calculate and obtain the index data extraction rules and calculation results that meet the requirements;

Boundary calculation logic: By comparing and analyzing the calculation results within the group with the specified boundary data, the boundary calculation logic and final calculation results that meet the requirements are obtained; the formula is:

Among them, MIN represents the minimum value of the boundary, MAX represents the maximum value of the boundary, and ∩ represents and

Constructing a scoring operation model: by physically matching the final operation result with the commonly used same group calculation logic, indicator data extraction rules, and boundary calculation logic, a scoring operation model that meets the requirements is formed, and the model contains one or more scoring operation formulas;

When the interval judgment includes the upper limit calculation, it is obtained through formula calculation, and the calculation formula is:

Among them, a, b, c represent the values after comparison, and x, y, z represent the values of comparison.

When the interval determination includes the lower limit operation, it is obtained through formula operation, and the operation formula is:

Among them, a, b, c represent the values after comparison, and x, y, z represent the values of comparison.

When performing the comparison operation, it is obtained through formula operation, and the operation formula is:
B(i)＝(x＝b?x＝n：x＝m)

Where x is the value to be compared, b is the value to be compared, n is the value that is compared successfully, and m is the value that is compared unsuccessfully.

? indicates a comparison operator, : indicates another case

When the excess and deficiency calculations are performed after the comparison, the calculation formula is obtained, and the calculation formula is:
B(i)＝(x＝b?x＝n::x＝m)±(x÷c*|(xb)|)

Where x is the value to be compared, b is the value to be compared, n is the value that is compared successfully, and m is the value that is compared unsuccessfully.

? represents a comparison operator, : represents another case, ± represents addition or subtraction, c represents the base of addition, and || represents taking the absolute value.

A credit evaluation method for construction enterprises, wherein the indicator model configuration comprises the following steps:

Create indicator nodes: Create indicator nodes according to the calculation formula and final calculation results in the scoring calculation model, where the created indicator nodes include: name, code, score, and indicator;

Observation point node: configure one or more observation nodes for the created indicator node, define the information items in the existing observation point node to complete the scoring criteria configuration, form the name in the scoring criteria, and configure the specific scoring criteria to the minimum score, and define the data drawing form to form the scoring data items. Verify the data content and duplication rate of the same observation point submitted by the same evaluation object; integrate the scoring criteria and data content of the observation point nodes through the association algorithm group to obtain the evaluation execution rules of a single observation node.

A credit evaluation method for construction enterprises, wherein the observation point node comprises the following steps:

Scoring rules configuration: configure one or more observation nodes for the created indicator node, define the information items in the existing observation point nodes to complete the scoring rules configuration, and form the name in the scoring rules and the specific scoring rules configuration up to the minimum score;

Data collection form: select whether to automatically generate the form. If it is determined to be yes, the form will be automatically generated. If it is determined not to be, the customized form will be connected and generated according to the specified parameters; according to the indicator data, the information items involved in the evaluation are selected for standardized configuration to generate the form and configure the data;

Duplicate check: set one or more keywords for the data content under the same observation point submitted by the same evaluation object, and obtain the duplication rate to support data collection and evaluation results by matching the keywords and their similar words and synonyms;

Scoring: The scoring rules and data content of the observation point nodes are integrated by associating the algorithm group to obtain the evaluation execution rules of a single observation node and the scoring algorithm group.

A credit evaluation method for construction enterprises, wherein the data aggregation comprises the following steps:

Unified authentication: Provide authentication modes based on the type of evaluation object, conduct qualification verification and authentication based on built-in data and the identity of the evaluation object, and conduct qualification verification and authentication through third-party authentication data sources and evaluation user-related identity information;

Data aggregation: The evaluation objects and users who have passed the unified authentication are sorted and aggregated into the evaluation qualification queue, and the indicator data are automatically extracted and interactively aggregated; for the evaluation objects and users who have not passed the unified authentication, historical information is saved;

Construct an evaluation data source: retrieve the information items in the data aggregation form from the indicator model configuration, process them with the information model stored in the big data warehouse to remove interference items, merge information models, and calculate information models to obtain the final evaluation data.

A credit evaluation method for construction enterprises, wherein the specific steps of data aggregation are: arranging and aggregating evaluation objects and evaluation users that have passed unified authentication into an evaluation qualification queue, and automatically extracting and interactively aggregating indicator data; for evaluation objects and evaluation users that have not passed unified authentication, saving historical information;

Aggregation qualifications: The evaluation objects and evaluation users who have passed the unified authentication are sorted and aggregated into the evaluation qualification queue, and the evaluation qualification queue is tested in real time; if the aggregation qualification of the evaluation object at the current time point is determined to be valid, the subsequent data aggregation work will continue; otherwise, the qualification of the evaluation object and evaluation user will be suspended, and the evaluation object and evaluation user will be enabled when they have the relevant qualifications again;

Automatic extraction: The evaluation objects and evaluation users who meet the aggregation qualifications are automatically included in the data extraction queue to extract the evaluation index model data, authenticate and connect the data interface through the public database, compare the evaluation objects, evaluation users and model data, and when the data comparison shows discrepancies Analyze and compare the difference information to obtain the difference information; when the data comparison results are consistent, proceed to the next step; search and match the objects of the evaluation object information with the data model in the public database one by one, synchronize the matching results to the local database, record the historical version of the local database and update it regularly; the evaluation index model data includes: enterprise data, personnel data, qualification data, and performance data;

