Taxpayer risk identification method and system based on inspection rule and relation graph

Abstract

The invention discloses a taxpayer risk identification method and system based on an inspection rule and a relation graph, and the method comprises the steps: obtaining historical invoice data, stored in an offline data warehouse, of the invoicing and transaction behaviors of a taxpayer, and carrying out the cleaning and structural processing of the historical invoice data, so as to obtain historical invoice processing data; configuring risk indexes and threshold values corresponding to the risk indexes according to the inspection rules of the different areas so as to construct a risk rule engine; constructing a transaction relation graph and a risk model of the taxpayer based on the historical invoice processing data and risk indexes in a risk rule engine; obtaining detailed invoice data of invoicing and transaction behaviors of the taxpayer in real time; and performing risk identification based on the detailed invoice data, the transaction relation graph and the risk model, and outputting a risk identification result in combination with a threshold value corresponding to each risk index in the risk rule engine.

Description

Technical Field

[0001] The present invention relates to the technical field of taxpayer risk identification, and more specifically, to a taxpayer risk identification method and system based on audit rules and relationship graphs. Background Art

[0002] With the continuous development of digitalization, data applications and flows have become more extensive and frequent within the government industry, and data flows have become more diversified, which will bring many security risks when data is shared. Especially in the field of taxation, how to identify and analyze the risks of invoicing and transaction behaviors of enterprises or taxpayers in an era of rapid information flow, and to make qualitative and quantitative estimates of the possibility and severity of adverse effects, and to effectively identify and control potential risks from different angles according to different strategies, has become the key to solving this problem. It is particularly important to identify and analyze data assets of corresponding public data, personal data, important data, etc., and to identify threats, identify vulnerabilities, and conduct risk analysis after passing the data security risk assessment.

[0003] How to analyze the invoicing and transaction behaviors of enterprises or taxpayers more reasonably, comprehensively and efficiently, so as to identify and control tax risks is particularly important. However, in practice, the methods of analyzing the transaction behaviors of enterprises or taxpayers and identifying tax risks still have the following shortcomings: 1. The analysis strategy of invoicing and transaction behaviors is not comprehensive enough: the invoicing and transaction behaviors of enterprises or taxpayers have diverse characteristics. When the information collected is more comprehensive, the situation is more comprehensive, and the analysis rules are more complex, the rules of behavior analysis will be more perfect; 2. The real-time performance of risk identification and monitoring is poor: some risk strategies of enterprises or taxpayers have high requirements for real-time performance. How to carry out real-time and effective risk identification and control under the premise of complex risk rules is particularly important. At the same time, it is more difficult to ensure the accuracy of risk assessment in real time; 3. The risk analysis and identification engine is relatively simple: in the field of tax audit, the technology stack of the tax risk analysis engine for enterprises or taxpayers is too simple, and it is not possible to effectively handle the comprehensive scenarios combining simple rules, real-time rules and complex rules. At this time, it is particularly important to introduce more analysis engines and algorithms to handle such complex tax scenarios.

[0004] The above situations are crucial in today's tax system. Only more comprehensive analysis, more timely judgment and more accurate analysis and identification can truly meet the needs of the taxpayer risk identification system. Therefore, a taxpayer risk identification method based on audit rules and relationship maps is needed. Summary of the invention

[0005] The present invention proposes a taxpayer risk identification method and system based on audit rules and relationship maps to solve the problem of how to identify taxpayer risks efficiently and accurately.

[0006] In order to solve the above problem, according to one aspect of the present invention, a taxpayer risk identification method based on audit rules and relationship graph is provided, the method comprising:

[0007] Obtain historical invoice data of taxpayers' invoicing and transaction behaviors stored in the offline data warehouse, and clean and structure the historical invoice data to obtain historical invoice processing data;

[0008] Configure risk indicators and thresholds corresponding to each risk indicator according to the audit rules of different regions to build a risk rule engine;

[0009] Building a taxpayer's transaction relationship map and risk model based on the historical invoice processing data and risk indicators in the risk rule engine;

[0010] Obtain detailed invoice data of taxpayers’ invoicing and transaction behaviors in real time;

[0011] Risk identification is performed based on the detailed invoice data, transaction relationship map and risk model, and the risk identification result is output in combination with the threshold value corresponding to each risk indicator in the risk rule engine.

[0012] Preferably, the transaction relationship map includes: a loop transaction risk map and a community transaction risk map.

[0013] Preferably, the loop transaction risk map is constructed using a depth-first search algorithm (DFS), while edge vertex optimization, bidirectional traversal optimization and low-value data screening are used to reduce the complexity of the algorithm, and then combined with multi-threaded parallel computing to output the transaction paths of enterprises or taxpayers with loop transaction risks.

