A method and system for trusted traceability of data throughout its entire lifecycle based on digital twins
By using the digital twin-based data lifecycle trusted traceability method, the problem of difficulty in identifying shifts in business meaning when field values remain unchanged has been solved in existing technologies, achieving accurate location and trusted traceability of data semantic changes.
Patent Information
- Application Number
- CN202611416661.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-14
- Publication Date
- 2026-11-10
AI Technical Summary
Existing data tracing systems cannot identify situations where field values remain unchanged but their business meanings have shifted, resulting in semantic changes going unrecognized. Risk analysis may be misjudged, and the tracing process struggles to determine the range and duration of semantic shifts.
A trusted data lifecycle traceability method based on digital twins is adopted. Through modules such as the same-value semantic original extraction module, the counterfactual translation residual generation module, the bidirectional semantic carrying offset evaluation module, the twin semantic crease value generation module, and the twin crease propagation traceability module, the bidirectional exchange and deduction of generation rules and business interpretation actions of the preceding and following nodes and the calculation of semantic offset are realized.
It can accurately identify and quantify situations where field values remain consistent but generation rules or business interpretation actions change, providing a basis for locating semantic changes, preventing misjudgments and the triggering of erroneous business states, and ensuring the credibility of the tracing process.
Smart Images

Figure CN123070129A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of information technology and data management technology, specifically to a method and system for trusted traceability of the entire data lifecycle based on digital twins. Background Technology
[0002] In the field of data governance, digital twins are increasingly being applied to trusted traceability throughout the entire data lifecycle. They are used to record the flow of data at various lifecycle nodes, including collection, processing, transmission, use, and archiving. The generation rules, judgment conditions, and business interpretation actions adopted when field values pass through different lifecycle nodes are the specific analytical objects for identifying whether the semantics of the fields have changed.
[0003] Existing data traceability systems typically track data flow based on field identifiers, field values, timestamps, node numbers, or hash records. When the normalized field values of the preceding and following nodes are identical, existing systems usually consider them semantically consistent. However, the same field values may be generated by different rules, or they may trigger different business interpretation actions at the preceding and following nodes. Simply comparing field values cannot verify whether the preceding node's generation rule can support the business interpretation action of the following node, nor can it verify whether the following node's generation rule can reverse-interpret the preceding node's business interpretation action. Therefore, it is difficult to identify situations where the numerical value remains unchanged but the business meaning has shifted.
[0004] The above situations typically arise from rule version changes, adjustments to the judgment scope, replacement of enumeration meanings, changes in state combinations, or business process migrations. When field values remain unchanged across nodes, numerical consistency can obscure the forks between the generated rules and the business interpretation actions; subsequent lifecycle nodes continue to reference the same field value and execute judgments according to the changed rules, causing semantic offsets to propagate along the field usage path. After such propagation occurs, the data source records may still remain continuous, but audit results may misjudge semantic consistency, business states may be triggered incorrectly, risk analysis may miss the starting node of the semantic change, and the tracing process may struggle to determine the range and duration of nodes traversed by the semantic offset. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides a reliable data lifecycle traceability method and system based on digital twins, solving the problems mentioned in the background section.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a data lifecycle trusted traceability system based on digital twins, including a same-value semantic original fragment extraction module, a counterfactual translation residual generation module, a bidirectional semantic bearing offset evaluation module, a twin semantic crease value generation module, and a twin crease propagation traceability module;
[0007] The same-value semantic original fragment extraction module performs a normalized comparison between the field value Vf of the preceding node and the field value Vb of the following node to obtain the common field value V. It then generates a rule fragment Rf for the preceding node, a rule fragment Rb for the following node, a business interpretation action Ef for the preceding node, and a business interpretation action Eb for the following node along the field assignment path, and forms the same-value semantic original fragment AW.
