A big data-based fiscal supervision data processing method and system

Abstract

The application discloses a kind of big data-based financial supervision data processing method and system.The method comprises: obtaining the project information related to project fund expenditure deviation from multiple internal information carriers;Content processing is carried out on the project information to identify key entities and the logical association between entities;According to the key entity and the logical association, a project semantic knowledge network reflecting the project situation is constructed;With fund expenditure deviation as the starting point, information paths that can explain the rationality of the deviation are sought in the project semantic knowledge network, and the rationality of the fund expenditure deviation is judged according to the evidence strength embodied by the information paths;And according to the rationality judgment result, the corresponding explanation report or risk prompt is generated.The application can verify the legality of fund expenditure deviation and reduce the generation of "false alarms".

Description

Technical Field

[0001] This application relates to the field of fiscal regulatory data processing, specifically to a method and system for automatic processing of fiscal regulatory data based on big data. Background Technology

[0002] In current fiscal oversight, automated data processing systems have been widely adopted to collect, clean, integrate, and analyze massive amounts of heterogeneous fiscal data through automation. This aims to improve compliance verification efficiency and promptly detect potential violations and abnormal fund flows. However, existing systems face significant challenges in determining the legality of deviations between project expenditures and initial plans. Many seemingly abnormal fund flows often involve compliant and reasonable internal adjustments, but this explanatory information is scattered across various non-standardized internal records, leading to frequent false alarms and severely impacting the efficiency and accuracy of oversight.

[0003] Specifically, when existing fiscal oversight systems identify discrepancies between actual project expenditures and the plans described in initial project approval documents, they are unable to effectively acquire and integrate legality explanations scattered across various unstructured or semi-structured data sources, such as project management software, internal knowledge bases, and internal communication records. This explanatory information is diverse in form, dynamically changing in content, and context-dependent. Examples include intermediate records in project change approval processes, internal expert review opinions, project manager's periodic reports, detailed interpretations and applicability statements of specific policy clauses, and various internal communication records generated during project execution. The system can only identify the existence of "deviations" but cannot further understand the "reasonableness" and "legality" behind these "deviations."

[0004] This situation leads to new inefficiencies and a waste of regulatory resources. While the system can identify more "anomalies," its inability to effectively "verify the legitimacy" of these anomalies results in a large number of "false alarms." Regulatory personnel are forced to spend considerable time and effort manually interviewing project leaders, reviewing scattered records in the project management platform, sifting through internal emails, and even contacting relevant departments for detailed inquiries to verify the legitimacy of each of these system-marked "deviations." This process is not only extremely tedious and time-consuming, but also, due to the sheer number of "false alarms" to process daily, genuinely concerning, potentially serious violations are buried under a sea of ​​false alarms. This severely slows down the overall regulatory response speed, significantly diminishing the advantages of the intelligent system. In some ways, the excessive amount of false alarms actually increases the burden of manual review, contradicting the system's original design intent. Summary of the Invention

[0005] This application provides a data processing method and system for fiscal supervision based on big data, which at least solves the problem that existing fiscal supervision systems cannot effectively obtain and integrate legality explanation information scattered in various unstructured or semi-structured data sources when judging the legality of deviations between project fund expenditures and initial plans, resulting in the system frequently issuing "false alarms" and seriously affecting the efficiency and accuracy of supervision work.

[0006] Firstly, this application provides a method for processing fiscal regulatory data, comprising the following steps: Obtain project information from multiple internal information carriers related to deviations in project fund expenditures; The project information is processed to identify key entities and the logical relationships between them. Based on the key entities and logical relationships, construct a project semantic knowledge network that reflects the project context. Starting from the aforementioned deviation in fund expenditure, information paths that can explain the rationality of the deviation are explored in the project semantic knowledge network, and the rationality of the deviation in fund expenditure is judged based on the strength of evidence reflected in the information paths. Based on the reasonableness assessment results, a corresponding explanatory report or risk warning will be generated.

[0007] Optionally, the step of using the deviation in fund expenditure as a starting point to explore information paths in the project semantic knowledge network that can explain the reasonableness of the deviation, and judging the reasonableness of the deviation in fund expenditure based on the strength of evidence reflected in the information paths, includes: Based on the type of the fund expenditure deviation and the stage of the project, adjust the initial credibility of evidence units from different sources and the connection strength of different relation types in the semantic knowledge network of the project; Calculate the sum of the products of the initial credibility and the adjusted connection strength for each evidence unit in the information path; The conflict intensity generated by the conflicting information existing in the information path is accumulated; The overall persuasiveness score of the information path is obtained based on the ratio of the product to the conflict intensity. The reasonableness of the deviation in fund expenditure is judged based on the comprehensive persuasiveness score.

[0008] Optionally, the step of adjusting the initial credibility of evidence units from different sources in the project semantic knowledge network based on the type of the fund expenditure deviation and the stage of the project includes: Identify evidence pairs in the project's semantic knowledge network that have mutual references or dependencies; For the evidence unit pair, determine the influence factor of the reference or dependency type on the degree of influence on credibility; The initial credibility of the cited or dependent evidence unit is adjusted based on the current credibility of the cited or dependent evidence unit and the influence factor.

[0009] Optionally, the step of determining the influence factor of the reference or dependency type on the credibility of the evidence unit pair includes: Obtain the business attribution information of the cited evidence unit and the cited evidence unit; Based on the business attribution information, query the business association rule table to obtain the impact factor adjustment coefficient; The final impact factor is obtained by multiplying the base impact factor of the reference or dependency type with the adjustment coefficient.

[0010] Optionally, the step of adjusting the connection strength of different relation types in the project semantic knowledge network according to the type of the fund expenditure deviation and the stage of the project includes: Obtain the business affiliation information of the entities connected by the relationship, as well as the frequency with which the relationship has been accepted in similar past scenarios; Based on the entity’s own business affiliation information and the frequency with which the relationship has been accepted in similar past scenarios, calculate the influence coefficient of the relationship on the connection strength. The final connection strength is obtained by multiplying the basic connection strength of the relationship type with the influence coefficient.