[0014] Preferably, when the method identifies risks based on the community transaction risk map, the input and output tax amounts and the degree of difference are calculated based on the taxpayer's invoices and transaction information, and based on this, the enterprises are divided into communities according to the degree of difference. If the transaction amount exceeds the preset amount and the taxpayer is in an abnormal degree of difference community, it is determined that the taxpayer has an abnormal community transaction risk.

[0015] Preferably, the risk model includes: abnormal risk model for purchased and sold goods, risk model for sudden increase in invoice quantity, risk model for sudden increase in invoice amount, abnormal risk model for invoice cancellation, abnormal risk model for invoicing time, abnormal risk model for associated enterprises, risk model for large-scale invoicing of newly registered enterprises, risk model for large-scale invoicing at the end of the month, risk model for false deductions, abnormal risk model for illegal warehousing, abnormal risk model for suspected sale of invoices and risk model for invoicing in a different place.

[0016] Preferably, when the method identifies risks based on the invoice quantity sudden increase risk model or the invoice amount sudden increase risk model, it collects statistics and monitors the taxpayer's invoice data or invoice amount within a preset time period, calculates the year-on-year and month-on-month growth rates of the invoice quantity or invoice amount, and if the year-on-year and month-on-month growth rates of the taxpayer exceed the preset threshold, it is determined that the taxpayer is at risk of a sudden increase in invoice quantity or a sudden increase in invoice amount.

[0017] Preferably, when performing risk identification based on the invoice red cancellation abnormal risk model, the method collects statistics on the amount and quantity of the taxpayer's red invoices and all invoices, and calculates the proportion of the amount and quantity of red invoices to all invoices. If the proportion of the taxpayer's red invoice amount and quantity exceeds a preset threshold, it is determined that the taxpayer has an invoice red cancellation abnormal risk.

[0018] Preferably, when the method identifies risks based on the abnormal risk model of associated enterprises, it counts and monitors the taxpayer's upstream and downstream trading enterprises. If there are abnormalities in the upstream and downstream enterprises, the number of transactions between the enterprise and the abnormal enterprise and the proportion of gold in the total transactions of the enterprise are calculated. If the number or amount ratio of input transactions, or the number or amount ratio of output transactions exceeds the threshold, the taxpayer has abnormal risk of associated enterprises.

[0019] Preferably, the method further comprises:

[0020] The taxpayer's transaction behavior is scored and rated based on the risk identification results.

[0021] According to another aspect of the present invention, a taxpayer risk identification system based on audit rules and relationship graphs is provided, the system comprising:

[0022] An invoice data acquisition unit is used to acquire historical invoice data of taxpayers' invoicing and transaction behaviors stored in the offline data warehouse, and to clean and structure the historical invoice data to acquire historical invoice processing data;

[0023] A risk rule engine construction unit is used to configure risk indicators and thresholds corresponding to each risk indicator according to the audit rules of different regions to build a risk rule engine;

[0024] A graph and risk model building unit, used to build a transaction relationship graph and risk model of the taxpayer based on the historical invoice processing data and the risk indicators in the risk rule engine;

[0025] The invoice data real-time acquisition unit is used to acquire the detailed invoice data of the taxpayer's invoicing and transaction behaviors in real time;

[0026] The risk identification unit is used to perform risk identification based on the detailed invoice data, the transaction relationship map and the risk model, and output the risk identification result in combination with the threshold value corresponding to each risk indicator in the risk rule engine.

[0027] Preferably, the transaction relationship map includes: a loop transaction risk map and a community transaction risk map.

[0028] Preferably, the loop transaction risk map is constructed using a depth-first search algorithm (DFS), while edge vertex optimization, bidirectional traversal optimization and low-value data screening are used to reduce the complexity of the algorithm, and then combined with multi-threaded parallel computing to output the transaction paths of enterprises or taxpayers with loop transaction risks.

[0029] Preferably, the risk identification unit, when performing risk identification based on the community transaction risk map, calculates the input and output tax amounts and the degree of difference based on the taxpayer's invoice and transaction information, and divides the enterprises into communities based on the degree of difference. If the transaction amount exceeds the preset amount and the taxpayer is in a community with abnormal degree of difference, it is determined that the taxpayer has an abnormal community transaction risk.

[0030] Preferably, the risk model includes: abnormal risk model for purchased and sold goods, risk model for sudden increase in invoice quantity, risk model for sudden increase in invoice amount, abnormal risk model for invoice cancellation, abnormal risk model for invoicing time, abnormal risk model for associated enterprises, risk model for large-scale invoicing of newly registered enterprises, risk model for large-scale invoicing at the end of the month, risk model for false deductions, abnormal risk model for illegal warehousing, abnormal risk model for suspected sale of invoices and risk model for invoicing in a different place.