[0008] The counterfactual translation residual generation module is based on the same-value semantic original AW. It performs bidirectional exchange and deduction of the rules and business interpretation actions before and after the translation, obtains the forward translation path difference Dsf and the backward translation path difference Dsb, and generates the semantic translation residual CM.
[0009] The bidirectional semantic load offset evaluation module is based on the semantic translation residual CM. It solves the reverse allowable domain of the common field value V in the preceding and following rules, obtains the forward load distance Qfo and the backward load distance Qbo, and generates the bidirectional load offset J.
[0010] The twin semantic crease value generation module generates a translation asymmetry quantity KU based on the forward translation path difference Dsf and the backward translation path difference Dsb, and combines it with the bidirectional load offset J, the forward load distance Qfo and the backward load distance Qbo to generate the twin semantic crease value F, and form the semantic crease result CMN.
[0011] The twin crease propagation tracing module interprets subsequent lifecycle nodes using dual rules based on the semantic crease result CMN, obtains the semantic propagation length G, and generates a full lifecycle semantic reliable tracing result T based on the common field value V, the twin semantic crease value F, and the semantic propagation length G.
[0012] Preferably, the same-value semantic original fragment extraction module includes a same-value normalization locking unit and a rule-action oriented extraction unit;
[0013] The same value normalization locking unit receives the previous node field value Vf from the previous lifecycle node and the subsequent node field value Vb from the subsequent lifecycle node;
[0014] First, read the data type identifier, character expression form, and encoding expression form of the preceding node field value Vf and the following node field value Vb, and generate the preceding node normalized value nVf and the following node normalized value nVb according to the processing order of "data type conversion - removal of characters without business meaning - encoding value parsing".
[0015] Among them, numeric fields directly retain the numeric result, character numbers are converted into numeric results, and coded fields are parsed into comparable values according to the actual coding table used by the current node;
[0016] Perform an equality operation on the normalized value nVf of the previous node and the normalized value nVb of the next node. After the normalization results are consistent, lock the current consistent result as the common field value V, and determine the assignment position of the previous node and the assignment position of the next node according to the execution record of the common field value V in the two life cycle nodes respectively.
[0017] The rule-action-oriented extraction unit reads the common field value V, the assignment position of the previous node, and the assignment position of the next node, and combines them with the existing business execution records in the two lifecycle nodes to construct a field execution dependency chain:
[0018] Starting from the current previous node's assignment position, traverse backwards along the data dependency direction, continuously reading the assignment statements, conditional statements, and intermediate operation statements that directly participate in the generation of the common field value V;
[0019] When the reverse traversal stops at the first conditional branch that can change the value of the common field V, the actual statements executed between the current conditional branch and the previous node's assignment position are compressed into the previous node's generated rule fragment Rf.
[0020] The same processing is performed on the subsequent node assignment positions to obtain the subsequent node generation rule fragment Rb;
[0021] After obtaining the previous node generation rule fragment Rf and the next node generation rule fragment Rb, the system traverses forward along the data usage direction, starting from the previous node assignment position and the next node assignment position respectively, skipping intermediate statements that only execute storage, forwarding, and format conversion, until the common field value V first enters the condition branch that can drive changes in business state. The execution action actually triggered by the current condition branch is extracted into the previous node business interpretation action Ef and the next node business interpretation action Eb.
[0022] The same-value semantic original fragment AW is formed by encapsulating the common field value V, the rule fragment generated by the previous node Rf, the rule fragment generated by the next node Rb, the business interpretation action of the previous node Ef, and the business interpretation action of the next node Eb in a fixed order.
[0023] Preferably, the counterfactual translation residual generation module includes a counterfactual action cross-deduction unit and a translation bifurcation quantification and residual condensation unit;
[0024] The counterfactual action cross-deduction unit is based on the semantic original AW of the same value, and performs bidirectional cross-deduction on the previous node's generated rule fragment Rf, the subsequent node's generated rule fragment Rb, the previous node's business interpretation action Ef, and the subsequent node's business interpretation action Eb, under the condition that the common field value V remains unchanged:
[0025] Substitute the common field value V into the rule fragment Rf generated by the previous node, and expand the judgment nodes step by step according to the execution order of the conditions already formed in the rule fragment Rf generated by the previous node. Retain the judgment nodes and action nodes that are actually activated by the common field value V from the rule entry to the business interpretation action Ef of the previous node, and form the original execution path Pff of the previous node.