Interactive aggregation: The good information models filled out by the evaluation objects and evaluation users who meet the aggregation qualifications and provide relevant supporting materials are submitted in accordance with the approval process. The auditors verify the relevant information through the human-computer interaction mode, store the good information models that meet the requirements in the big data warehouse, and return the good information models that do not meet the requirements for modification and re-audit. The auditors search for negative data models of the evaluation objects and evaluation users through the supervision records. When the relevant negative data models are found, the negative data of the relevant evaluation objects and evaluation users are extracted from the supervision records and filled into the negative data models and the negative data models are published. The evaluation objects and evaluation users check and confirm. When the relevant written data models cannot be found, the process ends.

A credit evaluation method for construction enterprises, wherein the special evaluation comprises the following steps:

Evaluation restriction: According to the evaluation rules, set evaluation prohibition clauses, and compare the characteristic data of the evaluation object and evaluation user one by one according to the prohibition clauses. If it is determined that the prohibition clauses are met, they will not be included in the list of evaluation objects and evaluation users. If it is determined that the prohibition clauses are not met, they will pass the restriction clause review process; according to the evaluation standards of the restriction clauses, compare the characteristic data of the evaluation object and evaluation user one by one. If it is determined that all the restriction clauses are met, the specified score will be given and they will be included in the list of evaluation objects and evaluation users. If it is determined that the restriction clauses of the newly included enterprise are met, the specified score will be given to the evaluation object and evaluation user and they will be included in the list of evaluation objects and evaluation users; if it is determined that the restriction clauses are partially met, wait for the evaluation operation;

Evaluation calculation: retrieve the user list of each evaluation object and the matching evaluation data and indicator model, and use them as the parameters of the calculation engine to perform multi-threaded parallel calculations, obtain the calculation results and process logs, and push abnormal log messages, and use unstructured distributed storage to store the entire process logs and calculation results;

Result push: define one or more push tasks and configure the information items, push time period, network environment and agreed encryption/decryption rules included in the task; push the configured push tasks;

The result delivery includes the following steps:

Information disclosure: retrieve good information models from the big data warehouse and incorporate them into the automatic disclosure process. If no objection is received during the specified disclosure period, the information model will be included in the evaluation data source;

Result sharing: The evaluation benchmark information and evaluation result information are shared to the public database through the data docking authentication interface for data sharing.

A credit evaluation method for construction enterprises, wherein the evaluation calculation comprises the following steps:

Obtain data source: retrieve final evaluation data, indicator model, and evaluation object evaluation user list to form parameter sets for calculation;

Computing engine: The list of users who evaluate the evaluation object is grouped in sequence and one or more data processors are deployed according to actual needs. The data processors are operated in multi-threaded mode. The data in the sequence are asynchronously calculated with the indicator model parameter loop, and the progress, message, status, and exceptions in the asynchronous process are monitored in real time to form a full-process log and calculation results; when abnormal tasks and abnormal results are displayed, the abnormal tasks and abnormal results are cyclically calculated after the calculation is completed. If the calculation result is determined to be an abnormal result, a full-process log is formed. If the calculation result is determined to be an abnormal result, the log is recorded and the log message is pushed to the manager at the same time;

formula:

Where score represents the final calculated score value, sources(i) is the data source, company is the enterprise data, scoring(i) is the calculation engine, arrows indicate the data flow, base(i) is the basic information, good(i) is good information, bad(i) is bad information, and management(i) is the management information;

Persistent storage: Use unstructured distributed storage mode to store the entire process log and calculation results.

It can be seen from this that:

A credit evaluation method for construction enterprises in an embodiment of the present invention provides a unified standard indicator analysis by accessing different construction enterprise qualification credit evaluation standards, mainly analyzing from the aspects of type, tracing time, data source, collection mode, etc. Through the indicator data analysis, a data model of indicator extraction data is formed, and standard indicator information is formed by taking a type of information as an example. The logical extraction of calculation aspects is performed through the extracted indicator data to form a general calculation rule; the calculation of the entire indicator is modularized and subdivided through calculation logic such as grouping, splitting, and restriction to form a set of configurable and modular calculation rule system. Based on the indicator model formed by the indicator rule analysis, a set of indicator model configuration based on institutionalized data is created; based on modular programming technology, the configuration of the indicator model is divided into scoring rules, data collection forms, duplicate verification, and scoring algorithms, realizing configurable programming, and subsequent standard changes only need to add corresponding modules; using deep learning technology, by modeling the existing indicator analysis, by training its self-learning ability, the module of the automatic optimization indicator model is gradually realized; based on the modular idea combined with the single responsibility principle, the functions of dynamic generation of input forms, global verification of input data, general configuration of scoring rules, and fine-grained scoring algorithm are realized. It provides two data collection methods: data extraction and autonomous data aggregation. Autonomous data aggregation verifies data validity through human-machine mutual review. Human review provides three levels of review (initial review, re-review, and confirmation). Human intervention is required for initial or complex information review. Machine review has a built-in machine learning module. Continuous learning of data models and accumulation of sample data, and the gradual implementation of machine review + manual review in the later stage. Based on Java pre-research and self-developed scoring calculation engine. Multi-threading technology realizes real-time calculation of tens of millions of data, and the use of distributed computing greatly improves the calculation efficiency of scoring; the calculation process is fully asynchronous, avoiding thread blocking, including asynchronous tasks, asynchronous messages, asynchronous exception handling, etc.; the breakpoint resume function is realized, and when the scoring is interrupted due to force majeure, a breakpoint resume function similar to breakpoint resume can be realized; the entire scoring process provides full process data traceability, including real-time monitoring of scoring data and logging of the entire evaluation process. In order to ensure the speed of data query, the system adopts a distributed big data storage system, and the use of unstructured data greatly improves the query speed of data; some configuration data uses structured data storage to improve data readability.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of the overall process of a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG2 is a flow chart of a standard modeling step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG3 is a flow chart of a standard indicator extraction and analysis step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG4 is a flow chart of an indicator calculation rule analysis step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG5 is a flow chart of an indicator model configuration step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG6 is a flow chart of an observation point node step in a credit evaluation method for a construction enterprise provided by an embodiment of the present invention;