[0031] Preferably, when performing risk identification based on the invoice quantity sudden increase risk model or the invoice amount sudden increase risk model, the risk identification unit collects statistics and monitors the taxpayer's invoicing data or invoice amount within a preset time period, and calculates the year-on-year and month-on-month growth rates of the invoice quantity or invoice amount. If the year-on-year and month-on-month growth rates of the taxpayer exceed the preset threshold, it is determined that the taxpayer is at risk of a sudden increase in the number of invoices or a sudden increase in the invoice amount.

[0032] Preferably, the risk identification unit, when performing risk identification based on the invoice cancellation abnormality risk model, counts the amount and quantity of the taxpayer's red invoices and all invoices, calculates the ratio of the amount and quantity of red invoices to all invoices; if the ratio of the amount and quantity of the taxpayer's red invoices exceeds a preset threshold, it is determined that the taxpayer has an invoice cancellation abnormality risk.

[0033] Preferably, the risk identification unit, when performing risk identification based on the abnormal risk model of associated enterprises, counts and monitors the taxpayer's upstream and downstream trading enterprises. If there are abnormalities in the upstream and downstream enterprises, the number of transactions between the enterprise and the abnormal enterprise and the proportion of gold in the total transactions of the enterprise are calculated. If the proportion of the number or amount of input transactions, or the proportion of the number or amount of output transactions exceeds the threshold, the taxpayer has abnormal risk of associated enterprises.

[0034] Preferably, the system further comprises:

[0035] A scoring unit is used to score and rate the taxpayer's transaction behavior based on the risk identification results.

[0036] The present invention provides a method and system for taxpayer risk identification based on audit rules and relationship maps, including: obtaining historical invoice data of taxpayers performing invoicing and transaction behaviors stored in an offline data warehouse, and cleaning and structuring the historical invoice data to obtain historical invoice processing data; configuring risk indicators and thresholds corresponding to each risk indicator according to audit rules of different regions to build a risk rule engine; building a taxpayer's transaction relationship map and risk model based on the historical invoice processing data and the risk indicators in the risk rule engine; obtaining detailed invoice data of taxpayers performing invoicing and transaction behaviors in real time; performing risk identification based on the detailed invoice data, the transaction relationship map and the risk model, and outputting risk identification results in combination with the thresholds corresponding to each risk indicator in the risk rule engine. The present invention combines big data, relationship maps and taxpayer risk identification, adopts algorithms plus rules through offline analysis, real-time calculation and graph algorithms, and uses a big data algorithm calculation benchmark based on the big data algorithm. Threshold control is then performed using a rule management engine to jointly realize multi-dimensional risk identification of taxpayers, greatly improving the accuracy and real-time nature of tax-related risk identification of enterprises or taxpayers. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0038] Figure 1 A flowchart of a taxpayer risk identification method 100 based on audit rules and relationship graphs according to an embodiment of the present invention;

[0039] Figure 2 A schematic diagram of a risk identification process according to an embodiment of the present invention;

[0040] Figure 3 It is a structural diagram of a taxpayer risk identification system 300 based on audit rules and relationship graphs according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0042] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0043] Figure 1 FIG. 1 is a flow chart of a taxpayer risk identification method 100 based on audit rules and relationship graphs according to an embodiment of the present invention. Figure 1 As shown, the taxpayer risk identification method based on audit rules and relationship maps provided by the embodiment of the present invention combines big data, relationship maps and taxpayer risk identification, and adopts algorithms plus rules through offline analysis, real-time calculation and graph algorithms, based on big data algorithm calculation benchmarks, and then uses the rule management engine for threshold control, to jointly achieve multi-dimensional risk identification of taxpayers, greatly improving the accuracy and real-time nature of tax-related risk identification of enterprises or taxpayers. The taxpayer risk identification method based on audit rules and relationship maps provided by the embodiment of the present invention 100 starts from step 101. In step 101, the historical invoice data of the taxpayer's invoicing and transaction behavior stored in the offline data warehouse is obtained, and the historical invoice data is cleaned and structured to obtain historical invoice processing data.

[0044] In step 102, risk indicators and thresholds corresponding to each risk indicator are configured according to the audit rules of different regions to build a risk rule engine.

[0045] In step 103, based on the historical invoice processing data and the risk indicators in the risk rule engine, a transaction relationship map and a risk model of the taxpayer are constructed.

[0046] Preferably, the transaction relationship map includes: a loop transaction risk map and a community transaction risk map.