[0026] Substitute the common field value V into the rule fragment Rb generated by the subsequent node, and obtain the original execution path Pbb of the subsequent node from the rule entry of the subsequent node to the business interpretation action Eb of the subsequent node according to the same process;
[0027] After obtaining the native execution path Pff of the preceding node and the native execution path Pbb of the following node, a cross-side exchange is performed on the business interpretation action without changing the common field value V:
[0028] The subsequent node's business interpretation action Eb is used as the counterfactual termination action of the preceding node's generated rule fragment Rf. The conditions that can trigger the current action are reverse-analyzed from the business state contained in the subsequent node's business interpretation action Eb. The parsed conditions are then linked to the judgment nodes of the preceding node's generated rule fragment Rf level by level.
[0029] For judgment nodes that can directly pass the common field value V, the deduction continues along the original branch. For judgment nodes that cannot directly pass the common field value V, the common field value V remains unchanged, and the search continues along another branch of the current judgment node to find the shortest control path that can reach the subsequent node's business interpretation action Eb, thus forming the pre-rule carrying post-interpretation path Pfb.
[0030] The preceding node's business interpretation action Ef is used as the counterfactual termination action of the subsequent node's generated rule fragment Rb. The same reverse action parsing, condition attachment, and fixed value deduction processing are used to form the subsequent rule carrying the preceding interpretation path Pbf.
[0031] Preferably, the mutual translation fork quantization and residual agglomeration unit is based on the original execution path Pff of the previous node, the original interpretation path Pfb of the previous rule, the original execution path Pbb of the subsequent node, and the original interpretation path Pbf of the subsequent rule, and quantifies the semantic deviation generated by bidirectional exchange based on three processing positions: "common path end position - length of independent path after fork - termination business action".
[0032] Starting from the rule entry point of the original execution path Pff of the previous node and the rule entry point of the interpretation path Pfb carried by the previous rule, compare the judgment nodes that have been passed level by level, locate the judgment node that the two paths passed together last time, and determine the current judgment node as the forward mutual translation hinge node Hf.
[0033] Starting from the forward translation hinge node Hf, proceed along the original execution path Pff of the previous node and the interpretation path Pfb of the previous rule bearer to their respective termination business actions. Count the judgment nodes passed after the fork according to the order of execution, and determine the total number of independent judgment nodes on both sides as the forward translation path difference Dsf.
[0034] Similarly, the last common judgment node is located in the original execution path Pbb of the back node and the interpretation path Pbf of the back rule, and the backward translation hinge node Hb is obtained. The number of judgment nodes that each of the two paths after the backward translation hinge node Hb passes through independently is counted, and the backward translation path difference Dsb is obtained.
[0035] The semantic translation residual CM is calculated by combining the forward translation path difference Dsf and the backward translation path difference Dsb, using the following formula: .
[0036] Where, forward translation path difference Dsf represents the control path displacement caused by the forward node generating rule fragment Rf shifting from carrying the forward node service interpretation action Ef to carrying the subsequent node service interpretation action Eb when the common field value V does not change.
[0037] The backward translation path difference Dsb represents the control path displacement caused by the subsequent node generating rule fragment Rb shifting from carrying the subsequent node service interpretation action Eb to carrying the preceding node service interpretation action Ef.
[0038] Semantic translation residuals (CM) then condense the path displacements in both directions into a single fundamental quantity used for subsequent bidirectional semantic carrying analysis.