FIG7 is a flow chart of a data aggregation step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG8 is a flow chart of a data aggregation step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG9 is a flow chart of a special evaluation step in a credit evaluation method for construction enterprises provided by an embodiment of the present invention;

FIG10 is a flow chart of an evaluation calculation step in a credit evaluation method for construction enterprises provided in an embodiment of the present invention.

Detailed ways

In order to make those skilled in the art better understand the present invention, the following will be described in conjunction with the accompanying drawings and specific The present invention will be described in detail with reference to the embodiments. The exemplary embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

Embodiment 1:

FIG1 is a credit evaluation method for construction enterprises. As shown in FIG1 , the method includes the following steps:

Standard modeling: extract the data type, duration, source, and collection mode characteristics of the indicators, complete the indicator data extraction and analysis, refine the calculation rules of specific indicators based on the indicator characteristics, and establish indicator nodes, observation nodes, scoring rules, aggregation forms, repeated verification, and scoring algorithm configuration in the indicator model to form a standard model library;

Data aggregation: Conduct real-time detection of aggregation qualifications for evaluation objects and evaluation users that have passed unified authentication. After passing the detection, they will automatically be qualified for aggregation and automatically extract and interactively aggregate the evaluation indicator model to complete the construction of the evaluation data source and form the final evaluation data; perform data aggregation for this evaluation cycle;

Special evaluation: For the evaluation objects and evaluation users defined by the evaluation, the final evaluation data, indicator model and other data sources are retrieved as parameters and brought into the calculation engine for process logging, calculation results and persistent storage.

As shown in FIG2 , a credit evaluation method for construction enterprises is provided, wherein the standard modeling includes the following steps:

Standard indicator extraction and analysis: sort out and group the evaluation standards to form data types, analyze the validity period keywords in the data types, analyze the accuracy of the data sources to form rating data sources, and classify the rating data sources according to the enterprise's declared data, shared data, and data crawling types to form early reserves for data aggregation;

Indicator data extraction and analysis: Analyze the types in the indicator data to obtain the specified indicator data items, and then analyze the specified indicator data according to real-time data and specified timeliness through tracing time analysis; determine the data source of the specified indicator data items through data source analysis, and complete the collection of the specified indicator data items through automatic docking, data sharing, and human-machine integration; form indicator data extraction rules;

Indicator calculation rule analysis: After obtaining the logic extraction of the extracted specified indicator data through the calculation formula to form a general calculation rule, the calculation of the entire indicator is modularized and subdivided through calculation logic such as grouping, splitting, and restriction to form a set of configurable scoring calculation models;

Indicator model configuration: Create indicator nodes according to the calculation formula and final calculation results in the scoring calculation model, configure one or more observation nodes for the created indicator nodes, define the information items in the existing observation point nodes to complete the scoring rules configuration, form the name in the scoring rules, and configure the specific scoring rules to the minimum score, and define the data drawing form to form the scoring data items, and verify the data content and repetition rate under the same observation point submitted by the same evaluation object; integrate and calculate the scoring rules and data content of the observation point nodes through the association algorithm group to obtain the evaluation results of a single observation node;

Standard model library management: The scoring rules configuration, generated forms, evaluation execution rules, and duplicate verification rules constitute the indicator model and store it in the database; when a new indicator model is added, the new indicator is automatically The model is integrated with the indicator model in the database for data expansion.

As shown in FIG3 , a credit evaluation method for construction enterprises is provided, wherein the standard index extraction and analysis comprises the following steps:

Analyze data to obtain data types: Sort out the evaluation content and scoring criteria in the evaluation criteria to obtain sorted information types, and then group the relevant information by type to obtain four data types: basic information, business performance, good information, and bad behavior information;

Retrospective duration analysis: Analyze the validity period of the time-sensitive data in each data type based on keywords;

Data source analysis: Analyze the accuracy of the sources of various types of data in each evaluation indicator, obtain the evaluation data source, and make preliminary preparations for data collection;

Data collection pattern analysis: The evaluation data source formed by data source analysis is classified and collected through enterprise declaration data, shared data, and data crawling, and the data after format unification is cached to make preliminary reserves for data collection.