[0047] Preferably, the loop transaction risk map is constructed using a depth-first search algorithm (DFS), while edge vertex optimization, bidirectional traversal optimization and low-value data screening are used to reduce the complexity of the algorithm, and then combined with multi-threaded parallel computing to output the transaction paths of enterprises or taxpayers with loop transaction risks.

[0048] Preferably, the risk model includes: abnormal risk model for purchased and sold goods, risk model for sudden increase in invoice quantity, risk model for sudden increase in invoice amount, abnormal risk model for invoice cancellation, abnormal risk model for invoicing time, abnormal risk model for associated enterprises, risk model for large-scale invoicing of newly registered enterprises, risk model for large-scale invoicing at the end of the month, risk model for false deductions, abnormal risk model for illegal warehousing, abnormal risk model for suspected sale of invoices and risk model for invoicing in a different place.

[0049] In step 104, detailed invoice data of the taxpayer's invoicing and transaction activities is obtained in real time.

[0050] In step 105, risk identification is performed based on the detailed invoice data, the transaction relationship map and the risk model, and the risk identification result is output in combination with the threshold value corresponding to each risk indicator in the risk rule engine.

[0051] In order to comprehensively, timely and comprehensively prevent tax risks of enterprises and taxpayers, the present invention eliminates the taxpayer risk identification method based on audit rules and relationship maps, and combines Figure 2 As shown, based on the real-time data of the enterprise's tax-related information combined with the offline analysis results, the tax-related risks of the enterprise or taxpayer can be effectively identified under the rule engine. The main steps are as follows:

[0052] 1. Enterprises or taxpayers issue invoices and conduct transactions, obtain detailed invoice data of the current behavior as input for real-time analysis, and periodically write such real-time invoice records into the offline data warehouse.

[0053] 2. The invoicing behavior of enterprises or users and the structuring of invoice information constitute a periodically synchronized offline data warehouse, providing a data foundation for the construction of risk models and relationship maps.

[0054] 3. The risk indicator management system composed of audit rules and the configurable threshold management system constitute the risk rule engine. This part is mainly managed by professionals, who can build risk indicators according to the actual conditions and policies of different regions, and adjust and configure the thresholds of various indicators.

[0055] 4. Based on the data of offline data warehouse, combined with the relationship graph related indicators in the risk indicator system, after data processing such as cleaning and structuring, the invoicing and transaction relationship graph of enterprises or taxpayers is formed, providing a data basis for subsequent graph algorithm analysis. This includes loop transaction risks and community transaction risks.

[0056] 5. Also based on the data from the offline data warehouse, combined with the risk indicators related to the audit rules in the risk indicator system, after data cleaning and analysis, a rule-driven risk model is constructed, which includes the abnormal risks of purchased and sold goods, the risk of sudden increase in the number or amount of invoices, the risk of abnormal invoicing time, and the risk of large-scale invoicing at the end of the month.

[0057] 6. The above real-time invoice data, transaction relationship map and risk model are analyzed through big data real-time analysis, graph algorithm analysis and big data offline analysis respectively, combined with the threshold management system of the risk rule engine to output the risk results.

[0058] Among them, the loop transaction risk in the graph algorithm uses the depth-first search algorithm (DFS), and simultaneously adopts edge vertex optimization, bidirectional traversal optimization and low-value data screening to reduce the complexity of the algorithm as much as possible. Finally, combined with multi-threaded parallel computing, the transaction path of enterprises or taxpayers with loop transaction risks is output.

[0059] Preferably, when the method identifies risks based on the community transaction risk map, the input and output tax amounts and the degree of difference are calculated based on the taxpayer's invoices and transaction information, and based on this, the enterprises are divided into communities according to the degree of difference. If the transaction amount exceeds the preset amount and the taxpayer is in an abnormal degree of difference community, it is determined that the taxpayer has an abnormal community transaction risk.

[0060] Preferably, when the method identifies risks based on the invoice quantity sudden increase risk model or the invoice amount sudden increase risk model, it collects statistics and monitors the taxpayer's invoice data or invoice amount within a preset time period, calculates the year-on-year and month-on-month growth rates of the invoice quantity or invoice amount, and if the year-on-year and month-on-month growth rates of the taxpayer exceed the preset threshold, it is determined that the taxpayer is at risk of a sudden increase in invoice quantity or a sudden increase in invoice amount.

[0061] Preferably, when performing risk identification based on the invoice red cancellation abnormal risk model, the method collects statistics on the amount and quantity of the taxpayer's red invoices and all invoices, and calculates the proportion of the amount and quantity of red invoices to all invoices. If the proportion of the taxpayer's red invoice amount and quantity exceeds a preset threshold, it is determined that the taxpayer has an invoice red cancellation abnormal risk.

