Decision trace system based on semantic manifest structure and semantic compression loss evaluation

Abstract

The embodiment of the application provides a decision tracing system and method based on a semantic list structure and semantic compression loss evaluation, and a computer readable storage medium, and belongs to the technical field of artificial intelligence and data governance. The system comprises a semantic list construction module, a semantic compression loss evaluation module, a governance rule execution module, a decision tracing chain recording module, a governance feedback closed-loop control module and an entity life cycle management module. Through the above scheme, data semantic consistency management, semantic information loss evaluation and decision process tracing recording can be realized.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence and data governance technology, and in particular to a decision tracing system, method, and computer-readable storage medium based on semantic inventory structure and semantic compression loss assessment. Background Technology

[0002] With the rapid development of artificial intelligence technology and data-driven applications, more and more systems rely on data from different data sources for automated decision-making.

[0003] During data integration, different data sources often employ different data structures and semantic representations, easily leading to semantic inconsistencies within the system. For example, different systems may use different attribute names or relational structures for the same entity, resulting in semantic ambiguity or conflict during data integration and processing. Existing data governance technologies typically focus only on the uniformity of data structure, lacking a unified description at the semantic level, thus making it difficult to guarantee semantic consistency across system environments. During the transformation of data from its original structure to a semantic structure, information compression or semantic loss often occurs, and current technologies generally lack methods for quantifying this loss of semantic information. Furthermore, many systems only record the final decision result, lacking a complete record of the decision rule triggering process and execution path, making it difficult to effectively audit and trace the AI ​​decision-making process.

[0004] Therefore, how to unify the semantic description of data, assess the information loss in the semantic mapping process, and achieve traceable recording of the decision-making process has become one of the technical problems that need to be addressed in the field of artificial intelligence data governance.

[0005] It should be noted that the above content is not necessarily prior art, nor is it intended to limit the scope of patent protection of this application. Summary of the Invention

[0006] This application provides a decision tracing system, method, and computer-readable storage medium based on semantic list structure and semantic compression loss assessment to solve or alleviate one or more of the technical problems mentioned above.

[0007] One aspect of this application provides a decision tracing system based on a semantic inventory structure and semantic compression loss assessment. The system includes: a semantic inventory construction module, a semantic compression loss assessment module, a governance rule execution module, a decision tracing chain recording module, a governance feedback closed-loop control module, and an entity lifecycle management module; wherein: The semantic list construction module is used to: receive raw data from multiple data sources, perform semantic parsing on the raw data, and perform unified semantic structure modeling to form a semantic list structure; wherein, the semantic list structure includes a key-value structure, an object structure, a semantic graph structure, and a knowledge representation structure; The semantic compression loss evaluation module is used to: quantitatively evaluate the degree of semantic information preservation during the process of converting the original data from the original data structure to the semantic structure; and generate a semantic compression loss evaluation result based on the evaluation result; wherein the semantic compression loss evaluation result is used to characterize the degree of semantic information preservation or the degree of semantic difference. The governance rule execution module is used to: perform governance decision operations on the semantic entities in the semantic list structure based on the semantic compression loss evaluation results and / or other semantic consistency evaluation results and preset governance rules, so as to obtain governance decision results; The decision traceability chain recording module is used to: construct and record the decision node information in the process of executing the governance rules in a structured manner; and to perform association modeling based on the execution order relationship and / or logical association relationship between multiple decision nodes to generate a decision traceability chain structure; wherein, the decision traceability chain structure is used to represent the decision path and the association relationship between its nodes; The entity lifecycle management module is used to: perform lifecycle management on the semantic entity based on the governance decision results; The governance feedback closed-loop control module is used to: evaluate and analyze the governance decision results and generate feedback signals based on the decision tracing chain structure, semantic compression loss assessment results and historical governance decision data; and adjust the governance rules, decision paths or strategy parameters based on the feedback signals to form a closed-loop governance mechanism of "decision-evaluation-feedback-optimization". The decision tracing chain structure, the semantic list structure, and the semantic compression loss assessment results are interconnected, and this interconnection is used for path reconstruction, causal analysis, and audit verification in the governance decision-making process. Each module interacts with other modules through a data interface. The semantic list structure, semantic compression loss assessment results, decision tracing chain structure, and feedback signals are stored, transmitted, and processed in the system as structured data to achieve automated data processing and governance control in a computer environment. The feedback signals are used to drive adjustments to data flow relationships or reconstruction of processing paths between modules within the system.

[0008] Optionally, the semantic list construction module is further used to perform semantic parsing and structured modeling on the input data to generate a unified semantic list data structure; the semantic list construction module includes: a data parsing submodule, a semantic mapping submodule, and a semantic relationship construction submodule; wherein: The data parsing submodule is used to perform semantic parsing processing on the raw data from multiple data sources in order to extract semantic entities, attribute information and their relational structures. The semantic mapping submodule is used to perform semantic alignment and mapping modeling on the data after semantic parsing based on a predefined and / or dynamically generated semantic structure model, so as to generate a unified semantic description structure. The semantic relationship construction submodule is used to construct a network of relationships between semantic entities based on the semantic entities; and to generate a semantic inventory data structure through the network of relationships; wherein the network of relationships includes hierarchical relationships, reference relationships, dependency relationships, and semantic relationships; The semantic inventory data structure is used for unified semantic representation across data sources, subsequent semantic consistency assessment, and governance decision processing; the semantic inventory data structure includes data associations established based on entity identifiers, semantic mapping relationships, and / or decision node reference relationships.

[0009] Optionally, the semantic compression loss evaluation module is further used to quantitatively model and evaluate the degree of semantic information loss generated during the semantic mapping process, so as to determine the degree of semantic information preservation and / or the degree of semantic difference; the semantic compression loss evaluation module includes: an information integrity calculation submodule, an attribute preservation rate evaluation submodule, and a relation consistency evaluation submodule; wherein: The information integrity calculation submodule is used to calculate the degree of semantic information retention during the semantic mapping process; The attribute retention rate evaluation submodule is used to evaluate the retention ratio between the mapped semantic entity attributes and the original semantic entity attributes. The relation consistency evaluation submodule is used to evaluate the degree of consistency between the mapped semantic entity relation structure and the original semantic entity relation structure; The semantic compression loss evaluation results are also used as the decision-making basis for the governance rule execution module and as evaluation information nodes in the construction of the decision traceability chain structure.

[0010] Optionally, the governance rule execution module is further configured to perform rule matching, condition judgment, and governance decision-making operations based on the semantic inventory data structure and semantic compression loss evaluation results, so as to achieve automated semantic governance; the governance rule execution module includes: a rule matching submodule, a condition judgment submodule, and a decision execution submodule; wherein: The rule matching submodule is used to match preset governance rules based on semantic entities, attribute information and relational structure; The condition judgment submodule is used to determine the current semantic state based on preset rule triggering conditions in order to determine whether the governance execution conditions are met. The decision execution submodule is used to execute corresponding governance decision operations when the governance execution conditions are met; wherein the governance decision operations include: data tagging, status update, access control, process triggering or policy adjustment operations; The governance rule execution module is also used to analyze the semantic consistency status, semantic anomaly detection results, and semantic compression loss evaluation results, and output the decision-making process as traceable decision node information to realize dynamic governance decisions for semantic entities.

[0011] Optionally, the decision tracing chain recording module is further used for: structurally constructing and recording decision node information during the execution of governance rules; performing association modeling based on the execution order relationship and / or logical association relationship between multiple decision nodes to generate a decision tracing chain structure, so as to realize path tracing and analysis of the decision process; the decision tracing chain recording module includes: a decision path recording submodule, an audit log submodule, and an association modeling submodule; wherein: The decision path recording submodule is used to construct and record the information of each decision node and its execution order during the execution of the governance rules; The audit log submodule is used to record the execution behavior, execution results and audit information of governance operations; The association modeling submodule is used to establish node association relationships based on the temporal sequence relationship, dependency relationship and / or causal relationship between the decision nodes, so as to form a decision tracing chain structure; the decision tracing chain structure is used for decision path reconstruction, cross-node causal relationship analysis and / or unified tracing of cross-system decision behavior.

[0012] Optionally, the governance feedback closed-loop control module is further configured to form a closed-loop optimization mechanism based on governance decision results during system operation, thereby enabling continuous evaluation, anomaly identification, and strategy optimization of the governance decision-making process. This allows the system to possess closed-loop governance capabilities based on historical decision results, including adaptive adjustment, dynamic strategy updates, and continuous optimization. The governance feedback closed-loop control module includes: a feedback signal generation submodule, a feedback analysis submodule, a strategy adjustment submodule, and a closed-loop control submodule. The feedback signal generation submodule is used to generate structured feedback signals based on the decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data. The feedback analysis submodule is used to analyze the feedback signals to identify abnormal patterns, consistency deviations, or optimization opportunities in the governance decision-making process. The strategy adjustment submodule is used to update or optimize governance rules, strategy parameters, or decision execution paths based on feedback analysis results. The closed-loop control submodule is used to apply the optimized governance strategy back to the subsequent governance decision-making process, thereby forming a closed-loop governance mechanism of "decision-evaluation-feedback-optimization". The feedback signal includes structured control parameters generated based on statistical analysis results, model output results, or rule evaluation results.

[0013] Optionally, the entity lifecycle management module is further used to perform structured management and dynamic control of the semantic entity's full lifecycle state in the system, so as to realize lifecycle scheduling based on governance decisions; the entity lifecycle management module includes: an entity creation submodule, a state update submodule, an entity disposal submodule, and a lifecycle control submodule; wherein: The entity creation submodule is used to create the semantic entity and initialize the lifecycle state of the semantic entity; wherein, the initialization includes: configuring entity attributes based on contextual semantic relationships; The state update submodule is used to update the lifecycle state of semantic entities based on governance decision results and semantic state changes; the update includes state transition, risk level adjustment and / or governance tag change; The entity processing submodule is used to perform processing operations on semantic entities that meet preset conditions; the processing operations include archiving, freezing, isolation, deletion and / or access restriction; The lifecycle control submodule is used to schedule and control the lifecycle flow of semantic entities based on governance rules and system status. The lifecycle state is associated with the decision node information in the decision tracing chain structure, and the association is used for tracing and analyzing the semantic entity evolution process.