As shown in FIG4 , a credit evaluation method for construction enterprises is provided, wherein the index calculation rule analysis includes the following steps:

Same group calculation logic: According to the indicator data extraction rules, the specified indicator data is divided into the same group, and then the calculation logic of the data in the group is calculated; the formula is:

Where GROUP(i) represents the grouping calculation rules for evaluation data. Indicates the specific calculation rules between groups

Indicator data extraction rules: assign values, split, classify, and time out the specified indicator data in the group after the same group calculation logic according to actual needs, and calculate one or more data extraction rules to obtain indicator data extraction rules and calculation results that meet the requirements;

Boundary calculation logic: By comparing and analyzing the calculation results within the group with the specified boundary data, the boundary calculation logic and final calculation results that meet the requirements are obtained; the formula is:

Among them, MIN represents the minimum value of the boundary, MAX represents the maximum value of the boundary, and ∩ represents and

Constructing a scoring operation model: by physically matching the final operation result with the commonly used same group calculation logic, indicator data extraction rules, and boundary calculation logic, a scoring operation model that meets the requirements is formed, and the model contains one or more scoring operation formulas;

When the interval judgment includes the upper limit calculation, it is obtained through formula calculation, and the calculation formula is:

Among them, a, b, c represent the values after comparison, and x, y, z represent the values of comparison.

When the interval determination includes the lower limit operation, it is obtained through formula operation, and the operation formula is:

Among them, a, b, c represent the values after comparison, and x, y, z represent the values of comparison.

When performing the comparison operation, it is obtained through formula operation, and the operation formula is:
B(i)＝(x＝b?x＝n：x＝m)

Where x is the value to be compared, b is the value to be compared, n is the value that is compared successfully, and m is the value that is compared unsuccessfully.

? indicates a comparison operator, : indicates another case

When the excess and deficiency calculations are performed after the comparison, the calculation formula is obtained, and the calculation formula is:
B(i)＝(x＝b?x＝n::x＝m)±(x÷c*|(xb)|)

Where x is the value to be compared, b is the value to be compared, n is the value that is compared successfully, and m is the value that is compared unsuccessfully.

? represents a comparison operator, : represents another case, ± represents addition or subtraction, c represents the base of addition, and || represents taking the absolute value.

As shown in FIG5 , a credit evaluation method for construction enterprises is provided, wherein the index model configuration includes the following steps:

Create indicator nodes: Create indicator nodes according to the calculation formula and final calculation results in the scoring calculation model, where the created indicator nodes include: name, code, score, and indicator;

Observation point node: configure one or more observation nodes for the created indicator node, define the information items in the existing observation point node to complete the scoring criteria configuration, form the name in the scoring criteria, and configure the specific scoring criteria to the minimum score, and define the data drawing form to form the scoring data items, and verify the data content and repetition rate under the same observation point submitted by the same evaluation object; integrate the scoring criteria and data content of the observation point node through the association algorithm group to obtain the evaluation execution rules of a single observation node.

As shown in FIG6 , a credit evaluation method for construction enterprises is provided, wherein the observation point node comprises the following steps:

Scoring rules configuration: configure one or more observation nodes for the created indicator node, define the information items in the existing observation point nodes to complete the scoring rules configuration, and form the name in the scoring rules and the specific scoring rules configuration up to the minimum score;

Data collection form: select whether to automatically generate the form. If it is determined to be yes, the form will be automatically generated. If it is determined not to be, the customized form will be connected and generated according to the specified parameters; according to the indicator data, the information items involved in the evaluation are selected for standardized configuration to generate the form and configure the data;

Duplicate check: set one or more keywords for the data content under the same observation point submitted by the same evaluation object, and obtain duplicates by matching the keywords and their similar words and synonyms. The rate supports data collection and evaluation results;

Scoring: The scoring rules and data content of the observation point nodes are integrated by associating the algorithm group to obtain the evaluation execution rules of a single observation node and the scoring algorithm group.

As shown in FIG. 7 , a credit evaluation method for construction enterprises is provided, wherein the data aggregation includes the following steps:

Unified authentication: Provide authentication modes based on the type of evaluation object, conduct qualification verification and authentication based on built-in data and the identity of the evaluation object, and conduct qualification verification and authentication through third-party authentication data sources and evaluation user-related identity information;

Data aggregation: The evaluation objects and users who have passed the unified authentication are sorted and aggregated into the evaluation qualification queue, and the indicator data are automatically extracted and interactively aggregated; for the evaluation objects and users who have not passed the unified authentication, historical information is saved;

Construct an evaluation data source: retrieve the information items in the data aggregation form from the indicator model configuration, process them with the information model stored in the big data warehouse to remove interference items, merge information models, and calculate information models to obtain the final evaluation data.