[0062] Preferably, when the method identifies risks based on the abnormal risk model of associated enterprises, it counts and monitors the taxpayer's upstream and downstream trading enterprises. If there are abnormalities in the upstream and downstream enterprises, the number of transactions between the enterprise and the abnormal enterprise and the proportion of gold in the total transactions of the enterprise are calculated. If the number or amount ratio of input transactions, or the number or amount ratio of output transactions exceeds the threshold, the taxpayer has abnormal risk of associated enterprises.

[0063] The behavior analysis of enterprises or taxpayers in the present invention is divided into multiple dimensions according to the audit rules and is implemented based on the following multiple models:

[0064] (1) Loop Transaction Abnormal Risk Model

[0065] A graph algorithm model structures data and builds a relationship graph based on an offline data warehouse, and uses graph algorithms to mine enterprises or taxpayers with loop transactions. The transaction amount threshold and the threshold for the number of enterprises in the loop can be modified and maintained in the threshold management system.

[0066] (2) Community transaction abnormal risk model

[0067] A graph algorithm model, similarly, structures the data and builds a relationship map based on the offline data warehouse, calculates the input and output tax and the difference degree based on the invoice and transaction information of the enterprise or taxpayer, and divides the enterprise into communities based on the difference degree. Enterprises with transaction amounts exceeding a certain amount and in the abnormal difference degree community are judged as abnormal. The transaction amount threshold and the difference degree interval threshold can be modified and maintained in the threshold management system.

[0068] (3) Abnormal risk model for purchase and sales items

[0069] An offline analysis model that calculates the difference between the main input commodities and the main marketing commodities of the taxpayers based on the commodity transaction data of the enterprises or taxpayers in the offline data warehouse, and finds out the taxpayers with abnormal input and sales commodities. The difference threshold can be modified and maintained in the threshold management system.

[0070] (4) Risk model for sudden increase in invoice quantity

[0071] An offline analysis model that counts and monitors the number of invoices issued each month based on the invoice data of enterprises or taxpayers in the offline data warehouse, and calculates the year-on-year and month-on-month changes in the number of invoices issued. Taxpayers with significant year-on-year and month-on-month increases are classified as abnormal taxpayers. The year-on-year and month-on-month thresholds TV1 and TV2 in the model can be modified and maintained in the threshold management system.

[0072] (5) Risk model for sudden increase in invoice amount

[0073] An offline analysis model that counts and monitors the monthly invoice amount of enterprises or taxpayers based on the invoice data in the offline data warehouse, and calculates the year-on-year and month-on-month changes in the invoice amount. Taxpayers with significant year-on-year and month-on-month increases are listed as abnormal taxpayers. The year-on-year and month-on-month thresholds TV1 and TV2 in the model can be modified and maintained in the threshold management system.

[0074] (6) Invoice cancellation abnormal risk model

[0075] An offline analysis model that counts the amount and quantity of red-letter invoices and all invoices based on the invoice and red-letter invoice cancellation data of enterprises or taxpayers in the offline data warehouse, calculates the ratio of the amount and quantity of red-letter invoices to all invoices, and lists taxpayers with excessively large amounts and quantities of red-letter invoices as abnormal taxpayers. The threshold TV1 for the proportion of invoice amount and the threshold TV2 for the proportion of invoice quantity in the model can be modified and maintained in the threshold management system.

[0076] (7) Risk model for abnormal billing time

[0077] An offline and real-time analysis model, based on the real-time and offline data warehouse historical invoicing data, counts and analyzes the number of invoices and invoicing time of enterprises or taxpayers, and calculates the proportion of invoicing quantity in each period and each day within a day and a month. Taxpayers whose proportion of invoicing quantity in abnormal time periods is higher than the risk threshold are classified as abnormal. The abnormal time period T1, abnormal time period proportion abnormal threshold TV1, abnormal date T2, and abnormal date proportion abnormal threshold TV2 in the model can be modified and maintained in the threshold management system.

[0078] (8) Abnormal risk model of related enterprises

[0079] An offline analysis model, based on the invoice data of enterprises and taxpayers in the offline data warehouse, counts and monitors the upstream and downstream trading enterprises of an enterprise. If there are abnormalities in the upstream and downstream enterprises, the number of transactions between the enterprise and the abnormal enterprise and the proportion of the gold in the total transactions of the enterprise are calculated. If the proportion of the number or amount of input transactions, or the proportion of the number or amount of output transactions exceeds the threshold, the enterprise is also listed as an abnormal enterprise. The input transaction amount proportion threshold TV1, input transaction quantity proportion threshold TV2, output transaction amount proportion threshold TV3 and output transaction quantity proportion threshold TV4 in the model can be modified and maintained in the threshold management system.