[0011] Optionally, the step of accumulating the conflict intensity generated by the conflict information existing in the information path includes: Obtain the hierarchical identifier of each conflicting information source in the information path; Based on the hierarchical identifier, determine the hierarchical weight of each conflicting information source; For conflicting information sources that influence each other in the information path, identify their influence type and adjust the hierarchical weight of the affected conflicting information sources according to the influence type. The conflict intensity of each conflicting information source is calculated by multiplying it with its adjusted hierarchical weight. The product is summed to obtain the conflict intensity caused by conflicting information in the information path.

[0012] Optionally, the step of identifying the influence type of conflicting information sources that influence each other in the information path, and adjusting the hierarchical weight of the affected conflicting information sources according to the influence type, includes: Obtain the identity information of the entity that publishes the conflict information source, the historical credit record of the entity, and the criticality identifier of the business process associated with the conflict information source; Based on the identity information of the publishing entity, the historical credit records, and the criticality indicator of the business process, identify the impact type of the conflicting information source; Based on the identified types of impact, the degree of their influence on the hierarchical weight of the affected conflict information sources is quantified to obtain the weight adjustment range; Based on the aforementioned weight adjustment range, the hierarchical weights of the affected conflicting information sources are adjusted.

[0013] Optionally, the step of obtaining the historical credit records of the publishing entity includes: Identify the identity of the publishing entity in different business systems; According to the preset mapping rules, the identity identifiers in the different business systems are unified into a unique global identity identifier; For the global identity identifier, historical behavior data related to the publishing entity is periodically acquired from the multiple business systems according to a preset data collection strategy. The historical behavior data includes the publishing entity's operation records such as information publishing, approval, modification, and business process participation in each business system. The acquired historical behavior data is format-converted and content-parsed to extract key credit elements, which include the accuracy of information release, compliance of approval, efficiency of decision execution, and records of abnormal behavior. Based on the key credit elements and combined with a preset set of credit assessment rules, the historical behavior of the publishing entity is quantitatively assessed to generate a local credit score for the publishing entity in each business system. Based on the local credit scores of the publishing entity in each business system, and combined with the weight of each business system in the overall fiscal supervision, the local credit scores of the publishing entity are weighted and summarized to obtain the unified historical credit record of the publishing entity.

[0014] Optionally, the step of accumulating the conflict intensity generated by the conflict information existing in the information path includes: Obtain the hierarchical identifier of each conflicting information source in the information path, and determine the hierarchical weight of each conflicting information source based on the hierarchical identifier; Obtain the identity information of the publishing entity of each conflict information source, the organizational structure relationship between the publishing entities, and the frequency of historical collaborative behavior between the publishing entities; Based on the organizational structure and the frequency of historical collaborative behavior, identify whether there is collusion or conspiracy among the publishing entities; If collusion or conspiracy is identified, the conflict information sources of the collusion or conspiracy are deduplicated, and the conflict intensity of the conflict information sources is adjusted according to the nature of the collusion or conspiracy. The conflict intensity of each conflicting information source is calculated by multiplying it with its adjusted hierarchical weight. The product is summed to obtain the conflict intensity caused by conflicting information in the information path.

[0015] Secondly, this application provides a big data-based fiscal supervision data processing system for processing fiscal supervision data, the system comprising: The information acquisition module is used to acquire project information related to deviations in project fund expenditures from multiple internal information carriers; The content processing module is used to process the project information to identify key entities and logical relationships between them. The knowledge network construction module is used to construct a project semantic knowledge network that reflects the project context based on the key entities and the logical relationships. The reasonableness judgment module is used to explore information paths in the project semantic knowledge network that can explain the reasonableness of the deviation, starting from the deviation in fund expenditure; and to judge the reasonableness of the deviation in fund expenditure based on the strength of evidence reflected in the information paths. The report generation module is used to generate corresponding explanatory reports or risk warnings based on the reasonableness judgment results.

[0016] Compared with related technologies, the big data-based fiscal supervision data processing method and system provided in this application have at least the following technical advantages: By acquiring project information from multiple internal information carriers and processing this information to identify key entities and logical connections, a project semantic knowledge network reflecting the project context is constructed. Based on this, starting with deviations in fund expenditures, information paths explaining the rationality of these deviations are explored within the semantic knowledge network. The rationality of the deviations is judged based on the strength of evidence embodied in these information paths, and an explanation report or risk warning is generated based on the judgment results. This method effectively solves the problem of existing fiscal supervision systems failing to effectively acquire and integrate explanatory information scattered across various non-standardized internal records when judging the legality of deviations between project fund expenditures and initial plans, leading to frequent false alarms. By constructing a project semantic knowledge network, this application structurally integrates scattered, unstructured explanatory information and utilizes knowledge reasoning mechanisms to conduct in-depth analysis of the rationality of deviations. This significantly improves the accuracy of judging the legality of fund expenditure deviations, reduces the number of false alarms, greatly enhances the efficiency and accuracy of fiscal supervision, and avoids the waste of supervisory resources.

[0017] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart illustrating a fiscal regulatory data processing method according to an exemplary embodiment.

[0019] Figure 2 This is a flowchart illustrating step S4 of exploring information paths and determining their reasonableness according to an exemplary embodiment.

[0020] Figure 3 This is a flowchart illustrating step S43, which involves accumulating conflict information, according to an exemplary embodiment.

[0021] Figure 4 This is a flowchart illustrating step S43322 of obtaining the historical credit records of the publishing entity according to an exemplary embodiment.

[0022] Figure 5 This is a flowchart illustrating step S43, collusion identification and processing, according to an exemplary embodiment.

[0023] Figure 6 This is a block diagram illustrating a fiscal regulatory data processing system according to an exemplary embodiment. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0025] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0026] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0027] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0028] Example 1 This application provides a method for processing fiscal regulatory data based on big data. Figure 1This is a flowchart illustrating a fiscal regulatory data processing method according to an exemplary embodiment. For example... Figure 1 As shown, the method includes the following steps: S1. Obtain project information related to deviations in project fund expenditures from multiple internal information carriers.