As shown in FIG8 , a credit evaluation method for construction enterprises is described, wherein the specific steps of data aggregation are: arranging and aggregating the evaluation objects and evaluation users that have passed the unified authentication into the evaluation qualification queue, and automatically extracting and interactively aggregating the index data; for the evaluation objects and evaluation users that have not passed the unified authentication, saving the historical information;

Aggregation qualifications: The evaluation objects and evaluation users who have passed the unified authentication are sorted and aggregated into the evaluation qualification queue, and the evaluation qualification queue is tested in real time; if the aggregation qualification of the evaluation object at the current time point is determined to be valid, the subsequent data aggregation work will continue; otherwise, the qualification of the evaluation object and evaluation user will be suspended, and the evaluation object and evaluation user will be enabled when they have the relevant qualifications again;

Automatic extraction: The evaluation objects and evaluation users that meet the aggregation qualifications are automatically included in the data extraction queue, and the evaluation index model data is extracted. The data interface is authenticated and connected through the public database, and the evaluation objects, evaluation users and model data are compared. When the data comparison shows difference information, the difference information is analyzed and compared to obtain the difference information; when the data comparison results are consistent, the next step is carried out; the objects of the evaluation object information are searched and matched with the data model in the public database one by one, and the matching results are synchronized to the local database. The local database records the historical version and updates it regularly; the evaluation index model data includes: enterprise data, personnel data, qualification data, and performance data;

Interactive aggregation: The good information models filled out by the evaluation objects and evaluation users who meet the aggregation qualifications and provide relevant supporting materials are submitted in accordance with the approval process. The auditors verify the relevant information through the human-computer interaction mode, store the good information models that meet the requirements in the big data warehouse, and return the good information models that do not meet the requirements for modification and re-audit process; the auditors search for negative data models of the evaluation objects and evaluation users through the supervision records. When the relevant negative data models are searched, the negative data of the relevant evaluation objects and evaluation users are extracted from the supervision records and filled into the negative data models and the negative data models are published. The evaluation objects and evaluation users check and confirm. If the relevant written data cannot be found, the negative data models are published. The process ends when the model is reached.

As shown in FIG9 , a credit evaluation method for construction enterprises is provided, wherein the special evaluation includes the following steps:

Evaluation restriction: According to the evaluation rules, set evaluation prohibition clauses, and compare the characteristic data of the evaluation object and evaluation user one by one according to the prohibition clauses. If it is determined that the prohibition clauses are met, they will not be included in the list of evaluation objects and evaluation users. If it is determined that the prohibition clauses are not met, they will pass the restriction clause review process; according to the evaluation standards of the restriction clauses, compare the characteristic data of the evaluation object and evaluation user one by one. If it is determined that all the restriction clauses are met, the specified score will be given and they will be included in the list of evaluation objects and evaluation users. If it is determined that the restriction clauses of the newly included enterprise are met, the specified score will be given to the evaluation object and evaluation user and they will be included in the list of evaluation objects and evaluation users; if it is determined that the restriction clauses are partially met, wait for the evaluation operation;

Evaluation calculation: retrieve the user list of each evaluation object and the matching evaluation data and indicator model, and use them as the parameters of the calculation engine to perform multi-threaded parallel calculations, obtain the calculation results and process logs, and push abnormal log messages, and use unstructured distributed storage to store the entire process logs and calculation results;

Result push: define one or more push tasks and configure the information items, push time period, network environment and agreed encryption/decryption rules included in the task; push the configured push tasks;

The result delivery includes the following steps:

Information disclosure: retrieve good information models from the big data warehouse and incorporate them into the automatic disclosure process. If no objection is received during the specified disclosure period, the information model will be included in the evaluation data source;

Result sharing: The evaluation benchmark information and evaluation result information are shared to the public database through the data docking authentication interface for data sharing.

As shown in FIG10 , a credit evaluation method for construction enterprises is provided, wherein the evaluation calculation includes the following steps:

Obtain data source: retrieve final evaluation data, indicator model, and evaluation object evaluation user list to form parameter sets for calculation;

Computing engine: The list of users of the evaluation object is grouped in sequence and more than one data processor is deployed according to actual needs. The data processor performs multi-threaded operation mode. The data in the sequence is asynchronously operated in a loop according to the evaluation data parameters and the indicator model parameters. The progress, messages, status and exceptions in the asynchronous process are monitored in real time to form a full-process log and operation results. When abnormal tasks and abnormal results are displayed, the abnormal tasks and abnormal results are cyclically operated after the operation is completed. If the operation result is determined to be abnormal, a full-process log is formed. If the operation result is determined to be an abnormal result, the log is recorded and the log message is pushed to the administrator at the same time.

formula:

Where score represents the final calculated score value, sources(i) is the data source, company is the enterprise data, scoring(i) is the calculation engine, arrows indicate the data flow, base(i) is the basic information, good(i) is good information, bad(i) is bad information, and management(i) is the management information;

Persistent storage: Use unstructured distributed storage mode to store the entire process log and calculation results.