[0080] (9) Risk model for large-scale invoicing by newly registered enterprises

[0081] An offline analysis model that counts and monitors the registration time and invoicing of enterprises or taxpayers based on the registration information and invoice data of enterprises and taxpayers in the offline data warehouse. If the registration time is within a certain period of time and the number and amount of invoices issued by the enterprise in recent months are too large, there is a risk. In the model, the new enterprise registration time threshold is T. Within the last T1 month, the number of invoices exceeds TV1 and the invoice amount exceeds TV2, or within the last T2 months, the number of invoices exceeds TV3 and the invoice amount exceeds TV4. The above thresholds can be modified and maintained in the threshold management system.

[0082] (10) Abnormal risk model for large number of invoices at the end of the month

[0083] An offline analysis model, based on the invoice data of enterprises and taxpayers in the offline data warehouse, if the input tax of the invoice issued exceeds a certain amount in the last few days or one day of the month, and the number and amount of invoices account for too large a proportion of the total number of invoices and the total amount of the month, there is a risk. The thresholds of time, input tax, number of invoices and amount ratio in the model can all be modified and maintained in the threshold management system.

[0084] (11) Abnormal risk model for large-scale taxpayer false deductions

[0085] An offline analysis model that counts and analyzes the business turnover and input and output tax of an enterprise or taxpayer based on the invoice data of the enterprise and taxpayer in the offline data warehouse. If the turnover in the current fiscal year is greater than a certain amount, it is a large-scale taxpayer, and the corresponding tax payable after deducting the input tax from the output tax is within a certain range, it is judged to be at risk. In the model, the turnover threshold and tax payable threshold range can be modified and maintained in the threshold management system.

[0086] (12) Abnormal risk model of taxpayers’ illegal deposits

[0087] An offline analysis model that counts and analyzes the amount of goods entering warehouses and the amount of goods entering warehouses through invoices based on the goods entering warehouse data of enterprises and taxpayers in the offline data warehouse. If the total amount of goods entering warehouses of registered taxpayers exceeds a certain amount in the current fiscal year, but the amount of goods entering warehouses through invoices is less than a certain amount, it is judged to have a certain risk. The threshold of the amount of goods entering warehouses and the threshold of the amount of goods entering warehouses through invoices in the model can be modified and maintained in the threshold management system.

[0088] (13) Risk model for suspected abnormal sales of invoices

[0089] An offline analysis model that conducts statistical analysis on the number and amount of invoices based on the invoice data of enterprises and taxpayers in the offline data warehouse. According to a list of illegal invoice purchases provided by management personnel, if the number of invoices issued by an enterprise or taxpayer to the enterprises on the list exceeds a certain value, or the amount of the invoice exceeds a certain amount in the current fiscal year, then such taxpayers are at risk of selling invoices. The invoice quantity and amount thresholds in the model can be modified and maintained in the threshold management system.

[0090] (14) Abnormal risk model for invoicing in different places

[0091] An offline analysis model that conducts statistical analysis on the number and amount of invoices based on the registration and invoice data of enterprises and taxpayers in the offline data warehouse. If the number and amount of invoices issued by an enterprise or taxpayer to a different place exceeds a certain value in the past month, the taxpayer on the selling side is at risk. The thresholds for the number and amount of invoices in the model can be modified and maintained in the threshold management system.

[0092] Preferably, the method further comprises:

[0093] The taxpayer's transaction behavior is scored and rated based on the risk identification results.

[0094] The risk results output by the method of the present invention can also be connected to a risk level assessment system to score and rate the transaction behavior of enterprises or taxpayers, thereby implementing more accurate risk management and control measures.

[0095] The taxpayer risk identification method based on audit rules and relationship maps provided by the present invention can identify and perceive tax-related risks in a combination of offline and real-time. According to the risk indicators constructed according to the audit rules of different regions, the risk assessment of enterprises or taxpayers can be carried out from multiple dimensions, which not only enhances the real-time performance but also ensures the accuracy. And through the indicator management and threshold management in the risk rule engine, the scalability and controllability of risk assessment are greatly increased.

[0096] Figure 3 FIG. 3 is a schematic diagram of the structure of a taxpayer risk identification system 300 based on audit rules and relationship graphs according to an embodiment of the present invention. Figure 3 As shown, the taxpayer risk identification system 300 based on audit rules and relationship graphs provided in an embodiment of the present invention includes: an invoice data acquisition unit 301, a risk rule engine construction unit 302, a graph and risk model construction unit 303, an invoice data real-time acquisition unit 304 and a risk identification unit 305.