[0029] In this embodiment, project expenditure deviation refers to the difference between actual project expenditure and the initial plan or budget. These deviations may stem from various reasons, such as project changes, policy adjustments, and market fluctuations. Internal information carriers encompass various data sources generated during project management, including but not limited to project management systems, financial systems, contract management systems, internal communication platforms, and document management systems. When the system detects over-budget expenditure in a specific funding category for a project, it can automatically trigger an information retrieval process. This involves automatically extracting project progress reports and budget execution status from the project management system via API interfaces; obtaining actual payment vouchers and contract texts from the financial system via data integration tools; or retrieving data from multiple sources, such as project change requests and approval records, from the internal document management system using web scraping technology.

[0030] S2. Perform content processing on the project information to identify key entities and the logical relationships between them.

[0031] In this embodiment, content processing can employ Natural Language Processing (NLP) techniques, such as Named Entity Recognition (NER) models to identify key entities like project names, funding categories, suppliers, approvers, contract numbers, and policy documents from unstructured text. Relationship extraction models further identify the logical connections between these entities, such as "Project A is related to Contract B," "Contract B involves Supplier C," and "Supplier C's payment was approved by Approver D." Simultaneously, entities and relationships can be directly extracted from structured or semi-structured data through predefined rules and template matching.

[0032] S3. Based on the key entities and the logical relationships, construct a project semantic knowledge network that reflects the project context.

[0033] In this embodiment, the project semantic knowledge network is a graph structure, where nodes represent key entities and edges represent logical relationships between entities. For example, a graph database (such as Neo4j) can be used to store and manage this knowledge network. Each entity node contains its attribute information; for example, a project name node may contain attributes such as project number, start date, and person in charge. Each relationship edge may contain attributes such as relationship type, timestamp, and source of evidence. In this way, heterogeneous information scattered across different information carriers can be integrated into a unified, structured knowledge representation, thus providing a foundation for subsequent analysis.

[0034] S4. Starting from the deviation in fund expenditure, explore information paths in the project semantic knowledge network that can explain the rationality of the deviation, and judge the rationality of the deviation in fund expenditure based on the strength of evidence reflected in the information paths.

[0035] In this embodiment, when a discrepancy is found between a expenditure and the budget, this expenditure can be used as the starting node. A graph traversal or path search is then performed within the knowledge network to find a series of interconnected entities and relationships that collectively explain the cause of the discrepancy. For example, an information path might be: "Expenditure Discrepancy" → "Related Contracts" → "Contract Change Approval Records" → "Approver" → "Policy Document Interpretation" → "Project Manager Report". Subsequently, based on factors such as the credibility of the evidentiary units contained in the path, the strength of the relationships, and the presence of conflicting information, the evidentiary strength of the path is comprehensively evaluated, thereby determining the reasonableness of the discrepancy.

[0036] S5. Based on the reasonableness judgment result, generate a corresponding explanation report or risk warning.

[0037] In this embodiment, if an information path contains multiple official approval documents and authoritative policy interpretations without conflicting information, the path has high evidentiary strength, the deviation is judged as reasonable, and the system automatically generates a detailed explanation report; conversely, if the evidence is weak, there is conflicting information, or a complete explanation path cannot be found, the deviation may be judged as unreasonable, the system generates a risk warning and suggests manual intervention for review.

[0038] In the technical solution of the above embodiments, the fiscal supervision data processing method of this application first comprehensively collects project information related to deviations in fund expenditures through multi-source information acquisition; secondly, it uses natural language processing and other technical means to accurately identify key entities and their logical relationships in the information; then, it constructs a project semantic knowledge network to integrate scattered and heterogeneous information into a unified structured knowledge representation; based on this, it explores explanatory information paths in the knowledge network starting from the deviation, and makes an objective judgment on the rationality of the deviation based on the strength of evidence; finally, it outputs an explanation report or risk warning. The entire method forms a complete closed-loop link from information collection, semantic understanding, knowledge modeling to reasoning and judgment, overcoming the fundamental defect of traditional supervision systems that can only identify deviations but cannot verify their rationality.

[0039] In one possible design, Figure 2 This is a flowchart illustrating step S4, which involves exploring information paths and determining their reasonableness, according to an exemplary embodiment. (See attached diagram.) Figure 2 Step S4 includes: S41. Based on the type of the fund expenditure deviation and the stage of the project, adjust the initial credibility of evidence units from different sources and the connection strength of different relation types in the project semantic knowledge network.

[0040] In this step, the specific business context must be fully considered when judging the reasonableness of expenditure deviations. Types of expenditure deviations include, but are not limited to, over-budget expenditures, expenditures with mismatched purposes, and expenditures lagging behind schedule. Different types of deviations may require attention to different aspects of evidence. The project stage includes the initiation stage, implementation stage, acceptance stage, and final settlement stage, etc., and the information characteristics and risk points differ at each stage. By combining this contextual information, the initial credibility of each evidentiary unit can be dynamically adjusted (e.g., appropriately increasing the credibility of evidence highly related to the current deviation type) and the connection strength of different relationship types in different contexts (e.g., the connection strength of the evidence chain related to contract execution is higher in the project implementation stage).

[0041] S42. Calculate the sum of the product of the initial credibility and the adjusted connection strength of each evidence unit in the information path.

[0042] In this step, along the information path, the current credibility of each piece of evidence on the path is multiplied by the connection strength of its subsequent relation types, and these products are summed. This sum of products can be seen as a quantitative representation of the positive support provided by the information path, reflecting the cumulative support strength of all evidence and relations along the path.

[0043] S43. Accumulate the conflict intensity generated by the conflicting information existing in the information path.

[0044] In this step, contradictory or inconsistent information may be encountered while exploring the information path. This conflicting information can weaken the overall persuasiveness of the information path. Therefore, it is necessary to identify this conflicting information and quantify its intensity. The intensity of conflict can be determined and accumulated based on the source, nature, severity, and impact on the core facts of the conflicting information to reflect the sum of negative evidence.

[0045] S44. Based on the product and the ratio of the conflict intensity, obtain the comprehensive persuasiveness score of the information path.

[0046] In this step, a comprehensive score, namely the comprehensive persuasiveness score, is obtained by calculating the ratio of the product representing positive support to the conflict intensity representing negative impact. This score can comprehensively measure the overall ability of an information path to explain the rationality of deviations in fund expenditure. The higher the score, the more persuasive the explanation provided by the path.

[0047] S45. Based on the comprehensive persuasiveness score, determine the reasonableness of the deviation in fund expenditure.