It can be seen from this that: a credit evaluation method for construction enterprises in the embodiment of the present invention: by accessing different construction enterprise qualification credit evaluation standards, a unified standard indicator analysis is provided, mainly from the aspects of type, tracing time, data source, collection mode, etc. Through the indicator data analysis, a data model of indicator extraction data is formed, and standard indicator information is formed by taking a type of information as an example. Through the extracted indicator data, the logic extraction of calculation is carried out to form a general calculation rule; through the calculation logic such as grouping, splitting, and restriction, the calculation of the entire indicator is modularized and subdivided to form a set of configurable and modular calculation rule system. Based on the indicator model formed by the indicator rule analysis, a set of indicator model configuration based on institutionalized data is created; based on the modular programming technology, the configuration of the indicator model is divided into scoring rules, data collection forms, duplicate verification, and scoring algorithms, realizing configurable programming, and subsequent standard changes only need to add corresponding modules; using deep learning technology, by modeling the existing indicator analysis, by training its self-learning ability, the module of the automatic optimization indicator model is gradually realized; based on the modular idea combined with the single responsibility principle, the functions of dynamic generation of input forms, global verification of input data, general configuration of scoring rules, and fine-grained scoring algorithm are realized. Two data collection methods are provided: data extraction and autonomous data aggregation. Among them, autonomous data aggregation verifies the validity of data through human-machine mutual review. Human review provides three levels of review (initial review, re-review, and confirmation). Initial or complex information requires human intervention and review. The machine review has a built-in machine learning module. Through continuous learning of human review data and data models and accumulation of sample data, it will gradually implement machine review + manual review in the later stage. Based on Java pre-research and self-developed scoring calculation engine. Multi-threading technology realizes real-time calculation of tens of millions of data, and the use of distributed computing greatly improves the calculation efficiency of scoring; the calculation process is fully asynchronous, avoiding thread blocking, including asynchronous tasks, asynchronous messages, asynchronous exception handling, etc.; the breakpoint resumption function is realized, and when the scoring is interrupted due to force majeure, a breakpoint resumption function similar to breakpoint resumption can be realized; the entire scoring process provides full process data traceability, including real-time monitoring of scoring data and logging of the entire evaluation process. In order to ensure the speed of data query, the system adopts a distributed big data storage system, and the use of unstructured data greatly improves the data query speed; some configuration data adopts Structured data storage improves data readability.

Although the embodiments of the present invention have been described by way of example, those skilled in the art will appreciate that there are many modifications and variations of the present invention without departing from the spirit of the present invention. It is intended that the appended claims include these modifications and variations without departing from the spirit of the present invention.

Claims (10)

1. A credit evaluation method for construction enterprises, characterized in that the method comprises the following steps:

   Standard modeling: extract the data type, duration, source, and collection mode characteristics of the indicators, complete the indicator data extraction and analysis, refine the calculation rules of specific indicators based on the indicator characteristics, and establish indicator nodes, observation nodes, scoring rules, aggregation forms, repeated verification, and scoring algorithm configuration in the indicator model to form a standard model library;

   Data aggregation: Conduct real-time detection of aggregation qualifications for evaluation objects and evaluation users that have passed unified authentication. After passing the detection, they will automatically be qualified for aggregation and automatically extract and interactively aggregate the evaluation indicator model to complete the construction of the evaluation data source and form the final evaluation data; perform data aggregation for this evaluation cycle;

   Special evaluation: For the evaluation objects and evaluation users defined by the evaluation, the final evaluation data, indicator model and other data sources are retrieved as parameters and brought into the calculation engine for process logging, calculation results and persistent storage.
2. The credit evaluation method for construction enterprises according to claim 1 is characterized in that the standard modeling comprises the following steps:

   Standard indicator extraction and analysis: sort out and group the evaluation standards to form data types, analyze the validity period keywords in the data types, analyze the accuracy of the data sources to form rating data sources, and classify the rating data sources according to the enterprise's declared data, shared data, and data crawling types to form early reserves for data aggregation;

   Indicator data extraction and analysis: Analyze the types in the indicator data to obtain the specified indicator data items, and then analyze the specified indicator data according to real-time data and specified timeliness through tracing time analysis; determine the data source of the specified indicator data items through data source analysis, and complete the collection of the specified indicator data items through automatic docking, data sharing, and human-machine integration; form indicator data extraction rules;

   Indicator calculation rule analysis: After obtaining the logic extraction of the extracted specified indicator data through the calculation formula to form a general calculation rule, the calculation of the entire indicator is modularized and subdivided through calculation logic such as grouping, splitting, and restriction to form a set of configurable scoring calculation models;

   Indicator model configuration: Create indicator nodes according to the calculation formula and final calculation results in the scoring calculation model, configure one or more observation nodes for the created indicator nodes, define the information items in the existing observation point nodes to complete the scoring rules configuration, form the name in the scoring rules, and configure the specific scoring rules to the minimum score, and define the data drawing form to form the scoring data items, and verify the data content and repetition rate under the same observation point submitted by the same evaluation object; integrate and calculate the scoring rules and data content of the observation point nodes through the association algorithm group to obtain the evaluation results of a single observation node;

   Standard model library management: The scoring criteria configuration, generated forms, evaluation execution rules, and duplicate verification rules are combined into an indicator model and stored in the database; when a new indicator model is added, the new indicator is automatically The model is integrated with the indicator model in the database for data expansion.
3. According to a credit evaluation method for construction enterprises according to claim 2, it is characterized in that the standard indicator extraction and analysis comprises the following steps:

   Analyze data to obtain data types: Sort out the evaluation content and scoring criteria in the evaluation criteria to obtain sorted information types, and then group the relevant information by type to obtain four data types: basic information, business performance, good information, and bad behavior information;