[0097] Preferably, the invoice data acquisition unit 301 is used to acquire historical invoice data of taxpayers' invoicing and transaction behaviors stored in an offline data warehouse, and clean and structure the historical invoice data to acquire historical invoice processing data.

[0098] Preferably, the risk rule engine construction unit 302 is used to configure risk indicators and thresholds corresponding to each risk indicator according to the audit rules of different regions to construct a risk rule engine.

[0099] Preferably, the graph and risk model building unit 303 is used to build a taxpayer's transaction relationship graph and risk model based on the historical invoice processing data and the risk indicators in the risk rule engine.

[0100] Preferably, the transaction relationship map includes: a loop transaction risk map and a community transaction risk map.

[0101] Preferably, the loop transaction risk map is constructed using a depth-first search algorithm (DFS), while edge vertex optimization, bidirectional traversal optimization and low-value data screening are used to reduce the complexity of the algorithm, and then combined with multi-threaded parallel computing to output the transaction paths of enterprises or taxpayers with loop transaction risks.

[0102] Preferably, the risk model includes: abnormal risk model for purchased and sold goods, risk model for sudden increase in invoice quantity, risk model for sudden increase in invoice amount, abnormal risk model for invoice cancellation, abnormal risk model for invoicing time, abnormal risk model for associated enterprises, risk model for large-scale invoicing of newly registered enterprises, risk model for large-scale invoicing at the end of the month, risk model for false deductions, abnormal risk model for illegal warehousing, abnormal risk model for suspected sale of invoices and risk model for invoicing in a different place.

[0103] Preferably, the real-time invoice data acquisition unit 304 is used to acquire detailed invoice data of the taxpayer's invoicing and transaction behaviors in real time.

[0104] Preferably, the risk identification unit 305 is used to perform risk identification based on the detailed invoice data, transaction relationship map and risk model, and output risk identification results in combination with thresholds corresponding to each risk indicator in the risk rule engine.

[0105] Preferably, the risk identification unit 305, when performing risk identification based on the community transaction risk map, calculates the input and output tax amounts and the degree of difference based on the taxpayer's invoice and transaction information, and divides the enterprise into communities based on the degree of difference. If the transaction amount exceeds the preset amount and the taxpayer is in an abnormal degree of difference community, it is determined that the taxpayer has an abnormal community transaction risk.

[0106] Preferably, the risk identification unit 305, when performing risk identification based on the invoice quantity sudden increase risk model or the invoice amount sudden increase risk model, collects statistics and monitors the taxpayer's invoicing data or invoice amount within a preset time period, calculates the year-on-year and month-on-month growth rates of the invoice quantity or invoice amount, and if the year-on-year and month-on-month growth rates of the taxpayer exceed the preset threshold, it is determined that the taxpayer is at risk of a sudden increase in invoice quantity or a sudden increase in invoice amount.

[0107] Preferably, the risk identification unit 305, when performing risk identification based on the invoice cancellation abnormality risk model, counts the amount and quantity of the taxpayer's red invoices and all invoices, calculates the ratio of the amount and quantity of red invoices to all invoices; if the ratio of the amount and quantity of the taxpayer's red invoices exceeds a preset threshold, it is determined that the taxpayer has an invoice cancellation abnormality risk.

[0108] Preferably, the risk identification unit 305, when performing risk identification based on the abnormal risk model of associated enterprises, counts and monitors the taxpayer's upstream and downstream transaction enterprises. If there are abnormalities in the upstream and downstream enterprises, the number of transactions between the enterprise and the abnormal enterprise and the proportion of gold in the total transactions of the enterprise are calculated. If the proportion of the number or amount of input transactions, or the proportion of the number or amount of output transactions exceeds the threshold, the taxpayer has abnormal risk of associated enterprises.

[0109] Preferably, the system further comprises:

[0110] A scoring unit is used to score and rate the taxpayer's transaction behavior based on the risk identification results.

[0111] The taxpayer risk identification system 300 based on audit rules and relationship maps of an embodiment of the present invention corresponds to the taxpayer risk identification method 100 based on audit rules and relationship maps of another embodiment of the present invention, which will not be repeated here.

[0112] The invention has been described above with reference to a few embodiments. However, it is readily apparent to a person skilled in the art that other embodiments than the ones disclosed above are equally within the scope of the invention, as defined by the appended patent claims.

[0113] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise therein. All references to "a / said / the [means, components, etc.]" are to be openly interpreted as at least one instance of said means, components, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not necessarily have to be performed in the exact order disclosed, unless explicitly stated otherwise.