[0014] Optionally, the system further includes: an extended governance function module for extending and enhancing the capabilities of the semantic governance process; the extended governance function module includes: an entity version management module, a governance report generation module, a rule version management module, a rule generation and optimization module, a semantic consistency monitoring module, and a decision tracking and analysis module; wherein: The entity version management module is used to record and manage the version information of semantic entities in order to realize version tracking and change analysis of semantic data. The governance report generation module is used to generate a structured governance report based on the execution results of the governance rules and the decision traceability chain structure. The rule version management module is used to control and update the governance rules, so as to realize the switching and comparative analysis between different rule versions; The rule generation and optimization module is used to optimize the governance rules based on historical decision data and external strategies. The semantic consistency monitoring module is used to continuously monitor the semantic consistency status in the semantic list structure. The decision tracking and analysis module is used to query, analyze, and reconstruct the decision path based on the decision tracing chain structure.

[0015] Optionally, the semantic compression loss evaluation module is further used to quantitatively model the semantic compression loss value based on the degree of preservation of semantic information or the degree of semantic difference during the semantic mapping process. The semantic compression loss value is calculated based on multiple evaluation metrics and modeling the differences between the original data structure and the semantic structure. It is used to participate in the construction of evaluation nodes in the governance decision-making and decision tracing chain structure. The evaluation metrics include information integrity, attribute retention rate, relationship consistency and semantic structure similarity.

[0016] Optionally, the semantic compression loss evaluation module is further configured to: calculate the semantic compression loss value; trigger corresponding governance rules or audit processes based on a preset semantic compression loss threshold and / or loss penalty parameter; the semantic compression loss value is determined based on a preset threshold, a dynamically adjusted threshold, or an adaptive learning threshold.

[0017] Optionally, the system further includes a verification module, which is used to perform consistency verification and / or integrity verification on the decision node information and the decision traceability chain structure.

[0018] Optionally, the governance rule execution module is further configured to: perform governance control operations on the data input, model inference results, and / or automated decision-making results in the artificial intelligence system based on the semantic compression loss evaluation results; wherein the governance control operations include quality control of model input data, risk verification of model output results, and / or intervention control of automated decision-making behavior.

[0019] Optionally, the semantic list structure may further include: structured semantic objects, hierarchical structures, or other data structures that can express semantic entities and their relationships.

[0020] Optionally, the system is further configured to: Semantic representations are obtained by performing semantic layer processing on the data; Based on the differences or information changes between the semantic representation and the original data, the semantic consistency state is determined. Automated decision control or governance operations are performed based on the semantic consistency state; wherein, the semantic consistency state is determined based on the semantic compression loss evaluation result; The decision control process is recorded to form a traceable decision traceability chain structure; The decision traceability chain structure is used for cross-system sharing, cross-system traceability, or cross-system verification to achieve collaborative decision-making and auditing between different systems.

[0021] Another aspect of this application provides a decision tracing method based on semantic inventory structure and semantic compression loss evaluation, the method comprising: Semantic parsing and structured modeling are performed on data from external data sources to generate a unified semantic inventory structure; Based on the semantic list structure, the semantic information loss of data in the semantic mapping process is calculated and quantified to obtain the semantic compression loss evaluation result; Based on the semantic compression loss evaluation results and preset governance rules, automated governance decision-making operations are performed on semantic entities. The decision-making node information in the governance decision-making process is structured and associated with modeling to generate a decision traceability chain structure; Based on the aforementioned decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data, feedback analysis is performed on the governance decision results and feedback signals are generated. Based on the feedback signals, the governance rules, strategy parameters, or decision execution paths are updated or optimized to form a closed-loop governance mechanism. Based on the updated governance strategy, the lifecycle state of semantic entities is updated and managed.

[0022] Another aspect of this application provides an artificial intelligence semantic governance and decision tracing method based on semantic inventory structure and semantic compression loss evaluation, the method comprising: Semantic parsing and structured modeling are performed on data from external data sources to generate a unified semantic list structure; wherein, the semantic list structure includes: semantic object structure, semantic graph structure, knowledge representation structure or other data structures used to express semantic entities and their relationships; Based on the semantic list structure, the semantic information loss of data in the semantic mapping process is calculated and quantified to obtain the semantic compression loss evaluation result. Based on the semantic compression loss assessment results and the preset governance rules, governance decision operations are performed on semantic entities; the governance decision operations include: data correction, data tagging, access control and / or policy adjustment; The decision-making node information in the governance decision-making process is structured and associated with modeling to generate a decision traceability chain structure; the decision traceability chain structure includes: chain structure, graph structure, event flow structure or other data structures used to represent the relationship between decision nodes; Based on the aforementioned decision tracing chain structure, the lifecycle state of semantic entities is updated and managed.

[0023] Optionally, the method further includes: Perform semantic parsing on the raw data; Establish semantic mapping relationships; The original data is structured and modeled to generate semantic entities and their relational structures. The semantic entities and their relational structures are used for semantic compression loss assessment and governance decision modeling.

[0024] Optionally, the method further includes: The triggering conditions, execution paths, and execution results of governance rules are structurally constructed and associated models are built to generate a decision traceability chain structure.

[0025] Optionally, the method further includes: Based on the semantic compression loss assessment results and / or semantic consistency assessment results, trigger the corresponding governance rules and / or audit processes; The governance rules are used to perform governance operations based on semantic consistency status, semantic anomaly detection results, or preset strategies.

[0026] Another aspect of this application provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the method described above.

[0027] The embodiments of this application employing the above-described technical solution may include the following beneficial effects: 1. It provides a unified semantic description mechanism to improve data semantic consistency: By constructing a semantic list structure, it provides unified semantic description and standardized expression for data from different data sources, thereby solving the problem of semantic inconsistency between different data systems and improving the semantic consistency and interpretability of data in artificial intelligence systems.

[0028] 2. A semantic compression loss assessment mechanism is proposed to quantify semantic information loss: By quantifying the information loss of data in the semantic mapping process, semantic missing or semantic deviation generated in the semantic conversion process can be identified, thereby improving the accuracy and reliability of data semantic mapping.

[0029] 3. Automated governance rule execution is achieved, improving system governance efficiency: Through the governance rule execution module, automated rule matching and condition judgment based on semantic list structure are realized, enabling the system to automatically perform data governance operations, thereby reducing manual intervention and improving the data governance efficiency of the artificial intelligence system.

[0030] 4. Constructing a decision traceability chain structure improves the transparency of AI system decision-making: The decision traceability chain recording module records decision node information during the execution of governance rules and generates a complete decision traceability chain structure, thereby enabling traceable recording of the AI ​​system's decision-making process and improving the system's decision transparency and auditing capabilities.

[0031] 5. An entity lifecycle management mechanism is introduced to achieve full lifecycle governance of semantic entities: The entity lifecycle management module manages the creation, updating, governance, and disposal processes of semantic entities, enabling the system to control semantic entities throughout their entire lifecycle, thereby improving the system's data management capabilities.

[0032] 6. Modular design facilitates system expansion and maintenance: Modular design, including semantic inventory construction, semantic compression loss assessment, governance rule execution, decision tracing chain recording, and entity lifecycle management, facilitates system function expansion, deployment adjustment, and operation and maintenance. Attached Figure Description

[0033] The accompanying drawings exemplify embodiments and form part of the specification, serving together with the textual description to explain exemplary implementations of the embodiments. The illustrated embodiments are for illustrative purposes only and do not limit the scope of the claims. Throughout the drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0034] Figure 1 This illustration schematically shows the overall architecture of a decision tracing system based on semantic list structure and semantic compression loss evaluation according to Embodiment 1 of this application; Figure 2 This illustration schematically shows a flowchart of semantic compression loss modeling and governance signal generation for a decision tracing system based on semantic list structure and semantic compression loss evaluation according to Embodiment 1 of this application; Figure 3 This illustration schematically shows a closed-loop flowchart of governance decision execution and feedback for a decision tracing system based on semantic list structure and semantic compression loss assessment according to Embodiment 1 of this application. Figure 4 This illustration schematically shows a system deployment structure diagram of a decision tracing system based on semantic list structure and semantic compression loss evaluation according to Embodiment 1 of this application; Figure 5 This illustration shows a schematic diagram of the semantic list data structure of a decision tracing system based on semantic list structure and semantic compression loss evaluation according to Embodiment 1 of this application; Figure 6This illustration shows a schematic diagram of the decision tracing data structure of a decision tracing system based on semantic list structure and semantic compression loss evaluation according to Embodiment 1 of this application; Figure 7 The illustration shows the semantic entity lifecycle data structure and state transition diagram of the semantic list structure and semantic compression loss assessment decision tracing system according to Embodiment 1 of this application. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with 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 in this application without inventive effort are within the scope of protection of this application.

[0036] It should be noted that the descriptions involving "first," "second," etc., in the embodiments of this application are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0037] It should be noted that, in any stage of this application involving the collection, storage, use, transmission, and processing of data, each stage strictly adheres to the laws, regulations, industry standards, and regulatory requirements of the data source, usage location, and relevant countries and regions to ensure the legality and compliance of data activities. In the collection stage, the purpose, method, and scope of collection are clearly communicated to the data subject in a prominent manner. Collection is conducted only after obtaining the data subject's legal authorization, ensuring that the collection process follows the "minimum necessary" principle and does not exceed the scope of data collection. In the storage stage, storage periods are limited, and data is promptly deleted or anonymized / encrypted after the storage purpose is achieved. In the usage stage, a strict data security protection mechanism is implemented, using field-level desensitization technology and processing the original data according to preset desensitization rules. For different types of data, multiple desensitization strategies, such as data generalization, data anonymization, and data encryption, are employed to effectively mitigate the risk of sensitive information leakage and ensure that all data used is securely processed and desensitized, comprehensively protecting the rights and interests of data subjects and data security. In the transmission and processing stages, the confidentiality and security of data are ensured during transmission and processing.

[0038] In the description of this application, it should be understood that the numerical labels before the steps do not indicate the order of the steps, but are only used to facilitate the description of this application and to distinguish each step, and therefore should not be construed as a limitation of this application.

[0039] First, a definition of the terminology used in this application is provided: Semantic inventory: refers to a unified semantic description structure formed after semantic parsing of data from multiple data sources, including semantic entities, entity attributes, and the relationship structure between entities.

[0040] Semantic compression loss refers to the degree of information loss caused by data abstraction, structural transformation, or semantic mapping during the transformation of data from its original data structure to a semantic structure. This semantic compression loss can be obtained by quantifying the differences between the original semantic features and the mapped semantic features. These differences include at least one attribute difference, relational structure difference, and semantic coverage difference.

[0041] Governance rules: refer to the set of rules preset or dynamically generated in the system for data governance, which are used to perform governance operations such as data verification, data correction or data tagging on semantic entities.