   Retrospective duration analysis: Analyze the validity period of the time-sensitive data in each data type based on keywords;

   Data source analysis: Analyze the accuracy of the sources of various types of data in each evaluation indicator, obtain the evaluation data source, and make preliminary preparations for data collection;

   Data collection pattern analysis: The evaluation data source formed by data source analysis is classified and collected through enterprise declaration data, shared data, and data crawling, and the data after format unification is cached to make preliminary reserves for data collection.
4. According to the credit evaluation method for construction enterprises according to claim 2, it is characterized in that the indicator calculation rule analysis includes the following steps:

   Same group calculation logic: According to the indicator data extraction rules, the specified indicator data is divided into the same group, and then the calculation logic of the data in the group is calculated; the formula is:
   

   Where GROUP(i) represents the grouping calculation rules for evaluation data. Indicates the specific calculation rules between groups

   Indicator data extraction rules: assign values, split, classify, and time out the specified indicator data in the group after the same group calculation logic according to actual needs, and calculate one or more data extraction rules to obtain indicator data extraction rules and calculation results that meet the requirements;

   Boundary calculation logic: By comparing and analyzing the calculation results within the group with the specified boundary data, the boundary calculation logic and final calculation results that meet the requirements are obtained; the formula is:
   

   Among them, MIN represents the minimum value of the boundary, MAX represents the maximum value of the boundary, and ∩ represents and

   Constructing a scoring operation model: by physically matching the final operation result with the commonly used same group calculation logic, indicator data extraction rules, and boundary calculation logic, a scoring operation model that meets the requirements is formed, and the model contains one or more scoring operation formulas;

   When the interval judgment includes the upper limit calculation, it is obtained through formula calculation, and the calculation formula is:
   

   Among them, a, b, c represent the values after comparison, and x, y, z represent the values of comparison.

   When the interval determination includes the lower limit operation, it is obtained through formula operation, and the operation formula is:
   

   Among them, a, b, c represent the values after comparison, and x, y, z represent the values of comparison.

   When performing the comparison operation, it is obtained through formula operation, and the operation formula is:
   B(i)＝(x＝b?x＝n：x＝m)

   Where x is the value to be compared, b is the value to be compared, n is the value that is compared successfully, and m is the value that is compared unsuccessfully.

   ? indicates a comparison operator, : indicates another case

   When the excess and deficiency calculations are performed after the comparison, the calculation formula is obtained, and the calculation formula is:
   B(i)＝(x＝b?x＝n：x＝m)±(x÷c*|(xb)|)

   Where x is the value to be compared, b is the value to be compared, n is the value that is compared successfully, and m is the value that is compared unsuccessfully.

   ? represents a comparison operator, : represents another case, ± represents addition or subtraction, c represents the base of addition, and || represents taking the absolute value.
5. According to a credit evaluation method for construction enterprises according to claim 2, it is characterized in that: the indicator model configuration includes the following steps:

   Create indicator nodes: Create indicator nodes according to the calculation formula and final calculation results in the scoring calculation model, where the created indicator nodes include: name, code, score, and indicator;

   Observation point node: configure one or more observation nodes for the created indicator node, define the information items in the existing observation point node to complete the scoring criteria configuration, form the name in the scoring criteria, and configure the specific scoring criteria to the minimum score, and define the data drawing form to form the scoring data items, and verify the data content and repetition rate under the same observation point submitted by the same evaluation object; integrate the scoring criteria and data content of the observation point node through the association algorithm group to obtain the evaluation execution rules of a single observation node.
6. According to a credit evaluation method for construction enterprises according to claim 5, it is characterized in that the observation point node comprises the following steps:

   Scoring rules configuration: configure one or more observation nodes for the created indicator node, define the information items in the existing observation point nodes to complete the scoring rules configuration, and form the name in the scoring rules and the specific scoring rules configuration up to the minimum score;

   Data collection form: select whether to automatically generate the form. If it is determined to be yes, the form will be automatically generated. If it is determined not to be, the customized form will be connected and generated according to the specified parameters; according to the indicator data, the information items involved in the evaluation are selected for standardized configuration to generate the form and configure the data;

   Duplicate check: set one or more keywords for the data content under the same observation point submitted by the same evaluation object, and obtain duplicates by matching the keywords and their similar words and synonyms. The rate supports data collection and evaluation results;

   Scoring: The scoring rules and data content of the observation point nodes are integrated by associating the algorithm group to obtain the evaluation execution rules of a single observation node and the scoring algorithm group.
7. According to the credit evaluation method for construction enterprises according to claim 1, it is characterized in that the data aggregation comprises the following steps:

   Unified authentication: Provide authentication modes based on the type of evaluation object, conduct qualification verification and authentication based on built-in data and the identity of the evaluation object, and conduct qualification verification and authentication through third-party authentication data sources and relevant identity information of the evaluation user;

   Data aggregation: The evaluation objects and users who have passed the unified authentication are sorted and aggregated into the evaluation qualification queue, and the indicator data are automatically extracted and interactively aggregated; for the evaluation objects and users who have not passed the unified authentication, historical information is saved;