[0048] In this step, after obtaining the comprehensive persuasiveness score, the score can be compared with the preset judgment threshold or rule to make a final judgment on the reasonableness of the deviation in fund expenditure.

[0049] In the technical solution of the above embodiments, this application dynamically adjusts the credibility and relational connection strength of evidence units in the project semantic knowledge network by introducing considerations of the type of fund expenditure deviation and project stage, making the value assessment of evidence more context-adaptive. It is precisely this contextual adjustment that allows the subsequently calculated product sum to more accurately reflect the positive support provided by the path under a specific regulatory background. Simultaneously, by explicitly accumulating the conflict intensity generated by conflict information, the impact of negative evidence is effectively quantified. Finally, a comprehensive persuasiveness score is obtained by calculating the ratio of positive support to negative impact, providing a balanced and comprehensive quantitative indicator. This effectively solves the potential one-sidedness or subjectivity problems that may exist in traditional methods when quantifying the strength of evidence, making the judgment on the reasonableness of fund expenditure deviations more scientific and reliable.

[0050] In one example, suppose a financial oversight system detects an overspending discrepancy in a project currently in the implementation phase. First, the system adjusts the project's semantic knowledge network based on the discrepancy type "overspending" and the project phase "implementation phase": for evidence units related to budget execution, contract changes, and project progress (such as financial statements, contract texts, and project supervision reports), their initial credibility is appropriately increased to 0.85 (an increase of approximately 21% from the base value of 0.7); for relationship types such as "contract changes leading to increased costs" and "design changes causing budget adjustments," their connection strength is also enhanced to 0.92 (an increase of approximately 23% from the base value of 0.75). Next, the system explores information paths within the adjusted knowledge network: "overspending" → "contract changes" → "design changes" → "expert review opinions" → "approval records." The system calculates the sum of the products of the initial credibility and the adjusted connection strength for each evidence unit in this path to be 2.68. During this process, the system also identified significant differences in the completion status of the "progress report submitted by the project manager" and the "third-party supervision report," with the accumulated conflict intensity reaching 0.82. Subsequently, the system calculated the ratio of the product sum of 2.68 to the conflict intensity of 0.82, resulting in a comprehensive persuasiveness score of 3.27. Finally, this score was compared with a preset threshold of 2.5. Since 3.27 > 2.5, the system determined that the over-budget expenditure was reasonable and generated a corresponding explanatory report. In this way, this application provides a quantitative, objective, and contextualized basis for judgment.

[0051] In one possible design, the step of adjusting the initial credibility of evidence units from different sources as described in S41 includes: S411. Identify evidence pairs in the project's semantic knowledge network that have mutual references or dependencies.

[0052] In this step, by analyzing the semantic links and structural relationships between evidence units in the project's semantic knowledge network, we discover situations where the content or validity of one evidence unit depends on one or more other evidence units. For example, a project budget report may cite data or conclusions from a project feasibility study report; in this case, the budget report and the feasibility study report constitute a pair of evidence units under a citation relationship.

[0053] S412. For the evidence unit pair, determine the influence factor of the citation or dependency type on the degree of influence of credibility.

[0054] In this step, different types of citations or dependencies (such as direct citations, data support, logical deductions, etc.) have different degrees of impact on credibility. This impact factor is used to reflect this difference. For example, direct citations may have a higher impact factor, such as 0.8, while indirect references may have a lower impact factor, such as 0.4.

[0055] S413. Based on the current credibility of the cited or dependent evidence unit and the influence factor, adjust the initial credibility of the cited or dependent evidence unit.

[0056] In this step, the initial credibility of the cited or relied-upon evidence unit is adjusted by combining its current assessed credibility with a pre-determined influence factor. This adjustment mechanism ensures the transmission and influence of credibility within the chain of evidence, enabling the final initial credibility to more accurately reflect the true reliability of the evidence unit within the entire knowledge network.

[0057] In the technical solutions of the above embodiments, this application introduces consideration of the reference or dependency relationships between evidence units. When the validity of one evidence unit largely depends on another evidence unit, if the credibility of the relied-upon evidence unit is low, the initial credibility of the evidence unit that relies on it is correspondingly weakened; conversely, if the relied-upon evidence unit has extremely high credibility, the initial credibility of the evidence unit that relies on it can be enhanced. This mechanism makes the initial credibility of evidence units no longer isolated but closely integrated with their context and relevance in the entire project's semantic knowledge network, thereby constructing a more realistic and reliable evidence credibility system.

[0058] In one possible design, the steps in S412 to determine the factors influencing the degree of influence of reference or dependency type on credibility include: S4121. Obtain the business attribution information of the cited evidence unit and the cited evidence unit.

[0059] In this step, business attribution information refers to the specific business context of the evidence unit, such as the department, function, or project to which it belongs. For example, the business attribution information of an evidence unit concerning a procurement contract can be identified as "procurement department".

[0060] S4122. Based on the business attribution information, query the business association rule table to obtain the impact factor adjustment coefficient.

[0061] In this step, the business association rule table is a pre-set database or rule set that stores the mapping relationship between business attribution information and influence factor adjustment coefficients. It can be built and maintained based on historical data analysis, expert experience, or industry standards.

[0062] S4123. Multiply the basic impact factor of the reference or dependency type with the adjustment coefficient to obtain the final impact factor.

[0063] In this step, by multiplying the preset basic impact factor for a specific reference or dependency type with the adjustment coefficient obtained from the business association rule table, a final impact factor that better fits the current business context is obtained. This makes the determination of the impact factor no longer a single static one, but can be dynamically adjusted according to the specific business context.

[0064] In the technical solution of the above embodiments, by introducing business affiliation information and business association rules, the influencing factors are dynamically adjusted, making the credibility correction process more in line with the actual business and enhancing the accuracy of the judgment.

[0065] In one possible design, the step of adjusting the connection strength of different relationship types as described in S41 includes: S421. Obtain the business affiliation information of the entity connected by the relationship, and the frequency with which the relationship has been accepted in similar past situations.

[0066] In this step, the entity business attribution information reflects the attribution of the two or more entities connected by the relationship in terms of organizational structure, functional departments, etc.; the acceptance frequency reflects the ratio of the number of times that the information carried by this type of relationship in similar past projects has been judged as reliable and valid by the system or by humans to the total number of times.