[0039] Preferably, the bidirectional semantic bearer offset evaluation module includes a rule-reverse allowable domain parsing unit and a bidirectional bearer distance calculation unit;
[0040] The rule reverse permission domain parsing unit takes the subsequent node business interpretation action Eb as the target action, traverses in reverse along the execution structure of the rule fragment Rf generated by the previous node, reads the judgment conditions that the subsequent node business interpretation action Eb needs to satisfy, including numerical range conditions, enumeration restriction conditions, state combination conditions and logical association conditions, and combines multiple judgment conditions according to the original rule execution order to form the previous rule reverse permission domain Lf.
[0041] Among them, the reverse permission field Lf of the previous rule indicates the value space of the common field value V that can support the business interpretation action Eb of the subsequent node in the previous node rule system;
[0042] The preceding node's business interpretation action Ef is used as the target action. The execution structure of the subsequent node's generated rule fragment Rb is traversed in reverse. The rule conditions that can trigger the preceding node's business interpretation action Ef are parsed in the same way, and the subsequent rule reverse permission domain Lb is formed.
[0043] Among them, the reverse permission field Lb of the post-rule indicates the value space that the common field value V can support for the business interpretation action Ef of the previous node in the post-node rule system;
[0044] Preferably, the bidirectional bearer distance calculation unit receives the preceding rule reverse allowable field Lf, the following rule reverse allowable field Lb, the common field value V, and the semantic translation residual CM. It calculates the rule offset distance generated by the common field value V when carrying the other party's business interpretation action in the two rule systems, and generates the bidirectional bearer offset J. The process is as follows:
[0045] Map the common field value V to the reverse allowable domain Lf of the previous rule and the reverse allowable domain Lb of the subsequent rule, and analyze the differences between the common field value V and the judgment conditions in the two allowable domains.
[0046] For the forward rule reverse allowable domain Lf, a path search is performed along the rule decision tree from the node where the common field value V is currently located to the target branch direction that can enter the forward rule reverse allowable domain Lf. The number of decision nodes that need to be crossed is counted to obtain the forward carrying distance Qfo.
[0047] For the backward allowed domain Lb, the same path search method is used to obtain the backward bearing distance Qbo;
[0048] Among them, the forward bearer distance Qfo represents the degree of rule path offset generated when the common field value VV supports the subsequent node's business interpretation action Eb in the rule fragment Rf generated by the front node;
[0049] The backward bearing distance Qbo represents the degree of rule path offset generated when the common field value V supports the business interpretation action Ef of the preceding node in the rule fragment Rb generated by the following node;
[0050] The forward load-bearing distance Qfo and the backward load-bearing distance Qbo are combined with the semantic translation residual CM to calculate the bidirectional load-bearing offset J, as shown in the following formula:
[0051] .
[0052] Preferably, the twin semantic crease value generation module includes a mutual translation direction difference calculation unit and a crease value fusion generation unit;
[0053] The mutual translation direction difference calculation unit is based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb, and analyzes the degree of rule interpretation and transfer to calculate and obtain the mutual translation asymmetry quantity KU.
[0054] Specifically, first, obtain the number of path bifurcations and path offset lengths in the forward translation path difference Dsf and the backward translation path difference Dsb, and then calculate the translation offset difference in the two directions respectively.
[0055] Since the forward translation path difference Dsf represents the degree of path change when the rule fragment Rf of the preceding node carries the business interpretation action Eb of the following node, and the backward translation path difference Dsb represents the degree of path change when the rule fragment Rb of the following node carries the business interpretation action Ef of the preceding node, the difference between the two directions can reflect the asymmetric change in the rule carrying capacity during the business interpretation migration process.
[0056] The mutual translation asymmetric quantity KU is obtained through the following formula:
[0057] ;
[0058] When the path changes of the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb are close, it indicates that the forward and backward rule systems have mutual interpretability.
[0059] When the change in one direction is greater than that in another direction, it indicates that the common field value V has generated a directional business interpretation offset during the lifecycle transmission process.