[0042] Decision traceability chain structure: refers to the decision traceability chain structure used to build and record decision node information in the process of governance rule execution and to associate them according to the execution order, so as to realize the traceability and auditing of the decision process.

[0043] Entity lifecycle management: refers to the mechanism for managing the creation, updating, governance, and disposal of semantic entities in a system.

[0044] Semantic consistency refers to the degree of semantic consistency between different data representations, and is used to measure the semantic preservation of data during semantic mapping or transformation.

[0045] Governance strategy: refers to a set of strategies used to guide the generation, selection or execution of governance rules, which can be determined based on preset logic, historical data or dynamic analysis results.

[0046] Governance feedback loop: refers to a mechanism for continuous evaluation, feedback analysis and strategy optimization of the governance decision-making process based on governance decision results, decision traceability chain structure and semantic compression loss assessment results.

[0047] Example 1 like Figure 1 As shown, this embodiment provides a decision tracing system based on a semantic inventory structure and semantic compression loss assessment, including: a semantic inventory construction module, a semantic compression loss assessment module, a governance rule execution module, a decision tracing chain recording module, a governance feedback closed-loop control module, and an entity lifecycle management module; wherein: The semantic list construction module is used to: receive raw data from multiple data sources, perform semantic parsing on the raw data, and perform unified semantic structure modeling to form a semantic list structure; wherein, the semantic list structure includes a key-value structure, an object structure, a semantic graph structure, and a knowledge representation structure; The semantic compression loss evaluation module is used to: quantitatively evaluate the degree of semantic information preservation during the process of converting the original data from the original data structure to the semantic structure; and generate a semantic compression loss evaluation result based on the evaluation result; wherein the semantic compression loss evaluation result is used to characterize the degree of semantic information preservation or the degree of semantic difference. The governance rule execution module is used to: perform governance decision operations on the semantic entities in the semantic list structure based on the semantic compression loss evaluation results and / or other semantic consistency evaluation results and preset governance rules, so as to obtain governance decision results; The decision traceability chain recording module is used to: construct and record the decision node information in the process of executing the governance rules in a structured manner; and to perform association modeling based on the execution order relationship and / or logical association relationship between multiple decision nodes to generate a decision traceability chain structure; wherein, the decision traceability chain structure is used to represent the decision path and the association relationship between its nodes; The entity lifecycle management module is used to: perform lifecycle management on the semantic entity based on the governance decision results; The governance feedback closed-loop control module is used to: evaluate and analyze the governance decision results and generate feedback signals based on the decision tracing chain structure, semantic compression loss assessment results and historical governance decision data; and adjust the governance rules, decision paths or strategy parameters based on the feedback signals to form a closed-loop governance mechanism of "decision-evaluation-feedback-optimization". The decision tracing chain structure, the semantic list structure, and the semantic compression loss assessment results are interconnected, and this interconnection is used for path reconstruction, causal analysis, and audit verification in the governance decision-making process. Each module interacts with other modules through a data interface. The semantic list structure, semantic compression loss assessment results, decision tracing chain structure, and feedback signals are stored, transmitted, and processed in the system as structured data to achieve automated data processing and governance control in a computer environment. The feedback signals are used to drive adjustments to data flow relationships or reconstruction of processing paths between modules within the system.

[0048] In some embodiments, such as Figure 5As shown, the semantic list construction module can also receive raw data from multiple data sources and perform semantic parsing and unified modeling on the data to generate a unified semantic list structure. The semantic list structure can include data units such as semantic entities, semantic attributes, and semantic relationships. This module can convert data from different systems into a unified semantic expression. The semantic compression loss evaluation module is used to evaluate the degree of compression loss of semantic information during the process of converting data from its original data structure to a semantic structure, thereby identifying potential information loss or semantic bias during semantic mapping. The governance rule execution module is used to perform automated governance operations on the semantic entities in the semantic list structure based on the semantic compression loss evaluation results and preset data governance rules, such as data consistency verification, data integrity detection, or data quality assessment. The semantic compression loss evaluation results can be used to participate in the construction of the decision path structure and serve as input for feedback analysis and strategy optimization by the governance feedback closed-loop control module. The semantic compression loss evaluation results participate in the construction and optimization feedback of the decision path structure. The decision traceability chain recording module is used to construct and record decision node information during the governance rule execution process, and associate multiple decision nodes according to their execution order to form a decision traceability chain structure, thereby achieving a complete record of the system governance decision-making process. The entity lifecycle management module is used to manage the lifecycle of semantic entities in the system, including their creation, updating, review, and disposal, based on governance decisions.

[0049] In some embodiments, within a data governance platform application scenario, the system can acquire data from multiple data sources, such as enterprise databases, external API interfaces, or machine learning model outputs. The system first performs semantic parsing on data from different data sources using a semantic inventory construction module, constructing a unified semantic inventory structure to ensure semantic consistency across different sources. Subsequently, a semantic compression loss assessment module evaluates the degree of information loss during the semantic mapping process to identify potential semantic distortion issues. After completing the semantic assessment, a governance rule execution module performs automated governance operations on the data in the semantic inventory structure according to preset data governance rules. During the governance process, a decision traceability chain recording module records the triggering conditions, execution paths, and execution results for each governance rule execution, thereby generating a complete decision traceability chain structure. Finally, an entity lifecycle management module updates the lifecycle status of semantic entities based on the governance results, such as creation, update, review, or obsolescence. Through this process, the system can output governance results and corresponding audit records, thereby achieving transparency and traceability management of the data governance process.

[0050] In some embodiments, after outputting governance results and corresponding audit records, the system can continuously optimize the governance process through a governance feedback closed-loop control module. Specifically, the governance feedback closed-loop control module generates feedback signals based on the decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data, and analyzes these feedback signals to identify abnormal patterns, consistency deviations, or potential optimization opportunities in the governance process. Based on the feedback analysis results, the system can dynamically adjust or optimize the semantic inventory construction process, governance rule execution strategies, and decision tracing chain generation paths, thereby continuously improving the consistency of semantic modeling, the accuracy of rule execution, and the rationality of decision paths in subsequent data governance processes, forming a closed-loop governance mechanism of "governance execution—result evaluation—feedback analysis—strategy optimization".

[0051] In this embodiment, a semantic inventory structure is constructed to perform unified semantic modeling on multi-source heterogeneous data. The degree of information preservation during semantic transformation is quantified through semantic compression loss assessment. This is then combined with a governance rule execution module for decision processing, and a decision traceability chain recording module to structurally record the entire process. Furthermore, an entity lifecycle management module provides continuous control, achieving a closed-loop process from data semantic modeling, semantic quality assessment, governance decision execution, decision process traceability, and entity lifecycle control. While ensuring semantic consistency and information integrity, this approach helps improve the interpretability, verifiability, and traceability of the governance decision-making process, thereby enhancing the reliability and auditability of data governance.

[0052] In an optional embodiment, the semantic list construction module is further used to perform semantic parsing and structured modeling on the input data to generate a unified semantic list data structure; the semantic list construction module includes: a data parsing submodule, a semantic mapping submodule, and a semantic relationship construction submodule; wherein: The data parsing submodule is used to perform semantic parsing processing on the raw data from multiple data sources in order to extract semantic entities, attribute information and their relational structures. The semantic mapping submodule is used to perform semantic alignment and mapping modeling on the data after semantic parsing based on a predefined and / or dynamically generated semantic structure model, so as to generate a unified semantic description structure. The semantic relationship construction submodule is used to construct a network of relationships between semantic entities based on the semantic entities; and to generate a semantic inventory data structure through the network of relationships; wherein the network of relationships includes hierarchical relationships, reference relationships, dependency relationships, and semantic relationships; The semantic inventory data structure is used for unified semantic representation across data sources, subsequent semantic consistency assessment, and governance decision processing; the semantic inventory data structure includes data associations established based on entity identifiers, semantic mapping relationships, and / or decision node reference relationships.

[0053] In this embodiment, a data parsing submodule extracts semantic entities and their attributes and relationships, a semantic mapping submodule achieves cross-data source semantic alignment, and a semantic relationship construction submodule establishes a multi-type association network between entities, thus realizing a unified semantic expression and structured organization of multi-source data. This breaks down data silos and semantic inconsistencies, enabling data from different sources to be integrated and interact within a unified semantic framework, thereby providing a stable and standardized data foundation for subsequent semantic consistency assessment and governance decisions.

[0054] In an optional embodiment, the semantic compression loss evaluation module is further used to quantitatively model and evaluate the degree of semantic information loss generated by the data during the semantic mapping process, so as to determine the degree of semantic information preservation and / or the degree of semantic difference; the semantic compression loss evaluation module includes: an information integrity calculation submodule, an attribute preservation rate evaluation submodule, and a relation consistency evaluation submodule; wherein: The information integrity calculation submodule is used to calculate the degree of semantic information retention during the semantic mapping process; The attribute retention rate evaluation submodule is used to evaluate the retention ratio between the mapped semantic entity attributes and the original semantic entity attributes. The relation consistency evaluation submodule is used to evaluate the degree of consistency between the mapped semantic entity relation structure and the original semantic entity relation structure; The semantic compression loss evaluation results are also used as the decision-making basis for the governance rule execution module and as evaluation information nodes in the construction of the decision traceability chain structure.

[0055] In some embodiments, such as Figure 2 As shown, the semantic compression loss evaluation process in this embodiment may include the following steps: Step S201: Raw data input. The system receives raw data from an external data source, which may include an enterprise database, an external API interface, or machine learning model output data.

[0056] Step S202: Extraction of raw semantic features. The system performs semantic feature extraction processing on the raw data to identify semantic entities, attribute information, and the relationship structure between entities in the raw data, thereby forming a set of raw semantic features.

[0057] Step S203: Semantic Inventory Mapping. The system performs semantic mapping on the original data according to a predefined semantic inventory structure, thereby generating semantic inventory data with a unified semantic description structure. The semantic inventory structure can be as follows: Figure 5 The diagram shows a structured representation.

[0058] Step S204: Semantic Feature Extraction After Mapping. The system extracts the mapped semantic feature information from the generated semantic inventory data to form a set of mapped semantic features.

[0059] Step S205: Semantic difference comparison. The system compares the original semantic feature set with the mapped semantic feature set to identify semantic changes or deviations that occur during the semantic mapping process.