[0067] S422. Calculate the influence coefficient of the relationship on the connection strength based on the entity's own business affiliation information and the frequency with which the relationship has been accepted in similar past scenarios.

[0068] In this step, a rule base or training model can be established, using business affiliation information and the frequency of acceptance as inputs and outputting a quantified influence coefficient. For example, if the entities connected by a relationship all belong to core business departments and have a high historical acceptance frequency, their influence coefficient may be greater than 1 to enhance the strength of their connection.

[0069] S423. Calculate the final connection strength by multiplying the basic connection strength of the relationship type with the influence coefficient.

[0070] In this step, the base connection strength is the default strength of this relationship type without any specific contextual information, usually preset by expert experience or derived from a large amount of historical data. By multiplying it by a dynamically calculated influence coefficient, the connection strength can be adaptively adjusted according to the specific business scenario while maintaining its inherent properties.

[0071] In the technical solutions of the above embodiments, this application introduces two dimensions—the entity's own business affiliation and the frequency of historical acceptance—to finely adjust the strength of relationship connections. This ensures that relationships that connect key business entities and have a good historical performance receive higher weight when exploring information paths, making the evaluation of information paths more closely reflect actual business logic and risk conditions.

[0072] In one possible design, Figure 3 This is a flowchart illustrating step S43, which involves accumulating conflict information, according to an exemplary embodiment. (Refer to the attached diagram.) Figure 3 Step S43 includes: S431. Obtain the hierarchical identifier of each conflicting information source in the information path.

[0073] In this step, the hierarchy identifier is used to identify the status or importance level of each conflicting information source in the overall financial oversight system. For example, information sources released by the audit department may be assigned a higher hierarchy identifier, while internal reports from project implementation departments may be assigned a lower hierarchy identifier.

[0074] S432. Determine the hierarchical weight of each conflicting information source based on the hierarchical identifier.

[0075] In this step, corresponding weight values ​​are assigned to conflict information sources identified at different levels based on preset rules or expert experience. The higher the level of the information source, the greater its weight is usually to reflect its influence in decision-making.

[0076] S433. For conflicting information sources that influence each other in the information path, identify their influence type and adjust the hierarchical weight of the affected conflicting information sources according to the influence type.

[0077] In this step, we analyze whether there are direct or indirect connections between conflicting information sources (such as confirmation, rebuttal, supplementation, etc.). When the impact of the rebuttal type is identified, the hierarchical weight of the refuted information source will be appropriately lowered; when the impact of the confirmation type is identified, the hierarchical weight of the confirmed information source will be appropriately increased to more accurately reflect its actual credibility in the current context.

[0078] S434. Calculate the product of the conflict intensity of each conflict information source and its adjusted hierarchical weight.

[0079] In this step, after considering the inherent conflict intensity of the information source, its hierarchical importance in the system, and their mutual influence, a weighted conflict intensity value is obtained.

[0080] S435. Accumulate the product to obtain the conflict intensity caused by the conflicting information in the information path.

[0081] In this step, all weighted conflict intensity values ​​are summarized to obtain a comprehensive and more representative total conflict intensity, which is used for subsequent rationality assessment.

[0082] In the technical solutions of the above embodiments, this application introduces hierarchical identifiers and weights for conflicting information sources, and further considers the mutual influence between conflicting information sources, so that information sources of different importance or authority are given different influences in the calculation of conflict intensity. When a higher-level information source refutes a lower-level information source, the weight of the lower-level information source will be reduced accordingly, thereby more accurately reflecting its actual conflict contribution in the current context. Finally, a more comprehensive and accurate total conflict intensity is obtained by calculating and accumulating the weighted product, providing a more reliable basis for subsequent judgments on the reasonableness of deviations in fund expenditures.

[0083] In one example, suppose there are three conflicting information sources in the information path when judging the reasonableness of a project's expenditure deviation: Conflicting information source A comes from an internal report from the project execution department claiming that a certain expenditure is reasonable, its hierarchical identifier is "internal department report" and its initial conflict intensity is 0.6; Conflicting information source B comes from the preliminary audit opinion of the audit department pointing out that the expenditure is questionable, its hierarchical identifier is "audit opinion" and its initial conflict intensity is 0.8; Conflicting information source C comes from the on-site records of the project supervision unit, which contradicts the content of conflicting information source A's report, its hierarchical identifier is "supervision record" and its initial conflict intensity is 0.7. First, the hierarchical weights are determined according to preset rules: internal department report 0.5, audit opinion 0.9, supervision record 0.7. Then, the mutual influence between conflicting information sources is identified: it is found that the audit opinion has a rebuttal effect on the internal report, and the supervision record also has a rebuttal effect on the internal report. Based on the type of rebuttal effect, the hierarchical weight of conflicting information source A is adjusted from 0.5 to 0.4 (a reduction of 20%). Then, the weighted conflict intensity is calculated: Conflicting information source A = 0.6 × 0.4 = 0.24; Conflicting information source B = 0.8 × 0.9 = 0.72; Conflicting information source C = 0.7 × 0.7 = 0.49. Finally, the total conflict intensity is obtained by summing the values: 0.24 + 0.72 + 0.49 = 1.45. This method yields the total conflict intensity that comprehensively considers the information source hierarchy, mutual influence, and inherent conflict intensity.

[0084] In one possible design, the steps in S433 to identify the impact type of conflicting information sources and adjust the hierarchical weights of the affected conflicting information sources include: S4331. Obtain the identity information of the publishing entity of the conflict information source, the historical credit record of the publishing entity, and the criticality identifier of the business process associated with the conflict information source.

[0085] In this step, the identity information of the publishing entity clarifies the responsible party for the source of the information (such as which department, individual, or system), the historical credit record reflects the comprehensive performance of the publishing entity in terms of the accuracy and compliance of its past information releases, and the business process criticality indicator indicates the importance of the business process involved in the conflicting information source in the overall fiscal supervision.

[0086] S4332. Obtain the historical credit records of the publishing entity.

[0087] In this step, a unified credit record is established by comprehensively collecting and quantitatively evaluating the historical behavior of the publishing entity across multiple business systems, providing a reliable basis for subsequent identification of impact types.