[0060] Preferably, the crease value fusion generation unit uses the bidirectional load offset J as the basic offset and uses the mutual translation asymmetric quantity KU to correct the direction of the basic offset so that the calculation results can reflect the degree of imbalance in the transfer of rule interpretation before and after.
[0061] Finally, the bidirectional common offset formed by the forward bearing distance Qfo and the backward bearing distance Qbo is added to the calculation to obtain the twin semantic crease value F;
[0062] The twin semantic crease value F is obtained using the following formula:
[0063] ;
[0064] Based on the twin semantic crease value F, the lifecycle node number that currently generates the crease value FF, the common field value V, the forward translation path difference Dsf, and the backward translation path difference Dsb are associated and fitted into the semantic crease result CMN.
[0065] Preferably, the twin crease propagation tracing module includes a crease node propagation verification unit and a full lifecycle semantic trustworthy result generation unit;
[0066] The crease node propagation verification unit extracts the semantic crease node positions in the semantic crease result CMN, and determines the starting node where the semantics of the current field changes based on the semantic crease node positions, and takes the node as the crease starting node;
[0067] Read the field usage records in subsequent nodes sequentially along the digital twin's lifecycle, filter out lifecycle nodes that still reference the common field value VV, and form a sequence of subsequent lifecycle nodes. ;
[0068] In the formula, nm represents the m-th subsequent node used in chronological order.
[0069] For each subsequent lifecycle node ni, read the actual business interpretation action executed by that node, and call the previous node generation rule fragment Rf saved before the crease start node and the subsequent node generation rule fragment Rb saved after the crease start node respectively to perform dual rule deduction for the same business interpretation action;
[0070] The subsequent node's business interpretation action is input into the previous node's generated rule fragment Rf, and matching is performed along the previous node's rule judgment path to obtain the previous rule deduction result Xf; at the same time, the same business interpretation action is input into the subsequent node's generated rule fragment Rb, and matching is performed along the subsequent node's rule judgment path to obtain the subsequent rule deduction result Xb.
[0071] Perform path consistency analysis on the derivation results Xf of the previous rule and Xb of the subsequent rule:
[0072] When both the current rule derivation result Xf and the subsequent rule derivation result Xb can reach the same business interpretation action, the current node is marked as a semantically unpropagated node.
[0073] If the current rule derivation result Xb can maintain the current business interpretation action, while Xf cannot generate the current business interpretation action along the original rule path, then the current node is marked as a crease propagation node.
[0074] The semantic propagation length G is generated based on the number of consecutively occurring crease propagation nodes, using the following formula: ;
[0075] In the formula, This represents the propagation identifier of the i-th subsequent lifecycle node, which is 1 when the node belongs to the crease propagation node, and 0 otherwise.
[0076] The full lifecycle semantic trustworthy result generation unit performs lifecycle association on the common field value VV, twin semantic crease value FF, crease start node position and semantic propagation length GG, takes the common field value VV as the tracking object, takes the crease start node as the semantic change location point, and takes the semantic propagation length GG as the influence range description quantity.
[0077] The full lifecycle semantic trustworthy traceability result T is generated according to the digital twin lifecycle sequence, as follows:
[0078] ;
[0079] In the formula, This represents the crease propagation path, used to record the nodes that generate semantic changes and subsequent propagation nodes.
[0080] During the generation process, each lifecycle node in the crease propagation path P is further associated with the corresponding rule version, business interpretation action, and node position, so that the final tracing result can be displayed simultaneously:
[0081] The flow of the common field value V throughout its lifecycle;
[0082] The location where the twin semantic crease value F is first generated;
[0083] The length G of the semantic crease propagating along subsequent nodes;
[0084] The complete lifecycle path of semantic interpretation changes.