   Construct an evaluation data source: retrieve the information items in the data aggregation form from the indicator model configuration, process them with the information model stored in the big data warehouse to remove interference items, merge information models, and calculate information models to obtain the final evaluation data.
8. According to claim 7, a credit evaluation method for construction enterprises is characterized in that the specific steps of data aggregation are: arranging and aggregating the evaluation objects and evaluation users that have passed the unified authentication into the evaluation qualification queue, and automatically extracting and interactively aggregating the indicator data; for the evaluation objects and evaluation users that have not passed the unified authentication, saving the historical information;

   Aggregation qualifications: The evaluation objects and evaluation users who have passed the unified authentication are sorted and aggregated into the evaluation qualification queue, and the evaluation qualification queue is tested in real time; if the aggregation qualification of the evaluation object at the current time point is determined to be valid, the subsequent data aggregation work will continue; otherwise, the qualification of the evaluation object and evaluation user will be suspended, and the evaluation object and evaluation user will be enabled when they have the relevant qualifications again;

   Automatic extraction: The evaluation objects and evaluation users that meet the aggregation qualifications are automatically included in the data extraction queue, and the evaluation index model data is extracted. The data interface is authenticated and connected through the public database, and the evaluation objects, evaluation users and model data are compared. When the data comparison shows difference information, the difference information is analyzed and compared to obtain the difference information; when the data comparison results are consistent, the next step is carried out; the objects of the evaluation object information are searched and matched with the data model in the public database one by one, and the matching results are synchronized to the local database. The local database records the historical version and updates it regularly; the evaluation index model data includes: enterprise data, personnel data, qualification data, and performance data;

   Interactive aggregation: The good information models filled out by the evaluation objects and evaluation users who meet the aggregation qualifications and provide relevant supporting materials are submitted in accordance with the approval process. The auditors verify the relevant information through the human-computer interaction mode, store the good information models that meet the requirements in the big data warehouse, and return the good information models that do not meet the requirements for modification and re-audit process; the auditors search for negative data models of the evaluation objects and evaluation users through the supervision records. When the relevant negative data models are searched, the negative data of the relevant evaluation objects and evaluation users are extracted from the supervision records and filled into the negative data models and the negative data models are published. The evaluation objects and evaluation users check and confirm. If the relevant written data cannot be found, the negative data models are published. The process ends when the model is reached.
9. According to a credit evaluation method for construction enterprises according to claim 1, it is characterized in that the special evaluation comprises the following steps:

   Evaluation restriction: set evaluation prohibition clauses according to the evaluation rules, compare the characteristic data of the evaluation object and evaluation user one by one according to the prohibition clauses, and when it is determined that the prohibition clauses are met, they will not be included in the list of evaluation objects and evaluation users; when it is determined that the prohibition clauses are not met, they will pass the restriction clause review process; compare the characteristic data of the evaluation object and evaluation user one by one according to the evaluation criteria of the restriction clauses, and when it is determined that all the restriction clauses are met, they will be assigned a specified score and included in the list of evaluation objects and evaluation users; when it is determined that the restriction clauses of the newly included enterprise are met, the evaluation object and evaluation user will be assigned a specified score and included in the list of evaluation objects and evaluation users; when it is determined that the restriction clauses are partially met, wait for the evaluation operation;

   Evaluation calculation: retrieve the user list of each evaluation object and the matching evaluation data and indicator model, and use them as the parameters of the calculation engine to perform multi-threaded parallel calculations, obtain the calculation results and process logs, and push abnormal log messages, and use unstructured distributed storage to store the entire process logs and calculation results;

   Result push: define one or more push tasks and configure the information items, push time period, network environment and agreed encryption/decryption rules included in the task; push the configured push tasks;

   The result delivery includes the following steps:

   Information disclosure: retrieve good information models from the big data warehouse and incorporate them into the automatic disclosure process. If no objection is received during the specified disclosure period, the information model will be included in the evaluation data source;

   Result sharing: The evaluation benchmark information and evaluation result information are shared to the public database through the data docking authentication interface for data sharing.
10. The credit evaluation method for construction enterprises according to claim 9 is characterized in that the evaluation calculation comprises the following steps:

    Obtain data source: retrieve final evaluation data, indicator model, and evaluation object evaluation user list to form parameter sets for calculation;

    Computing engine: The list of users of the evaluation object is grouped in sequence and more than one data processor is deployed according to actual needs. The data processor performs multi-threaded operation mode. The data in the sequence is asynchronously operated in a loop according to the evaluation data parameters and the indicator model parameters. The progress, messages, status and exceptions in the asynchronous process are monitored in real time to form a full-process log and operation results. When abnormal tasks and abnormal results are displayed, the abnormal tasks and abnormal results are cyclically operated after the operation is completed. If the operation result is determined to be abnormal, a full-process log is formed. If the operation result is determined to be an abnormal result, the log is recorded and the log message is pushed to the administrator at the same time.

    formula:
    
    
    

    Where score represents the final calculated score value, sources(i) is the data source, company is the enterprise data, scoring(i) is the calculation engine, arrows indicate the data flow, base(i) is the basic information, good(i) is good information, bad(i) is bad information, and management(i) is the management information;

    Persistent storage: Use unstructured distributed storage mode to store the entire process log and calculation results.