[0088] S4333. Identify the impact type of the conflicting information source based on the identity information of the publishing entity, the historical credit record, and the criticality identifier of the business process.

[0089] In this step, based on preset rules or models and combined with multidimensional factors, the source of conflicting information is identified as belonging to specific types such as authority conflict, data accuracy conflict, or process compliance conflict.

[0090] S4334. Based on the identified type of influence, quantify the degree of influence on the weight of the affected conflict information source hierarchy to obtain the weight adjustment range.

[0091] In this step, erroneous information from highly authoritative sources may have a greater negative impact on other information, and therefore the adjustment range will be larger. Finally, the hierarchical weights of the affected conflicting information sources are corrected based on the magnitude of the weight adjustment.

[0092] In the technical solution of the above embodiments, by deeply analyzing the identity of the publishing entity of the conflict information source, historical credit records, and the criticality of the business process, the identification of the impact type of the conflict information source is more accurate. Then, the degree of influence of the conflict information source on the hierarchical weight of the affected conflict information source is scientifically quantified and adjusted accordingly, which significantly improves the accuracy and reliability of conflict intensity calculation.

[0093] In one possible design, Figure 4 This is a flowchart illustrating step S43322, which involves obtaining the historical credit records of the publishing entity, according to an exemplary embodiment. (Refer to the attached diagram.) Figure 4 Step S43322 includes: S433221. Identify the identity of the publishing entity in different business systems.

[0094] In this step, given the complex environment of fiscal oversight, a publishing entity may have different accounts or user IDs in multiple business systems (such as budget management systems, procurement management systems, asset management systems, project management systems, etc.). This step aims to comprehensively collect these dispersed identity information.

[0095] S433222. According to the preset mapping rules, the identity identifiers in the different business systems are unified into a unique global identity identifier.

[0096] In this step, standardized mapping rules (such as matching based on unique information like name, ID number, and organization code) are established to associate and integrate the identity identifiers of the same publishing entity in different systems, forming a unique global identity identifier within the entire fiscal supervision system, eliminating information silos and ensuring comprehensive consistency in credit assessment.

[0097] S433223. For the global identity identifier, historical behavior data related to the publishing entity is periodically acquired from the multiple business systems according to a preset data collection strategy. The historical behavior data includes operation records such as information publishing, approval, modification, and business process participation of the publishing entity in each business system.

[0098] In this step, the system periodically (e.g., daily, weekly, or monthly) extracts all historical operation records associated with the global identity from all related business systems according to a preset collection strategy. These records reflect in detail the specific behaviors of the publishing entity in different business processes.

[0099] S433224. The obtained historical behavior data is format-converted and content-parsed to extract key credit elements, including the accuracy of information release, approval compliance, decision-making execution efficiency, and abnormal behavior records.

[0100] In this step, natural language processing technology is used to parse text descriptions and a rule engine is used to identify specific operation patterns, thereby quantifying the accuracy of the information release by the publishing entity (such as data error rate), the compliance of the approval process (such as whether regulations are violated), the efficiency of decision execution (such as response time), and whether there are any abnormal behaviors (such as frequent modification of key data, unauthorized operations, etc.).

[0101] S433225. Based on the key credit elements and combined with a preset set of credit assessment rules, the historical behavior of the publishing entity is quantitatively assessed to generate a local credit score for the publishing entity in each business system.

[0102] In this step, the credit assessment rule set includes a series of scoring standards and weights, such as information release accuracy accounting for 30%, approval compliance accounting for 40%, decision execution efficiency accounting for 20%, and abnormal behavior records accounting for 10%. Based on this, the system generates an independent local credit score for the publishing entity in each business system.

[0103] S433226. Based on the local credit scores of the publishing entity in each business system and combined with the weight of each business system in the overall financial supervision, the local credit scores of the publishing entity are weighted and summarized to obtain the unified historical credit record of the publishing entity.

[0104] In this step, different business systems may have varying degrees of importance in fiscal oversight; for example, the credit score of the budget system may have a higher weight than that of the asset management system. A weighted average can be used to obtain a comprehensive, unified historical credit record that reflects the issuing entity's creditworthiness throughout the entire fiscal oversight system.

[0105] In the technical solution of the above embodiments, this application systematically integrates the scattered information of the publishing entity in multiple business systems and performs multi-dimensional quantitative credit assessment of its historical behavior. First, it solves the problem of information dispersion by identifying and unifying the identity identifiers in different systems; second, it comprehensively captures the behavioral characteristics of the publishing entity by periodically collecting and parsing detailed historical behavior data; third, it avoids subjective bias by extracting key credit elements and combining them with assessment rules to achieve quantitative assessment; and finally, it considers the differences in importance between systems by weighting and summarizing the local credit scores of each business system, making the final unified historical credit record more representative and authoritative.

[0106] In one possible design, Figure 5 This is a flowchart illustrating step S43, collusion identification and processing, according to an exemplary embodiment. (Refer to the attached document.) Figure 5 Step S43 also includes another implementation: S43a. Obtain the hierarchical identifier of each conflicting information source in the information path, and determine the hierarchical weight of each conflicting information source based on the hierarchical identifier.

[0107] In this step, after obtaining the hierarchical identifier of each conflicting information source and determining its hierarchical weight, it is also necessary to further consider the potential collaborative relationships between the publishing entities.

[0108] S43b: Obtain the identity information of the publishing entity of each conflict information source, the organizational structure relationship between the publishing entities, and the frequency of historical collaborative behavior between the publishing entities.

[0109] In this step, identifying the identity information of the publishing entities clarifies the source of the conflicting information. Organizational structure relationships refer to the subordinate, collaborative, or hierarchical relationships of these publishing entities within the organization, which helps determine whether there is a natural basis for collaboration between them. The frequency of historical collaborative behavior reflects the degree of past joint actions by the publishing entities. High-frequency collaborative behavior may suggest the possibility of potential collusion.

[0110] S43c. Based on the organizational structure relationship and the frequency of the historical collaborative behavior, identify whether there is collusion or conspiracy among the publishing entities.

[0111] In this step, if the two publishing entities are under the same superior and their historical collaboration frequency exceeds a certain threshold, it can be preliminarily determined that there is a risk of collusion or conspiracy, and a preset rule set or machine learning model can be used for analysis and judgment.