[0060] Step S206: Semantic Compression Loss Calculation. The system calculates the semantic compression loss value based on the semantic difference comparison results. This value can be determined based on multiple semantic preservation metrics, such as attribute retention rate, relation consistency, and semantic completeness. Attribute retention rate represents the ratio between the number of valid attributes retained after semantic mapping and the original number of attributes. Relation consistency represents the degree of consistency between the mapped relationship structure and the original relationship structure. Semantic completeness represents the quantitative result of the degree to which the mapped semantic expression covers the original semantic information. In some embodiments, the semantic compression loss value can be obtained by weighting the above metrics, where the weights of each metric can be adjusted according to different data governance scenarios. When the semantic compression loss value exceeds a preset threshold, the system can trigger corresponding governance rules, such as marking semantic anomalies, initiating a manual audit process, and re-executing the semantic mapping.

[0061] Step S207: Generate semantic compression loss evaluation results. The system generates semantic compression loss evaluation results based on the calculated semantic compression loss values.

[0062] Step S208: Send to the governance rule execution module. The system sends the semantic compression loss evaluation result to the governance rule execution module for use in subsequent governance rule evaluation and governance decisions.

[0063] Step S209: Participate in the construction and optimization feedback of the decision tracing chain structure. The system participates in the construction and optimization adjustment of the decision tracing chain structure based on the semantic compression loss evaluation results. In some embodiments, the semantic compression loss evaluation results can further participate in the construction process of the decision tracing chain structure, optimizing the relationship between decision nodes to achieve optimization feedback on the decision path structure. This optimization feedback includes adjusting the weights, association strengths, or path selections of decision nodes based on the semantic compression loss evaluation results to improve the semantic consistency and interpretability of the decision path.

[0064] In some implementations, the semantic compression loss assessment results can also be used as input to the governance feedback closed-loop control module to generate feedback signals and perform feedback analysis, so as to achieve dynamic optimization of governance rules, strategy parameters or decision paths, thereby participating in the formation of a governance closed loop.

[0065] Through the above process, the system can quantitatively evaluate the degree of information compression generated during semantic mapping, thereby identifying potential semantic information loss problems and improving the data semantic consistency of the artificial intelligence system.

[0066] In some implementations, the semantic compression loss assessment results can also be used in the construction and optimization feedback of the decision tracing chain structure.

[0067] In this embodiment, a semantic compression loss assessment mechanism is used to quantitatively evaluate the information loss generated during the semantic mapping process, thereby improving semantic consistency and data reliability in the data governance process. This embodiment assesses the retention of semantic information through an information integrity calculation submodule, measures the mapping accuracy at the attribute level through an attribute retention rate assessment submodule, and judges the consistency of semantic relationship structures through a relation consistency assessment submodule, achieving multi-dimensional quantitative analysis of information loss during semantic mapping. Building upon traditional coarse-grained data quality assessment, a fine-grained semantic layer assessment mechanism is introduced, enabling the system to accurately identify the sources of semantic bias and provide quantifiable and comparable evidence for governance decisions.

[0068] In an optional embodiment, the governance rule execution module is further configured to perform rule matching, condition judgment, and governance decision operations based on the semantic inventory data structure and semantic compression loss evaluation results, so as to achieve automated semantic governance; the governance rule execution module includes: a rule matching submodule, a condition judgment submodule, and a decision execution submodule; wherein: The rule matching submodule is used to match preset governance rules based on semantic entities, attribute information and relational structure; The condition judgment submodule is used to determine the current semantic state based on preset rule triggering conditions in order to determine whether the governance execution conditions are met. The decision execution submodule is used to execute corresponding governance decision operations when the governance execution conditions are met; wherein the governance decision operations include: data tagging, status update, access control, process triggering or policy adjustment operations; The governance rule execution module is also used to analyze the semantic consistency status, semantic anomaly detection results, and semantic compression loss evaluation results, and output the decision-making process as traceable decision node information to realize dynamic governance decisions for semantic entities.

[0069] In some embodiments, such as Figure 3 As shown, the governance rule enforcement and decision traceability process may include the following steps: Step S301: Receive semantic manifest data. The system receives semantic manifest data that has undergone semantic standardization from the semantic manifest construction module.

[0070] Step S302: Read the governance rule base. The system reads the governance rules related to the current semantic entity from the governance rule base.

[0071] Step S303: Rule Matching. The system matches the entity attributes in the semantic manifest structure with the matching conditions in the governance rules to identify the applicable governance rules.

[0072] Step S304: Condition Judgment. The system judges the matching results based on the triggering conditions in the governance rules to determine whether a governance operation needs to be performed.

[0073] Step S305: Perform governance operations. When the rule triggering conditions are met, the system performs the corresponding governance operations, such as data correction, permission restriction, or anomaly marking. When the rule triggering conditions are not met, the system maintains the current data state and no governance operations are required.

[0074] Step S306: Record the nodes of the decision traceability chain. During each governance rule execution, the system records the corresponding decision node information, including rule identifier, triggering conditions, input semantic data summary, execution operation type, and timestamp information. The decision nodes are associated with each other through temporal order or logical dependency.

[0075] Step S307: Generate the decision traceability chain structure. The system connects multiple decision nodes according to the execution order to generate a complete decision traceability chain structure.

[0076] Step S308: Update the entity lifecycle state. The lifecycle state of relevant semantic entities can be updated based on the governance operation results, and the changes in lifecycle state are used as part of the feedback information for analysis and optimization by the subsequent governance feedback closed-loop control module. The lifecycle state management structure and its state transition relationships can be as follows: Figure 7 As shown.

[0077] Step S309: Output governance results and audit records. The system outputs the governance results and the corresponding decision traceability chain structure to the system interface for querying and auditing by system administrators or regulatory agencies. In some embodiments, each decision traceability chain structure node may include the following information: decision node identifier, trigger rule identifier, input semantic data summary, condition judgment result, execution operation type, and timestamp information. In some implementations, such as Figure 6As shown, the decision-making nodes in the decision tracing chain structure are associated through time sequence or logical dependency.

[0078] In some embodiments, after outputting governance results and audit records, the governance decision-making process can be continuously optimized through a governance feedback closed-loop control module. Specifically, the governance feedback closed-loop control module generates feedback signals based on the decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data, and analyzes these feedback signals to identify abnormal patterns, consistency deviations, or potential optimization opportunities in the governance process. Based on the feedback analysis results, the system can dynamically adjust or optimize the governance rule matching strategy, condition judgment logic, and decision tracing chain generation path, and reapply the optimized strategy to subsequent governance decision-making processes through a closed-loop scheduling mechanism, thereby forming a closed-loop governance process of "rule execution—result evaluation—feedback analysis—strategy optimization".

[0079] By following the steps above, an auditable decision log structure can be formed, thereby achieving a complete record of the decision-making process of the artificial intelligence system and improving the interpretability and auditability of the system.

[0080] In this embodiment, the traceable recording of the AI ​​decision-making process is achieved through a governance rule execution module and a decision traceability chain recording module, thereby supporting AI decision auditing application scenarios. This embodiment uses a rule matching submodule to perform rule matching on semantic entities and their attribute relationships, a condition judgment submodule to determine whether execution conditions are met, and a decision execution submodule to implement specific governance operations, realizing an automated rule-driven governance process based on semantic state. While reducing manual intervention, it enables the governance process to be dynamically triggered and executed according to semantic state, thereby improving governance efficiency, consistency, and the controllability of rule execution.

[0081] In optional embodiments, such as Figure 4As shown, the system deployment structure in this embodiment includes: an external data source, a data access layer, a semantic processing and governance service layer, a decision tracing chain structure service layer, a data storage layer, and an audit interface layer. The external data source provides raw data input to the system. In some embodiments, the data source may include: an enterprise database system, an external API interface, artificial intelligence model output data, and a data stream processing system. The system obtains data from the aforementioned data sources through the data access layer. The data access layer can receive data input from different data sources and perform unified access processing on the data. In some embodiments, the data access layer can be responsible for sending the received raw data to the semantic governance service layer for processing. The semantic governance service layer is the core computing layer of the system and may include a semantic inventory construction module, a semantic compression loss evaluation module, and a governance rule execution module. In some embodiments, this service layer may be deployed using a microservice architecture, with each module running as an independent service, such as a semantic inventory service, a semantic compression loss evaluation service, and a governance decision engine service. These services can communicate through internal APIs to achieve collaborative work between modules within the system. The decision tracing chain structure service layer may include a decision tracing chain recording module and an entity lifecycle management module. This service layer is used to construct and record governance decision information during system execution and maintain the lifecycle state of semantic entities. In some embodiments, the decision traceability chain structure can be stored using a chain-like data structure to ensure the complete recording of the governance decision path. In some embodiments, the system may also include a governance feedback closed-loop control mechanism, which can be deployed in the semantic governance service layer or the decision traceability chain service layer, or exist as an independent service. The governance feedback closed-loop control mechanism is used to generate feedback signals based on governance results, decision traceability chain structure, and semantic compression loss evaluation results, and to analyze and process the feedback signals to achieve dynamic optimization of governance rule execution strategies, decision path structures, and related parameters, thereby forming a continuously evolving closed-loop governance system at the system level. In some embodiments, the data storage layer can be used to store various types of data generated during system operation. The data storage layer may include a semantic manifest database, a governance rule database, a decision traceability chain structure log database, and an entity lifecycle database. This data can be stored in a relational database, a graph database, or a distributed storage system. The audit interface layer can be used to provide governance result query and audit functions to external systems. In some embodiments, the audit interface can support the following functions: decision traceability chain structure query, governance rule execution record query, and entity lifecycle state query. The audit interface can be accessed through an API or a graphical management interface.

[0082] In some embodiments, the system operation process may include the following steps: Step S1: The data access layer receives data from an external data source.

[0083] Step S2: The semantic list building module converts the raw data into a unified semantic list structure.

[0084] Step S3: The semantic compression loss evaluation module calculates the information loss during the semantic conversion process.

[0085] Step S4: The governance rule execution module executes automated governance decisions based on the semantic list and governance rules.

[0086] Step S5: The decision traceability chain recording module records the rule execution path and generates the decision traceability chain structure.

[0087] Step S6: The entity lifecycle management module updates the lifecycle status of the relevant semantic entities.

[0088] Step S7: The audit interface layer outputs the governance results and audit records.

[0089] In some embodiments, each module of the system can run on one or more computing nodes via network connection, thereby realizing the distributed deployment and scalability of the system.

[0090] In this embodiment, the AI ​​governance and decision tracing system based on semantic inventory structure and semantic compression loss assessment can be deployed on a cloud computing platform, distributed computing cluster, or enterprise data platform to support multi-data source access, semantic governance computation, and AI decision auditing functions.