[0114] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0115] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0116] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0117] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A taxpayer risk identification method based on audit rules and relationship graphs, characterized in that: The method comprises: Obtain historical invoice data of taxpayers' invoicing and transaction behaviors stored in the offline data warehouse, and clean and structure the historical invoice data to obtain historical invoice processing data; Configure risk indicators and thresholds corresponding to each risk indicator according to the audit rules of different regions to build a risk rule engine; Building a taxpayer's transaction relationship map and risk model based on the historical invoice processing data and risk indicators in the risk rule engine; Obtain detailed invoice data of taxpayers’ invoicing and transaction behaviors in real time; Risk identification is performed based on the detailed invoice data, transaction relationship map and risk model, and the risk identification result is output in combination with the threshold value corresponding to each risk indicator in the risk rule engine.

2. The method according to claim 1, characterized in that The transaction relationship map includes: a loop transaction risk map and a community transaction risk map.

3. The method according to claim 2, characterized in that The loop transaction risk map is constructed using the depth-first search algorithm DFS, while edge vertex optimization, bidirectional traversal optimization and low-value data screening are used to reduce the complexity of the algorithm, and then combined with multi-threaded parallel computing to output the transaction paths of enterprises or taxpayers with loop transaction risks.

4. The method according to claim 2, characterized in that: When performing risk identification based on the community transaction risk map, the method calculates the input and output tax amounts and the degree of difference based on the taxpayer's invoices and transaction information, and divides the enterprises into communities based on the degree of difference. If the transaction amount exceeds the preset amount and the taxpayer is in an abnormal degree of difference community, it is determined that the taxpayer has an abnormal community transaction risk.

5. The method according to claim 1, characterized in that The risk models described include: abnormal risk model for purchase and sale of goods, risk model for sudden increase in invoice quantity, risk model for sudden increase in invoice amount, abnormal risk model for invoice cancellation, abnormal risk model for invoicing time, abnormal risk model for affiliated enterprises, risk model for large number of invoices issued by newly registered enterprises, risk model for large number of invoices issued at the end of the month, risk model for false deductions, abnormal risk model for illegal warehousing, abnormal risk model for suspected sale of invoices and risk model for invoicing in other places.

6. The method according to claim 5, characterized in that When performing risk identification based on the invoice quantity sudden increase risk model or the invoice amount sudden increase risk model, the method collects statistics and monitors the taxpayer's invoice data or invoice amount within a preset time period, calculates the year-on-year and month-on-month growth rates of the invoice quantity or invoice amount, and if the year-on-year and month-on-month growth rates of the taxpayer exceed a preset threshold, it is determined that the taxpayer has a risk of a sudden increase in the number of invoices or a sudden increase in the invoice amount.

7. The method according to claim 5, characterized in that When performing risk identification based on the invoice red cancellation abnormal risk model, the method collects statistics on the amount and quantity of the taxpayer's red invoices and all invoices, calculates the ratio of the amount and quantity of the red invoices to all invoices, and if the ratio of the amount and quantity of the taxpayer's red invoices exceeds a preset threshold, it is determined that the taxpayer has an invoice red cancellation abnormal risk.

8. The method according to claim 5, characterized in that When the method identifies risks based on the abnormal risk model of related enterprises, the upstream and downstream trading enterprises of the taxpayer are counted and monitored. If there are abnormalities in the upstream and downstream enterprises, the number of transactions between the enterprise and the abnormal enterprise and the proportion of gold in the total transactions of the enterprise are calculated. If the number or amount ratio of input transactions, or the number or amount ratio of output transactions exceeds the threshold, the taxpayer has abnormal risk of related enterprises.

9. The method according to claim 1, characterized in that: The method further comprises: The taxpayer's transaction behavior is scored and rated based on the risk identification results.

10. A taxpayer risk identification system based on audit rules and relationship graphs, characterized in that: The system comprises: An invoice data acquisition unit is used to acquire historical invoice data of taxpayers' invoicing and transaction behaviors stored in the offline data warehouse, and to clean and structure the historical invoice data to acquire historical invoice processing data; A risk rule engine construction unit is used to configure risk indicators and thresholds corresponding to each risk indicator according to the audit rules of different regions to build a risk rule engine; A graph and risk model building unit, used to build a transaction relationship graph and risk model of the taxpayer based on the historical invoice processing data and the risk indicators in the risk rule engine; The invoice data real-time acquisition unit is used to acquire the detailed invoice data of the taxpayer's invoicing and transaction behaviors in real time; The risk identification unit is used to perform risk identification based on the detailed invoice data, the transaction relationship map and the risk model, and output the risk identification result in combination with the threshold value corresponding to each risk indicator in the risk rule engine.

Images