[0112] S43d. If collusion or conspiracy is identified, the conflict information sources of the collusion or conspiracy are deduplicated, and the conflict intensity of the conflict information sources is adjusted according to the nature of the collusion or conspiracy.

[0113] The purpose of deduplication in this step is to avoid inflated conflict intensity due to multiple sources of information repeatedly expressing the same conflicting viewpoint. For example, all sources of information are treated as a whole, and their conflict intensity is calculated only once. At the same time, the conflict intensity is appropriately adjusted according to the nature of the conspiracy or collusion (scope, duration, potential severity, etc.): the conflict intensity is amplified in cases of malicious conspiracy to reflect a more serious risk, while it is appropriately reduced or merged in cases of unintentional collusion.

[0114] S43e, Calculate the product of the conflict intensity of each conflict information source and its adjusted hierarchical weight.

[0115] In this step, the conflict information sources after collusion identification and processing are weighted and calculated.

[0116] S43f: Accumulate the product to obtain the conflict intensity caused by the conflicting information in the information path.

[0117] In this step, the final total conflict strength is obtained by summing all the weighted conflict strength values ​​after deduplication and adjustment.

[0118] In the technical solution of the above embodiments, by introducing the identification and processing of collusion or conspiracy among conflicting information sources, when multiple publishing entities collude or conspire, the conflicting information they publish often has consistency or coordination. If not distinguished, it will be regarded as independent evidence and simply accumulated, thus artificially increasing the intensity of the conflict. By obtaining the identity information, organizational structure relationship, and frequency of historical collaborative behavior of the publishing entities, the inherent relationship between conflicting information sources is analyzed in depth. Once such behavior is identified, the colluding information sources are deduplicated to ensure that each independent viewpoint is only effectively calculated once. At the same time, the intensity of the conflict is adjusted according to the nature of the collusion to make the assessment more accurate and objective.

[0119] In one example, suppose that when judging the reasonableness of a project's expenditure deviation, there are two approval records for a procurement contract in the information path: one submitted by Department A and the other by Department B. These two records conflict on key terms (procurement amount or supplier information). First, the system obtains the identity information of the issuing entities of Department A and Department B. By querying the organizational structure, it finds that Department A and Department B belong to the same superior Department C, and in the historical collaborative behavior records, the two departments have frequently engaged in joint approval and information sharing in 12 similar procurement projects, with a collaboration frequency exceeding the threshold of 80%. Based on this information, the system identifies a high risk of collusion or conspiracy between Department A and Department B. Once collusion is identified, the system deduplicates the two conflicting information sources, no longer treating them as two independent conflict points simply added together, but merging them into a single "collaborative conflict information source." Based on the nature of the collusion (determined to be unintentional collusion due to poor internal communication rather than malicious collusion), the intensity of the merged conflict is appropriately reduced. Finally, the system multiplies and sums the adjusted conflict intensity of each conflicting information source with its hierarchical weight to obtain the total conflict intensity in the information path. This method avoids inflated conflict intensity due to collusion between departments A and B, making the assessment of the reasonableness of deviations in the procurement contract's related expenditures more accurate and fair.

[0120] In summary, the big data-based fiscal supervision data processing method provided by this invention achieves a closed loop from multi-source heterogeneous information collection, semantic understanding, knowledge modeling to reasoning and judgment by connecting five core steps: acquiring project information related to project fund expenditure deviations, processing the project information to identify key entities and logical relationships, constructing a project semantic knowledge network, exploring information paths in the knowledge network starting from fund expenditure deviations and making reasonableness judgments based on the strength of evidence, and generating explanatory reports or risk warnings based on the judgment results. This effectively solves the fundamental defect of traditional fiscal supervision systems that can only identify deviations but cannot verify their reasonableness, ultimately reducing the number of false alarms.

[0121] Furthermore, by applying key technologies such as comprehensive persuasiveness score calculation, evidence unit credibility correction, refined accumulation of conflict intensity, and collusion identification and processing, a multi-dimensional, quantitative, and contextualized assessment of the reasonableness of fund expenditure deviations can be further realized, thereby improving the level of intelligence in fiscal supervision and decision support capabilities.

[0122] Example 2 Embodiment 2 of this application provides a financial supervision data processing system based on big data. Figure 6 This is a block diagram illustrating a fiscal regulatory data processing system according to an exemplary embodiment. Figure 6 As shown, the system includes: Information acquisition module 01 is used to acquire project information related to project fund expenditure deviations from multiple internal information carriers; Content processing module 02 is used to process the project information to identify key entities and logical relationships between them. The knowledge network construction module 03 is used to construct a project semantic knowledge network that reflects the project context based on the key entities and the logical associations. The reasonableness judgment module 04 is used to explore information paths that can explain the reasonableness of the deviation in the project semantic knowledge network, starting from the deviation in fund expenditure, and to judge the reasonableness of the deviation in fund expenditure based on the strength of evidence reflected in the information paths. The report generation module 05 is used to generate a corresponding explanatory report or risk warning based on the reasonableness judgment result.

[0123] The big data-based fiscal supervision data processing system provided in Embodiment 2 of this invention, through the coordinated operation of the information acquisition module 01, content processing module 02, knowledge network construction module 03, rationality judgment module 04, and report generation module 05, realizes automated in-depth analysis and judgment of the rationality of deviations in fiscal project fund expenditures. It solves the problem of frequent false alarms and low supervision efficiency caused by the inability to effectively acquire and integrate scattered explanatory information in existing fiscal supervision work, and improves the intelligence level and accuracy of fiscal supervision.

[0124] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for processing fiscal regulatory data based on big data, characterized in that, Includes the following steps: Obtain project information from multiple internal information carriers related to deviations in project fund expenditures; The project information is processed to identify key entities and the logical relationships between them. Based on the key entities and logical relationships, construct a project semantic knowledge network that reflects the project context. Starting from the aforementioned deviation in fund expenditure, information paths that can explain the rationality of the deviation are explored in the project semantic knowledge network, and the rationality of the deviation in fund expenditure is judged based on the strength of evidence reflected in the information paths. Based on the reasonableness assessment results, a corresponding explanatory report or risk warning will be generated.