[0091] In an optional embodiment, the decision tracing chain recording module is further configured to: structurally construct and record decision node information during the execution of governance rules; perform association modeling based on the execution order relationship and / or logical association relationship between multiple decision nodes to generate a decision tracing chain structure, so as to realize path tracing and analysis of the decision process; the decision tracing chain recording module includes: a decision path recording submodule, an audit log submodule, and an association modeling submodule; wherein: The decision path recording submodule is used to construct and record the information of each decision node and its execution order during the execution of the governance rules; The audit log submodule is used to record the execution behavior, execution results and audit information of governance operations; The association modeling submodule is used to establish node association relationships based on the temporal sequence relationship, dependency relationship and / or causal relationship between the decision nodes, so as to form a decision tracing chain structure; the decision tracing chain structure is used for decision path reconstruction, cross-node causal relationship analysis and / or unified tracing of cross-system decision behavior.

[0092] In some embodiments, such as Figure 6As shown, the decision traceability chain structure can be formed by connecting multiple decision nodes in execution order. Each decision node can include the following information: decision node identifier, trigger rule identifier, input semantic data summary, condition judgment result, execution operation type, execution result, and timestamp information. During system operation, when the governance rule execution module executes a governance rule, the system can generate a corresponding decision node and add it to the decision traceability chain structure. Multiple decision nodes are connected in execution order to form a complete decision traceability chain structure, which is used to construct and record the decision paths executed by the system during data governance.

[0093] In some embodiments, the decision traceability chain structure can be stored using a chained data structure, where each decision node can contain reference information pointing to the previous or next node, thereby forming a continuous decision traceability chain structure. This structure enables traceability of the decision-making process, queryability of the decision path, and auditability of the decision log.

[0094] In some embodiments, system administrators or regulatory bodies can query the decision tracing chain structure through an audit interface to obtain the governance rules executed by the system during data governance and their corresponding decision paths. In different implementations, the decision tracing chain structure can be stored in a relational database, a graph database, or a distributed log system.

[0095] In some implementations, decision nodes are connected through node associations, which include temporal order relationships, logical dependencies, or causal relationships. These associations may also include data association mechanisms based on entity identifiers, semantic mapping relationships, or decision node reference relationships, thereby achieving a unified association between the semantic inventory structure, the semantic compression loss assessment results, and the decision tracing chain structure.

[0096] In some implementations, the decision tracing chain structure can also serve as the data input basis for the governance feedback closed-loop control module, supporting the analysis, evaluation, and optimization feedback of historical decision paths.

[0097] In some implementations, the system may also include a verification module for verifying the consistency and integrity of decision node information and the decision traceability chain structure. This can be achieved through methods such as hash verification, chain structure verification, or node relationship verification to ensure that the decision traceability chain structure has not been tampered with. The verification module enables transparency, auditability, and consistency verifiability in the governance decision-making process of the artificial intelligence system.

[0098] In this embodiment, a decision traceability chain structure is constructed through a decision traceability chain recording module, enabling transparency, traceability, and auditability of the governance decision-making process in the artificial intelligence system, thereby improving the reliability and consistency verification capabilities of system governance. The decision path recording submodule records each decision node, the audit log submodule records execution behaviors and results, and the association modeling submodule establishes the temporal, dependency, and causal relationships between nodes, achieving structured chain modeling and recording of the governance process. This not only reconstructs the complete decision path but also enables cross-node causal analysis, providing a reliable basis for problem tracing, responsibility definition, and audit analysis.

[0099] In an optional embodiment, the governance feedback closed-loop control module is further configured to form a closed-loop optimization mechanism based on governance decision results during system operation, so as to achieve continuous evaluation, anomaly identification, and strategy optimization of the governance decision-making process, thereby enabling the system to have closed-loop governance capabilities based on historical decision results for adaptive adjustment, dynamic strategy updates, and continuous optimization; the governance feedback closed-loop control module includes: a feedback signal generation submodule, a feedback analysis submodule, a strategy adjustment submodule, and a closed-loop control submodule; wherein: The feedback signal generation submodule is used to generate structured feedback signals based on the decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data. The feedback analysis submodule is used to analyze the feedback signals to identify abnormal patterns, consistency deviations, or optimization opportunities in the governance decision-making process. The strategy adjustment submodule is used to update or optimize governance rules, strategy parameters, or decision execution paths based on feedback analysis results. The closed-loop control submodule is used to apply the optimized governance strategy back to the subsequent governance decision-making process, thereby forming a closed-loop governance mechanism of "decision-evaluation-feedback-optimization". The feedback signal includes structured control parameters generated based on statistical analysis results, model output results, or rule evaluation results.

[0100] In some embodiments, such as Figure 3As shown, the decision tracing system can further include a governance feedback closed-loop control module to continuously optimize the governance decision-making process, thereby forming a closed-loop governance mechanism. Specifically, during system operation, after the governance rule execution module completes a governance decision, the decision tracing chain recording module generates a corresponding decision tracing chain structure, while the semantic compression loss evaluation module outputs the semantic compression loss evaluation result. Based on the decision tracing chain structure, the semantic compression loss evaluation result, and historical governance decision data, the governance feedback closed-loop control module evaluates and analyzes the current governance decision result and generates feedback signals. The feedback signals may include: decision consistency evaluation results, semantic compression loss change trends, decision path anomaly detection results, and historical governance effect evaluation results. Based on the feedback signals, the governance feedback closed-loop control module adjusts or optimizes the governance rules, strategy parameters, or decision execution paths, and applies the optimized governance strategy to subsequent decision-making processes.

[0101] In some implementations, the governance feedback closed-loop control module can identify recurring abnormal patterns or high-loss semantic nodes in multiple consecutive decisions and generate targeted optimization strategies accordingly.

[0102] In this embodiment, the system forms a closed-loop governance process of "decision-evaluation-feedback-optimization", which transforms the artificial intelligence system from a single decision-making and execution mechanism into a dynamic governance system with continuous optimization capabilities, thereby improving decision consistency, system stability and long-term governance capabilities.

[0103] In an optional embodiment, the entity lifecycle management module is further used to perform structured management and dynamic control of the semantic entity's full lifecycle state in the system, so as to realize lifecycle scheduling based on governance decisions; the entity lifecycle management module includes: an entity creation submodule, a state update submodule, an entity disposal submodule, and a lifecycle control submodule; wherein: The entity creation submodule is used to create the semantic entity and initialize the lifecycle state of the semantic entity; wherein, the initialization includes: configuring entity attributes based on contextual semantic relationships; The state update submodule is used to update the lifecycle state of semantic entities based on governance decision results and semantic state changes; the update includes state transition, risk level adjustment and / or governance tag change; The entity processing submodule is used to perform processing operations on semantic entities that meet preset conditions; the processing operations include archiving, freezing, isolation, deletion and / or access restriction; The lifecycle control submodule is used to schedule and control the lifecycle flow of semantic entities based on governance rules and system status. The lifecycle state is associated with the decision node information in the decision tracing chain structure, and the association is used for tracing and analyzing the semantic entity evolution process.

[0104] In some embodiments, such as Figure 7 As shown, each semantic entity can contain the following basic information: entity identifier, entity type, version information, and lifecycle status. The system maintains the lifecycle status of semantic entities through the entity lifecycle management module.

[0105] In some embodiments, the lifecycle state of a semantic entity may include the following states: creation state, update state, review state, and obsolete state. When the system receives new semantic entity data, it can create an initial lifecycle state for that entity. During data governance, when a semantic entity undergoes attribute modification or semantic update, the system can update the entity state to the update state. After an entity has been verified by governance rules or manually reviewed, the system can update the entity state to the review state. When a semantic entity is no longer used or is determined to be invalid data, the system can update the entity state to the obsolete state.

[0106] In some embodiments, the system may also generate a corresponding lifecycle log record for each lifecycle state change. The lifecycle log may include the following information: entity identifier, lifecycle state change, type of operation performed, and timestamp information.

[0107] In some implementations, lifecycle state change records can also serve as one of the inputs to the governance feedback closed-loop control module, supporting the evaluation of semantic entity evolution processes and the optimization of governance strategies.

[0108] In some implementations, the lifecycle control submodule can control the lifecycle state transition path based on a preset state machine model or rule-based scheduling strategy, such as limiting the state transition order or triggering state transitions based on conditions. Through the above mechanisms, the system can achieve a complete record of the semantic entity lifecycle change process, thereby improving the system's data governance capabilities, state consistency control capabilities, and lifecycle-level auditability.

[0109] In some implementations, the entity lifecycle management module can also work in conjunction with the decision traceability chain recording module to record entity lifecycle change events into the decision traceability chain structure, thereby achieving unified auditing of the system governance decision-making process and the data lifecycle management process.

[0110] In this embodiment, semantic entities are initialized through an entity creation submodule, their states are dynamically updated through a state update submodule, and operations such as archiving or isolation are performed through an entity disposal submodule. The overall flow is scheduled through a lifecycle control submodule, achieving full-process management of semantic entities from creation to disposal. Governance decisions are directly applied to the evolution of entity states, making the changes in entities controllable and traceable, thereby improving the system's ability to manage risky data and abnormal states.

[0111] In an optional embodiment, the system further includes: an extended governance function module for extending and enhancing the capabilities of the semantic governance process; the extended governance function module includes: an entity version management module, a governance report generation module, a rule version management module, a rule generation and optimization module, a semantic consistency monitoring module, and a decision tracking and analysis module; wherein: The entity version management module is used to record and manage the version information of semantic entities in order to realize version tracking and change analysis of semantic data. The governance report generation module is used to generate a structured governance report based on the execution results of the governance rules and the decision traceability chain structure. The rule version management module is used to control and update the governance rules, so as to realize the switching and comparative analysis between different rule versions; The rule generation and optimization module is used to optimize the governance rules based on historical decision data and external strategies. The semantic consistency monitoring module is used to continuously monitor the semantic consistency status in the semantic list structure. The decision tracking and analysis module is used to query, analyze, and reconstruct the decision path based on the decision tracing chain structure.

[0112] In some embodiments, semantic entities can be used to represent data objects and include fields such as entity identifier, entity type, version information, and timestamp. Semantic attributes can be used to describe the attribute characteristics of semantic entities, including attribute name, attribute value, and data type. Semantic relationships can be used to represent the association between semantic entities, including relationship identifier, source entity, target entity, and relationship type, and optionally include relationship attributes to describe additional characteristic information of the relationship.

[0113] In some implementations, the semantic inventory data structure can support extended attribute definitions, including adding extended fields or extended structures to semantic entities, semantic relationships or semantic attributes, to support the semantic expression needs of different application scenarios and enhance the flexibility and scalability of semantic modeling.