[0085] A method for trusted traceability of data throughout its entire lifecycle based on digital twins includes the following steps:
[0086] Step 1: The same-value semantic original fragment extraction module performs a normalized comparison between the field value Vf of the preceding node and the field value Vb of the following node to obtain the common field value V. Then, along the field assignment path, the preceding node generates a rule fragment Rf, the following node generates a rule fragment Rb, the preceding node generates a business interpretation action Ef, and the following node generates a business interpretation action Eb, and forms the same-value semantic original fragment AW.
[0087] Step 2: The counterfactual translation residual generation module uses the same-value semantic original AW as a basis to perform bidirectional exchange and deduction of the preceding and following rules and business interpretation actions, obtain the forward translation path difference Dsf and the backward translation path difference Dsb, and generate the semantic translation residual CM.
[0088] Step 3: The bidirectional semantic bearer offset evaluation module, based on the semantic translation residual CM, solves the reverse allowable domain of the common field value V in the preceding and following rules, obtains the forward bearer distance Qfo and the backward bearer distance Qbo, and generates the bidirectional bearer offset J.
[0089] Step 4: The twin semantic crease value generation module generates the mutual translation asymmetry quantity KU based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb, and combines it with the bidirectional bearing offset J, the forward bearing distance Qfo and the backward bearing distance Qbo to generate the twin semantic crease value F, and form the semantic crease result CMN.
[0090] Step 5: The twin crease propagation tracing module performs dual-rule interpretation on subsequent lifecycle nodes based on the semantic crease result CMN, obtains the semantic propagation length G, and generates the full lifecycle semantic reliable tracing result T based on the common field value V, the twin semantic crease value F, and the semantic propagation length G.
[0091] This invention provides a method and system for trusted data lifecycle traceability based on digital twins, which has the following beneficial effects:
[0092] (1) During system operation, the mutual translation asymmetry quantity KU is calculated based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb. The mutual translation asymmetry quantity KU, the bidirectional bearer offset J, the forward bearer distance Qfo, and the backward bearer distance Qbo are included in the calculation of the twin semantic crease value F. Subsequently, the twin semantic crease value F is combined with the lifecycle node number, the common field value V, the forward mutual translation path difference Dsf, and the backward mutual translation path difference Dsb to form the semantic crease result CMN. The semantic crease result CMN can indicate the node where the field semantics begin to change and retain the path basis for forming the twin semantic crease value F, so that the semantic changes caused by rule version changes, judgment range adjustments, enumeration meaning replacements, state combination changes, or business process migrations have verifiable positioning basis.
[0093] (2) Starting from the assignment positions of the previous and subsequent nodes, the system traverses backward along the data dependency direction, continuously reading only the assignment statements, conditional statements, and intermediate operation statements that directly participate in the generation of the common field value V. The traversal stops when the first conditional branch that can change the value of the common field value V is reached, and the traversal is compressed to form the previous node generation rule fragment Rf and the subsequent node generation rule fragment Rb, respectively. The start and end range of the rule fragments are determined based on the actual executed statements, which can exclude program statements that do not participate in the formation process of the common field value V. The previous node generation rule fragment Rf and the subsequent node generation rule fragment Rb centrally represent the basis for the value of the common field value V in the two nodes, which is convenient for identifying the situation where the field values are the same but the generation conditions have changed.
[0094] (3) The judgment nodes are compared step by step from the rule entry points of the original execution path Pff of the preceding node and the interpretation path Pfb of the preceding rule. The forward translation articulation node Hf is determined by the judgment node that is last traversed together. The number of judgment nodes that each of the two paths after the forward translation articulation node Hf is counted to obtain the forward translation path difference Dsf. At the same time, the backward translation articulation node Hb and the backward translation path difference Dsb are obtained from the original execution path Pbb of the following node and the interpretation path Pbf of the following rule, and then the semantic translation residual CM is calculated from the forward translation path difference Dsf and the backward translation path difference Dsb. The above calculation is based on the fork position and the length of the independent path after the fork, so that the semantic deviation has a clear path source and can distinguish the different degrees of deviation generated when the preceding rule interprets the business interpretation action Eb of the following node and the following rule interprets the business interpretation action Ef of the following node.