2. The method according to claim 1, characterized in that, The steps of starting with the deviation in fund expenditure, exploring information paths in the project semantic knowledge network that can explain the reasonableness of the deviation, and judging the reasonableness of the deviation in fund expenditure based on the strength of evidence reflected in the information paths include: Based on the type of the fund expenditure deviation and the stage of the project, adjust the initial credibility of evidence units from different sources and the connection strength of different relation types in the semantic knowledge network of the project; Calculate the sum of the products of the initial credibility and the adjusted connection strength for each evidence unit in the information path; The conflict intensity generated by the conflicting information existing in the information path is accumulated; The overall persuasiveness score of the information path is obtained based on the ratio of the product to the conflict intensity. The reasonableness of the deviation in fund expenditure is judged based on the comprehensive persuasiveness score.

3. The method according to claim 2, characterized in that, The step of adjusting the initial credibility of evidence units from different sources in the project semantic knowledge network based on the type of the fund expenditure deviation and the stage of the project includes: Identify evidence pairs in the project's semantic knowledge network that have mutual references or dependencies; For the evidence unit pair, determine the influence factor of the reference or dependency type on the degree of influence on credibility; The initial credibility of the cited or dependent evidence unit is adjusted based on the current credibility of the cited or dependent evidence unit and the influence factor.

4. The method according to claim 3, characterized in that, The step of determining the influence factor of the reference or dependency type on the credibility of the evidence unit pair includes: Obtain the business attribution information of the cited evidence unit and the cited evidence unit; Based on the business attribution information, query the business association rule table to obtain the impact factor adjustment coefficient; The final impact factor is obtained by multiplying the base impact factor of the reference or dependency type with the adjustment coefficient.

5. The method according to claim 2, characterized in that, The step of adjusting the connection strength of different relation types in the project semantic knowledge network according to the type of the fund expenditure deviation and the stage of the project includes: Obtain the business affiliation information of the entities connected by the relationship, as well as the frequency with which the relationship has been accepted in similar past scenarios; Based on the entity’s own business affiliation information and the frequency with which the relationship has been accepted in similar past scenarios, calculate the influence coefficient of the relationship on the connection strength. The final connection strength is obtained by multiplying the basic connection strength of the relationship type with the influence coefficient.

6. The method according to claim 2, characterized in that, The step of accumulating the conflict intensity generated by the conflict information existing in the information path includes: Obtain the hierarchical identifier of each conflicting information source in the information path; Based on the hierarchical identifier, determine the hierarchical weight of each conflicting information source; For conflicting information sources that influence each other in the information path, identify their influence type and adjust the hierarchical weight of the affected conflicting information sources according to the influence type. The conflict intensity of each conflicting information source is calculated by multiplying it with its adjusted hierarchical weight. The product is summed to obtain the conflict intensity caused by conflicting information in the information path.

7. The method according to claim 6, characterized in that, The steps of identifying the influence types of conflicting information sources that influence each other in the information path, and adjusting the hierarchical weights of the affected conflicting information sources according to the influence types, include: Obtain the identity information of the entity that publishes the conflict information source, the historical credit record of the entity, and the criticality identifier of the business process associated with the conflict information source; Based on the identity information of the publishing entity, the historical credit records, and the criticality indicator of the business process, identify the impact type of the conflicting information source; Based on the identified types of impact, the degree of their influence on the hierarchical weight of the affected conflict information sources is quantified to obtain the weight adjustment range; Based on the aforementioned weight adjustment range, the hierarchical weights of the affected conflicting information sources are adjusted.

8. The method according to claim 7, characterized in that, The step of obtaining the historical credit records of the publishing entity includes: Identify the identity of the publishing entity in different business systems; According to the preset mapping rules, the identity identifiers in the different business systems are unified into a unique global identity identifier; For the global identity identifier, historical behavior data related to the publishing entity is periodically acquired from the multiple business systems according to a preset data collection strategy. The historical behavior data includes the publishing entity's operation records such as information publishing, approval, modification, and business process participation in each business system. The acquired historical behavior data is format-converted and content-parsed to extract key credit elements, which include the accuracy of information release, compliance of approval, efficiency of decision execution, and records of abnormal behavior. Based on the key credit elements and combined with a preset set of credit assessment rules, the historical behavior of the publishing entity is quantitatively assessed to generate a local credit score for the publishing entity in each business system. Based on the local credit scores of the publishing entity in each business system, and combined with the weight of each business system in the overall fiscal supervision, the local credit scores of the publishing entity are weighted and summarized to obtain the unified historical credit record of the publishing entity.

9. The method according to claim 6, characterized in that, The step of accumulating the conflict intensity generated by the conflict information existing in the information path includes: Obtain the hierarchical identifier of each conflicting information source in the information path, and determine the hierarchical weight of each conflicting information source based on the hierarchical identifier; Obtain the identity information of the publishing entity of each conflict information source, the organizational structure relationship between the publishing entities, and the frequency of historical collaborative behavior between the publishing entities; Based on the organizational structure and the frequency of historical collaborative behavior, identify whether there is collusion or conspiracy among the publishing entities; If collusion or conspiracy is identified, the conflict information sources of the collusion or conspiracy are deduplicated, and the conflict intensity of the conflict information sources is adjusted according to the nature of the collusion or conspiracy. The conflict intensity of each conflicting information source is calculated by multiplying it with its adjusted hierarchical weight. The product is summed to obtain the conflict intensity caused by conflicting information in the information path.

10. A big data-based fiscal supervision data processing system for processing fiscal supervision data, characterized in that: The system includes: The information acquisition module is used to acquire project information related to deviations in project fund expenditures from multiple internal information carriers; The content processing module is used to process the project information to identify key entities and logical relationships between them. The knowledge network construction module is used to construct a project semantic knowledge network that reflects the project context based on the key entities and the logical relationships. The reasonableness judgment module is used to explore information paths in the project semantic knowledge network that can explain the reasonableness of the deviation, starting from the deviation in fund expenditure; and to judge the reasonableness of the deviation in fund expenditure based on the strength of evidence reflected in the information paths. The report generation module is used to generate corresponding explanatory reports or risk warnings based on the reasonableness judgment results.

Images