[0114] In some embodiments, the entity version management module can also be used to parse and process semantic entities and manage entity version information. This module can perform structural parsing of semantic entities in the semantic manifest structure and establish a version identifier for each semantic entity. When a semantic entity is updated or a governance operation is performed, the system can generate new entity version information and record the version change history, thereby achieving version tracking and historical management of semantic entities. The governance report generation module can also be used to generate governance reports based on the execution results of governance rules. This module can obtain relevant data from the decision traceability chain recording module and the entity lifecycle management module, and generate governance reports according to preset report templates to display the system governance process and governance results. The rule version management module can also be used to perform version control and update management of governance rules. This module can establish version information for each governance rule and record the rule update history. When a governance rule is modified, the system can generate a new rule version and save the old version information, thereby supporting version backtracking and historical querying of governance rules. The version information includes at least one version identifier, version evolution relationship, and historical change records to support version tracing, version comparison, or version-based governance strategy execution. The rule generation and optimization module can also be used to generate new governance rules or optimize existing governance rules based on system operation results or governance analysis results. For example, when the system detects abnormal patterns in semantic data, it can generate new governance rules according to preset strategies and add them to the governance rule base. The semantic consistency monitoring module can also be used to continuously monitor semantic entities and their relationship structures in the semantic list structure. When the system detects semantic conflicts or inconsistencies between semantic entities, it can generate corresponding abnormal prompts and trigger corresponding data governance operations. The decision tracking and analysis module can be used to query and analyze decision information during the execution of governance rules. Based on the decision tracking chain structure generated by the decision tracking chain recording module, this module can retrieve decision nodes during the execution of governance rules and display the corresponding rule triggering conditions, execution paths, and execution results.

[0115] In some implementations, one or more modules within the extended governance module can work in conjunction with a governance feedback closed-loop control mechanism. Specifically, the semantic consistency monitoring module and decision tracking module provide the data input required for feedback analysis; the rule generation and update module generates or optimizes governance rules based on the feedback analysis results; and the entity version management module records version evolution information during the strategy optimization process, thus forming a system-level closed-loop governance optimization mechanism. These modules can be combined and deployed according to application requirements, and a cross-module information feedback path can be formed through the governance feedback closed-loop control mechanism to achieve adaptive optimization during system operation.

[0116] In some embodiments, the extended governance modules can be combined and deployed according to specific application requirements, thereby enabling flexible expansion of system functions.

[0117] In this embodiment, by introducing entity version management, rule version management, and rule optimization modules, and combining them with governance report generation, semantic consistency monitoring, and decision tracking and analysis modules, a multi-dimensional expansion of semantic governance capabilities is achieved. Beyond basic governance functions, version tracking, rule evolution, and governance effect analysis capabilities can be further provided, thereby supporting continuous optimization and system capability expansion in complex scenarios.

[0118] In an optional embodiment, the semantic compression loss evaluation module can also be used to quantitatively model the semantic compression loss value based on the degree of preservation of semantic information or the degree of semantic difference during the semantic mapping process. The semantic compression loss value is calculated based on multiple evaluation metrics and modeling the differences between the original data structure and the semantic structure, and is used to participate in the construction of evaluation nodes in the governance decision-making and decision tracing chain structure. The evaluation metrics include information completeness, attribute retention rate, relationship consistency, and semantic structure similarity.

[0119] In this embodiment, multiple evaluation metrics, including information completeness, attribute retention rate, relation consistency, and semantic structure similarity, are introduced to comprehensively model the semantic compression loss value, achieving a multi-dimensional quantitative evaluation of semantic differences. This avoids the bias of evaluation based on a single metric, making the semantic loss assessment more comprehensive and objective, thereby improving the accuracy and reliability of governance decisions.

[0120] In an optional embodiment, the semantic compression loss evaluation module can also be used to calculate the semantic compression loss value. Based on a preset semantic compression loss threshold and / or loss penalty parameter, corresponding governance rules or audit processes are triggered; the semantic compression loss value is determined based on a preset threshold, a dynamically adjusted threshold, or an adaptive learning threshold.

[0121] In this embodiment, by setting a semantic compression loss threshold and loss penalty parameters, and triggering corresponding governance rules or audit processes based on the loss value, automatic response based on the degree of semantic loss is achieved. By classifying and processing data according to the severity of semantic bias, the system can intervene in high-risk data in a timely manner, thereby improving the real-time performance of governance and risk control capabilities.

[0122] In an optional embodiment, the system may further include a verification module, which is used to perform consistency verification and / or integrity verification on decision node information and decision traceability chain structure.

[0123] In this embodiment, the consistency and integrity of the decision node information and the decision traceability chain structure are verified by the verification module, thereby realizing the verification and protection of decision record data. This can prevent the decision chain data from being tampered with or missing, and improve the data credibility and evidentiary value of the system in audit scenarios.

[0124] In an optional embodiment, the governance rule execution module is further configured to: perform governance control operations on the data inputs, model inference results, and / or automated decision-making results in the artificial intelligence system based on the semantic compression loss evaluation results; wherein the governance control operations include quality control of model input data, risk verification of model output results, and / or intervention control of automated decision-making behavior.

[0125] In this embodiment, by applying the semantic compression loss evaluation results to the data input, model inference results, and automated decision-making process of the artificial intelligence system, governance control of key AI aspects is achieved. This can identify and suppress low-quality inputs or abnormal outputs, improving the reliability and security of model decisions, thereby enhancing the controllability and compliance of the AI ​​system.

[0126] In an optional embodiment, the semantic inventory structure may further include: structured semantic objects, hierarchical structures, or other data structures that can express semantic entities and their relationships.

[0127] In this embodiment, by extending the data structure of the semantic manifest to support multiple expression methods such as structured semantic objects and hierarchical structures, the diversity of semantic modeling methods is achieved. This improves the system's ability to express complex semantic relationships while adapting to different business data structure requirements.

[0128] In an optional embodiment, the system is further configured to: Semantic representations are obtained by performing semantic layer processing on the data; Based on the differences or information changes between the semantic representation and the original data, the semantic consistency state is determined. Automated decision control or governance operations are performed based on the semantic consistency state; wherein, the semantic consistency state is determined based on the semantic compression loss evaluation result; The decision control process is recorded to form a traceable decision traceability chain structure; The decision traceability chain structure is used for cross-system sharing, cross-system traceability, or cross-system verification to achieve collaborative decision-making and auditing between different systems.

[0129] In this embodiment, semantic consistency is determined based on the difference between the semantic representation and the original data, and governance operations are performed accordingly. Simultaneously, a traceability chain is recorded throughout the entire process, achieving cross-system semantic governance and decision-making collaboration. Establishing a unified semantic governance and traceability mechanism across different systems enhances cross-system data consistency control and auditing capabilities.

[0130] To make the technical solution of this application easier to understand, several exemplary application scenarios are provided below to illustrate the application methods of the artificial intelligence semantic governance and decision tracing system provided in this application in different application environments. It should be noted that the following application scenarios are typical application methods based on the aforementioned semantic inventory data structure and do not constitute a limitation on the scope of protection of this application. Figures 1 to 4 The functional flow and deployment structure shown in this application indicate that the system provided can be applied to the following scenarios: Example 1: Enterprise data governance platform.

[0131] In one example application, the system provided in this application can be used in an enterprise data governance platform.

[0132] Enterprises typically have multiple data source systems, such as customer relationship management systems, enterprise resource planning systems, and data warehouse systems. Because these systems differ in data structure and semantic representation, semantic inconsistencies can easily arise. The system can use a semantic manifest building module to perform semantic parsing on data from different systems and construct a unified semantic manifest data structure, thereby achieving a unified semantic description of data across systems.

[0133] In some implementations, the semantic manifest object can record field identifiers, standardized semantic values, semantic mapping relationships, and data source information, thereby enabling semantic consistency management of enterprise data objects.

[0134] Through this exemplary application, enterprises can perform unified semantic management of data from different systems, thereby improving data governance capabilities and data interpretability.

[0135] Example 2: Artificial Intelligence Decision Audit System.

[0136] like Figure 4 As shown, the system provided in this application can be applied to an artificial intelligence decision auditing system.

[0137] When an AI model makes predictions or decisions based on input data, the system can use the semantic manifest construction module to perform semantic parsing and structured representation of the model's input data. The system can generate semantic manifest objects according to the semantic manifest construction process and provide a unified semantic description of the model's input data. Subsequently, the system can use the decision traceability chain recording module to record key decision-making node information during the model's decision-making process and generate corresponding structured record results of the decision process.

[0138] In some implementations, regulatory agencies or system administrators can query the corresponding decision traceability chain structure through an audit interface, thereby auditing the decision-making behavior of the artificial intelligence system.

[0139] This exemplary application allows for the complete recording of the execution path of an artificial intelligence system during the decision-making process, thereby enabling traceable recording of the artificial intelligence decision-making process.

[0140] Example 3: Data quality monitoring system.

[0141] The system provided in this application can be applied to data quality monitoring systems. The system can perform semantic structuring processing on input data through a semantic inventory building module, thereby constructing semantic inventory data objects.

[0142] In some implementations, the semantic manifest object can record semantic entity information, semantic mapping relationships, and semantic confidence information. The system can continuously monitor the semantic entities and their relationship structure in the semantic manifest structure through a semantic consistency monitoring mechanism. When semantic conflicts or data anomalies are detected, the system can trigger corresponding data governance rules to perform data correction or anomaly marking operations.

[0143] This exemplary application can improve the data quality and semantic consistency of a data system.

[0144] Example 4: Cross-system data integration platform.

[0145] The system provided in this application can be applied to cross-system data integration platforms. During data integration, different systems often differ in their data structures and semantic representations. For example... Figure 2 As shown, the system can use the semantic manifest building module to perform unified semantic parsing on data from different systems and generate a semantic manifest data structure. For example... Figure 3 As shown, a semantic manifest object can record raw data fields, standardized semantic fields, and semantic mapping relationships, thereby achieving semantic alignment of data fields across systems. In some implementations, such as... Figure 4As shown, the semantic processing and governance service layer in the system deployment structure can serve as a data semantic exchange layer between systems, enabling data semantic sharing between data governance systems, artificial intelligence systems, and decision-making systems.

[0146] This exemplary application allows for the establishment of a unified data semantic description model, thereby improving semantic accuracy in the cross-system data integration process.

[0147] In the above-mentioned application examples, the AI ​​governance and decision traceability system achieves structured expression, traceable recording, and continuous optimization of the AI ​​decision-making process through the synergistic effect of semantic inventory structure, semantic compression loss assessment, governance rule execution, decision traceability chain construction, and governance feedback closed-loop control mechanism. This makes it suitable for various high-complexity data processing and decision governance scenarios.

[0148] Example 2 In an optional embodiment, the decision tracing method based on semantic inventory structure and semantic compression loss evaluation includes: Semantic parsing and structured modeling are performed on data from external data sources to generate a unified semantic inventory structure; Based on the semantic list structure, the semantic information loss of data in the semantic mapping process is calculated and quantified to obtain the semantic compression loss evaluation result; Based on the semantic compression loss evaluation results and preset governance rules, automated governance decision-making operations are performed on semantic entities. The decision-making node information in the governance decision-making process is structured and associated with modeling to generate a decision traceability chain structure; Based on the aforementioned decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data, feedback analysis is performed on the governance decision results and feedback signals are generated. Based on the feedback signals, the governance rules, strategy parameters, or decision execution paths are updated or optimized to form a closed-loop governance mechanism. Based on the updated governance strategy, the lifecycle state of semantic entities is updated and managed.

[0149] In this embodiment, by associating the semantic inventory structure, semantic compression loss assessment results, decision tracing chain structure, and feedback signals, the path reconstruction, causal analysis, and continuous optimization of the governance decision-making process are achieved. The method is executed by a computer device and implemented by the processor calling instructions from memory. It then re-executes subsequent governance processes based on the optimized governance strategy to achieve continuous iterative execution of the closed-loop control process.

[0150] Example 3 In an optional embodiment, the AI-based semantic governance and decision tracing method based on semantic inventory structure and semantic compression loss evaluation includes: Semantic parsing and structured modeling are performed on data from external data sources to generate a unified semantic list structure; wherein, the semantic list structure includes: semantic object structure, semantic graph structure, knowledge representation structure or other data structures used to express semantic entities and their relationships; Based on the semantic list structure, the semantic information loss of data in the semantic mapping process is calculated and quantified to obtain the semantic compression loss evaluation result. Based on the semantic compression loss assessment results and the preset governance rules, governance decision operations are performed on semantic entities; the governance decision operations include: data correction, data tagging, access control and / or policy adjustment; The decision-making node information in the governance decision-making process is structured and associated with modeling to generate a decision traceability chain structure; the decision traceability chain structure includes: chain structure, graph structure, event flow structure or other data structures used to represent the relationship between decision nodes; Based on the aforementioned decision tracing chain structure, the lifecycle state of semantic entities is updated and managed.

[0151] In this embodiment, by sequentially executing semantic modeling, semantic compression loss assessment, governance decision-making, and decision traceability chain construction steps, a complete semantic governance and decision traceability methodology is achieved, thereby ensuring that the semantic governance process has consistency, reproducibility, and traceability.

[0152] In an optional embodiment, the method further includes: Perform semantic parsing on the raw data; Establish semantic mapping relationships; The original data is structured and modeled to generate semantic entities and their relational structures. The semantic entities and their relational structures are used for semantic compression loss assessment and governance decision modeling.

[0153] In this embodiment, by performing semantic parsing on the original data, establishing semantic mapping relationships, and performing structured modeling, the accurate construction of semantic entities and their relational structures is achieved. This improves the accuracy of semantic expression and provides high-quality input data for subsequent semantic loss assessment and governance decisions.

[0154] In an optional embodiment, the method further includes: The triggering conditions, execution paths, and execution results of governance rules are structurally constructed and associated models are built to generate a decision traceability chain structure.

[0155] In this embodiment, a complete expression of the decision tracing chain is achieved by structurally modeling the triggering conditions, execution paths, and execution results of governance rules. This not only records decision results but also fully reflects the decision triggering logic and execution process, thereby enhancing the depth and completeness of tracing analysis.

[0156] In an optional embodiment, the method further includes: Based on the semantic compression loss assessment results and / or semantic consistency assessment results, trigger the corresponding governance rules and / or audit processes; The governance rules are used to perform governance operations based on semantic consistency status, semantic anomaly detection results, or preset strategies.

[0157] In this embodiment, a semantically driven automated governance and auditing mechanism is achieved by triggering governance rules or audit processes based on semantic compression loss assessment results and semantic consistency status. The system can dynamically adjust governance strategies according to semantic changes, thereby improving the ability to identify semantic anomalies and the flexibility of governance responses.

[0158] Example 4 This application also provides a computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in the foregoing embodiments.

[0159] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device. Of course, the computer-readable storage medium may include both the internal storage unit and the external storage device of the computer device. In this embodiment, the computer-readable storage medium is typically used to store the operating system and various application software installed on the computer device, such as the program code of the method described in the foregoing embodiments. In addition, the computer-readable storage medium can also be used to temporarily store various types of data that have been output or will be output.

[0160] The above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit the scope of protection of this application. Those skilled in the art can make various modifications or substitutions to the technical solutions of this application without departing from the technical concept of this application, and such modifications or substitutions should all fall within the scope of protection of this application.

[0161] Furthermore, without departing from the technical concept of this application, those skilled in the art can implement various equivalent or alternative implementation methods based on the technical solutions disclosed in this application, including the following: 1. The semantic list building module can be implemented using rule-based, statistical, or machine learning model-based semantic parsing methods.

[0162] 2. The semantic list structure can be stored using relational databases, graph databases, document databases, key-value databases, or other data storage structures.

[0163] 3. The semantic compression loss evaluation module can be implemented using semantic difference evaluation methods based on semantic similarity calculation, statistical analysis, or machine learning models.

[0164] 4. The governance rule enforcement module can be implemented using a rule engine or an automated decision-making mechanism based on a strategy model.

[0165] 5. The decision traceability chain recording module can use, for example, a chained log structure, a log structure based on a distributed ledger, an event log system, or other data structures that support structured recording to realize decision path recording and tamper-proof storage.

[0166] 6. The entity lifecycle management module can use a state machine model or version control mechanism to implement semantic entity state management.

[0167] 7. The various modules of the system can be deployed using a microservice architecture or a monolithic application architecture.

[0168] 8. The system can be deployed on cloud computing platforms, local servers, enterprise data centers, distributed computing environments, or containerized computing platforms.

[0169] 9. The governance feedback closed-loop control module can be implemented using a feedback analysis mechanism based on rule analysis, statistical learning methods, or machine learning models. It is used to analyze historical decision data, semantic compression loss evaluation results, and decision tracing chain structure, and generate feedback signals to support governance rule optimization, strategy parameter adjustment, or decision path optimization.

[0170] As can be seen from the above examples, the technical solution of this application can be deployed and implemented in a variety of different ways, and can be flexibly adapted to different system architectures and application scenarios, thereby achieving closed-loop optimization of semantic modeling, semantic compression loss assessment, governance decision execution, decision tracing and governance feedback.

[0171] Without requiring inventive effort, all modifications, equivalent changes, or alternatives made by those skilled in the art based on the disclosure of this application shall fall within the protection scope of this application.

[0172] Obviously, those skilled in the art should understand that the modules or steps of the embodiments of this application described above can be implemented using general-purpose computer devices. They can be centralized on a single computer device or distributed across a network of multiple computer devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computer device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the embodiments of this application are not limited to any particular combination of hardware and software.

[0173] It should be noted that the above are merely preferred embodiments of this application and do not limit the scope of patent protection of this application. Any equivalent structural or procedural changes made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of this application.

Claims

1. A decision tracing system based on semantic inventory structure and semantic compression loss evaluation, characterized in that, include: The module comprises a semantic inventory construction module, a semantic compression loss assessment module, a governance rule execution module, a decision traceability chain recording module, a governance feedback closed-loop control module, and an entity lifecycle management module; among which: The semantic list construction module is used to: receive raw data from multiple data sources, perform semantic parsing on the raw data and perform unified semantic structure modeling to form a semantic list structure; wherein, the semantic list structure is a data structure used to represent semantic entities and their relationships, including key-value structure, object structure, semantic graph structure or knowledge representation structure; The semantic compression loss evaluation module is used to: quantitatively evaluate the degree of semantic information preservation based on a preset evaluation index during the process of converting the original data from the original data structure to the semantic structure; and generate a semantic compression loss evaluation result based on the evaluation result; wherein the semantic compression loss evaluation result is used to characterize the degree of semantic information preservation or the degree of semantic difference. The governance rule execution module is used to: perform governance decision operations on the semantic entities in the semantic list structure based on the semantic compression loss evaluation results and / or other semantic consistency evaluation results and preset governance rules, so as to obtain governance decision results; The decision traceability chain recording module is used to: construct and record the decision node information in the execution process of the governance rules in a structured manner; and to perform association modeling based on the execution order relationship and / or logical association relationship between multiple decision nodes to generate a decision traceability chain structure; wherein, the decision traceability chain structure is used to represent the data structure of decision nodes and their association relationships. The entity lifecycle management module is used to: perform lifecycle management on the semantic entity based on the governance decision results; The governance feedback closed-loop control module is used to: evaluate and analyze the governance decision results and generate feedback signals based on the decision tracing chain structure, semantic compression loss assessment results and historical governance decision data; and adjust the governance rules, decision paths or strategy parameters based on the feedback signals to form a closed-loop governance mechanism. The decision tracing chain structure, the semantic list structure, and the semantic compression loss evaluation results are related, and the relationship is used to support path reconstruction and node association analysis based on the decision tracing chain structure. Each module interacts with other modules through a data interface. The semantic list structure, semantic compression loss assessment results, decision tracing chain structure, and feedback signals are stored, transmitted, and processed in the form of structured data to achieve automated data processing and governance control in a computer environment. The feedback signals are used to drive the adjustment of data flow relationships or the reconstruction of processing paths between modules within the system.

2. The system according to claim 1, characterized in that, The semantic inventory building module is also used to perform semantic parsing and structured modeling on the input data to generate a unified semantic inventory data structure. The semantic inventory construction module includes: a data parsing submodule, a semantic mapping submodule, and a semantic relationship construction submodule; wherein: The data parsing submodule is used to perform semantic parsing processing on the raw data from multiple data sources in order to extract semantic entities, attribute information and their relational structures. The semantic mapping submodule is used to perform semantic alignment and mapping modeling on the data after semantic parsing based on a predefined and / or dynamically generated semantic structure model, so as to generate a unified semantic description structure. The semantic relationship construction submodule is used to construct a network of relationships between semantic entities based on the semantic entities; and to generate a semantic inventory data structure through the network of relationships; wherein the network of relationships includes hierarchical relationships, reference relationships, dependency relationships, and semantic relationships; The semantic inventory data structure is used for unified semantic representation across data sources, subsequent semantic consistency assessment, and governance decision processing; the semantic inventory data structure includes data associations established based on entity identifiers, semantic mapping relationships, and / or decision node reference relationships.

3. The system according to claim 1, characterized in that, The semantic compression loss assessment module is also used to quantitatively model and assess the degree of semantic information loss generated by the data during the semantic mapping process, so as to determine the degree of semantic information preservation and / or the degree of semantic difference. The semantic compression loss evaluation module includes: an information integrity calculation submodule, an attribute preservation rate evaluation submodule, and a relation consistency evaluation submodule; wherein: The information integrity calculation submodule is used to calculate the degree of semantic information retention during the semantic mapping process; The attribute retention rate evaluation submodule is used to evaluate the retention ratio between the mapped semantic entity attributes and the original semantic entity attributes. The relation consistency evaluation submodule is used to evaluate the degree of consistency between the mapped semantic entity relation structure and the original semantic entity relation structure; The semantic compression loss evaluation results are also used as the decision-making basis for the governance rule execution module and as evaluation information nodes in the construction of the decision traceability chain structure.

4. The system according to claim 1, characterized in that, The governance rule execution module is also used to perform rule matching, condition judgment and governance decision operations based on the semantic inventory data structure and semantic compression loss evaluation results, so as to realize automated semantic governance; The governance rule execution module includes: a rule matching submodule, a condition judgment submodule, and a decision execution submodule; wherein: The rule matching submodule is used to match preset governance rules based on semantic entities, attribute information and relational structure; The condition judgment submodule is used to determine the current semantic state based on preset rule triggering conditions in order to determine whether the governance execution conditions are met. The decision execution submodule is used to execute corresponding governance decision operations when the governance execution conditions are met; wherein the governance decision operations include: data tagging, status update, access control, process triggering or policy adjustment operations; The governance rule execution module is also used to analyze the semantic consistency status, semantic anomaly detection results, and semantic compression loss evaluation results, and output the decision-making process as traceable decision node information to realize dynamic governance decisions for semantic entities.

5. The system according to claim 1, characterized in that, The decision tracing chain recording module is also used for: constructing and recording the decision node information in the process of governance rule execution in a structured manner; and performing association modeling based on the execution order relationship and / or logical association relationship between multiple decision nodes to generate a decision tracing chain structure, so as to realize the path tracing and analysis of the decision process. The decision tracing chain recording module includes: a decision path recording submodule, an audit log submodule, and a correlation modeling submodule; wherein: The decision path recording submodule is used to construct and record the information of each decision node and its execution order during the execution of the governance rules; The audit log submodule is used to record the execution behavior, execution results and audit information of governance operations; The association modeling submodule is used to establish node association relationships based on the temporal sequence relationship, dependency relationship and / or causal relationship between the decision nodes, so as to form a decision tracing chain structure; the decision tracing chain structure is used for decision path reconstruction, cross-node causal relationship analysis and / or unified tracing of cross-system decision behavior.

6. The system according to claim 1, characterized in that, The governance feedback closed-loop control module is also used to form a closed-loop optimization mechanism based on governance decision results during system operation, so as to realize continuous evaluation, anomaly identification and strategy optimization of the governance decision process, thereby enabling the system to have a closed-loop governance capability based on historical decision results for adaptive adjustment, dynamic strategy update and continuous optimization. The governance feedback closed-loop control module includes: a feedback signal generation submodule, a feedback analysis submodule, a strategy adjustment submodule, and a closed-loop control submodule; wherein: The feedback signal generation submodule is used to generate structured feedback signals based on the decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data. The feedback analysis submodule is used to analyze the feedback signals to identify abnormal patterns, consistency deviations, or optimization opportunities in the governance decision-making process. The strategy adjustment submodule is used to update or optimize governance rules, strategy parameters, or decision execution paths based on feedback analysis results. The closed-loop control submodule is used to apply the optimized governance strategy back to the subsequent governance decision-making process, thereby forming a closed-loop governance mechanism of "decision-evaluation-feedback-optimization". The feedback signal includes structured control parameters generated based on statistical analysis results, model output results, or rule evaluation results.

7. The system according to claim 1, characterized in that, The entity lifecycle management module is also used for structured management and dynamic control of the full lifecycle state of semantic entities in the system, so as to realize lifecycle scheduling based on governance decisions; the entity lifecycle management module includes: an entity creation submodule, a state update submodule, an entity disposal submodule, and a lifecycle control submodule; wherein: The entity creation submodule is used to create the semantic entity and initialize the lifecycle state of the semantic entity; wherein, the initialization includes: configuring entity attributes based on contextual semantic relationships; The state update submodule is used to update the lifecycle state of semantic entities based on governance decision results and semantic state changes; the update includes state transition, risk level adjustment and / or governance tag change; The entity processing submodule is used to perform processing operations on semantic entities that meet preset conditions; the processing operations include archiving, freezing, isolation, deletion and / or access restriction; The lifecycle control submodule is used to schedule and control the lifecycle flow of semantic entities based on governance rules and system status. The lifecycle state is associated with the decision node information in the decision tracing chain structure, and the association is used for tracing and analyzing the semantic entity evolution process.

8. The system according to claim 1, characterized in that, The system further includes: an extended governance function module, used to extend and enhance the capabilities of the semantic governance process; the extended governance function module includes: an entity version management module, a governance report generation module, a rule version management module, a rule generation and optimization module, a semantic consistency monitoring module, and a decision tracking and analysis module; wherein: The entity version management module is used to record and manage the version information of semantic entities in order to realize version tracking and change analysis of semantic data. The governance report generation module is used to generate a structured governance report based on the execution results of the governance rules and the decision traceability chain structure. The rule version management module is used to control and update the governance rules, so as to realize the switching and comparative analysis between different rule versions; The rule generation and optimization module is used to optimize the governance rules based on historical decision data and external strategies. The semantic consistency monitoring module is used to continuously monitor the semantic consistency status in the semantic list structure. The decision tracking and analysis module is used to query, analyze, and reconstruct the decision path based on the decision tracing chain structure.

9. The system according to claim 1, characterized in that, The semantic compression loss evaluation module is also used to quantitatively model the semantic compression loss value based on the degree of preservation of semantic information or the degree of semantic difference during the semantic mapping process. The semantic compression loss value is calculated based on multiple evaluation metrics and modeling the differences between the original data structure and the semantic structure. It is used to participate in the construction of evaluation nodes in the governance decision-making and decision tracing chain structure. The evaluation metrics include information integrity, attribute retention rate, relationship consistency and semantic structure similarity.

10. The system according to claim 1, characterized in that, The semantic compression loss assessment module is also used to: calculate the semantic compression loss value; trigger corresponding governance rules or audit processes based on the preset semantic compression loss threshold and / or loss penalty parameters; the semantic compression loss value is judged based on the preset threshold, the dynamically adjusted threshold, or the adaptive learning threshold.

11. The system according to claim 1, characterized in that, The system also includes a verification module, which is used to perform consistency verification and / or integrity verification on decision node information and decision traceability chain structure.

12. The system according to claim 1, characterized in that, The governance rule execution module is also used to: perform governance control operations on the data input, model inference results, and / or automated decision-making results in the artificial intelligence system based on the semantic compression loss evaluation results; wherein, the governance control operations include quality control of model input data, risk verification of model output results, and / or intervention control of automated decision-making behavior.

13. The system according to claim 1, characterized in that, The semantic inventory structure also includes: structured semantic objects, hierarchical structures, or other data structures that can express semantic entities and their relationships.

14. The system according to claim 1, characterized in that, The system is also configured to: Semantic representations are obtained by performing semantic layer processing on the data; Based on the differences or information changes between the semantic representation and the original data, the semantic consistency state is determined. Automated decision control or governance operations are performed based on the semantic consistency state; wherein, the semantic consistency state is determined based on the semantic compression loss evaluation result; The decision control process is recorded to form a traceable decision traceability chain structure; The decision traceability chain structure is used for cross-system sharing, cross-system traceability, or cross-system verification to achieve collaborative decision-making and auditing between different systems.

15. A decision tracing method based on semantic inventory structure and semantic compression loss evaluation, characterized in that, The method includes: Semantic parsing and structured modeling are performed on data from external data sources to generate a unified semantic inventory structure; Based on the semantic list structure, the semantic information loss of data in the semantic mapping process is calculated and quantified to obtain the semantic compression loss evaluation result; Based on the semantic compression loss evaluation results and preset governance rules, automated governance decision-making operations are performed on semantic entities. The decision-making node information in the governance decision-making process is structured and associated with modeling to generate a decision traceability chain structure; Based on the aforementioned decision tracing chain structure, semantic compression loss assessment results, and historical governance decision data, feedback analysis is performed on the governance decision results and feedback signals are generated. Based on the feedback signals, the governance rules, strategy parameters, or decision execution paths are updated or optimized to form a closed-loop governance mechanism. Based on the updated governance strategy, the lifecycle state of semantic entities is updated and managed.

16. An AI-based semantic governance and decision tracing method based on semantic inventory structure and semantic compression loss evaluation, characterized in that, The method includes: Semantic parsing and structured modeling are performed on data from external data sources to generate a unified semantic list structure; wherein, the semantic list structure includes: semantic object structure, semantic graph structure, knowledge representation structure or other data structures used to express semantic entities and their relationships; Based on the semantic list structure, the semantic information loss of data in the semantic mapping process is calculated and quantified to obtain the semantic compression loss evaluation result. Based on the semantic compression loss assessment results and the preset governance rules, governance decision operations are performed on semantic entities; the governance decision operations include: data correction, data tagging, access control and / or policy adjustment; The decision-making node information in the governance decision-making process is structured and associated with modeling to generate a decision traceability chain structure; the decision traceability chain structure includes: chain structure, graph structure, event flow structure or other data structures used to represent the relationship between decision nodes; Based on the aforementioned decision tracing chain structure, the lifecycle state of semantic entities is updated and managed.

17. The method according to claim 16, characterized in that, The method further includes: Perform semantic parsing on the raw data; Establish semantic mapping relationships; The original data is structured and modeled to generate semantic entities and their relational structures. The semantic entities and their relational structures are used for semantic compression loss assessment and governance decision modeling.

18. The method according to claim 16, characterized in that, The method further includes: The triggering conditions, execution paths, and execution results of governance rules are structurally constructed and associated models are built to generate a decision traceability chain structure.

19. The method according to claim 16, characterized in that, The method further includes: Based on the semantic compression loss assessment results and / or semantic consistency assessment results, trigger the corresponding governance rules and / or audit processes; The governance rules are used to perform governance operations based on semantic consistency status, semantic anomaly detection results, or preset strategies.

20. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method as described in any one of claims 15 to 19.

Images