[0095] (4) Determine the starting node of the semantic crease from the position of the semantic crease node in the semantic crease result CMN, read the field usage records of subsequent nodes in the digital twin lifecycle sequence, and filter the lifecycle nodes that still reference the common field value V; for each filtered lifecycle node, call the previous node to generate rule fragment Rf and the subsequent node to generate rule fragment Rb to deduce the same business interpretation action, and obtain the previous rule deduction result Xf and the subsequent rule deduction result Xb. By comparing whether the previous rule deduction result Xf and the subsequent rule deduction result Xb can reach the same business interpretation action, the ordinary continued use of the common field value V and the semantic crease propagation can be distinguished, preventing the semantics from being considered consistent simply because subsequent nodes continue to use the same field value. Attached Figure Description
[0096] Figure 1 This is a schematic diagram of the block diagram of a trusted traceability system for the entire lifecycle of data based on digital twins according to the present invention.
[0097] Figure 2 This is a schematic diagram illustrating the steps of a trusted data lifecycle traceability method based on digital twins according to the present invention. Detailed Implementation
[0098] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0099] Example 1
[0100] This invention provides a trusted data lifecycle traceability system based on digital twins. Please refer to [link / reference]. Figure 1 It includes a semantic original extraction module, a counterfactual translation residual generation module, a bidirectional semantic load offset evaluation module, a twin semantic crease value generation module, and a twin crease propagation tracing module;
[0101] The same-value semantic original fragment extraction module performs a normalized comparison between the field value Vf of the preceding node and the field value Vb of the following node to obtain the common field value V. It then generates a rule fragment Rf for the preceding node, a rule fragment Rb for the following node, a business interpretation action Ef for the preceding node, and a business interpretation action Eb for the following node along the field assignment path, and forms the same-value semantic original fragment AW.
[0102] The counterfactual translation residual generation module is based on the same-value semantic original AW. It performs bidirectional exchange and deduction of the rules and business interpretation actions before and after the translation, obtains the forward translation path difference Dsf and the backward translation path difference Dsb, and generates the semantic translation residual CM.
[0103] The bidirectional semantic load offset evaluation module is based on the semantic translation residual CM. It solves the reverse allowable domain of the common field value V in the preceding and following rules, obtains the forward load distance Qfo and the backward load distance Qbo, and generates the bidirectional load offset J.
[0104] The twin semantic crease value generation module generates a translation asymmetry quantity KU based on the forward translation path difference Dsf and the backward translation path difference Dsb, and combines it with the bidirectional load offset J, the forward load distance Qfo and the backward load distance Qbo to generate the twin semantic crease value F, and form the semantic crease result CMN.
[0105] The twin crease propagation tracing module interprets subsequent lifecycle nodes using dual rules based on the semantic crease result CMN, obtains the semantic propagation length G, and generates a full lifecycle semantic reliable tracing result T based on the common field value V, the twin semantic crease value F, and the semantic propagation length G.
[0106] In this embodiment, by reading the data type identifier, character representation, and encoded representation of the preceding node field value Vf and the following node field value Vb, the common field value V is obtained in the order of data type conversion, removal of characters without business meaning, and parsing of encoded values. Subsequently, the dependency chain is executed along the field to extract the preceding node generation rule fragment Rf, the following node generation rule fragment Rb, the preceding node business interpretation action Ef, and the following node business interpretation action Eb, forming the semantic original fragment AW with the same value. Based on the above processing, the tracing object is no longer limited to field identifiers, field values, and flow records, but extends to the formation process and business usage process of field values, which can identify situations where the common field value V remains consistent but the generation rule or business interpretation action changes.
[0107] Under the condition that the common field value V remains unchanged, the subsequent node service interpretation action Eb is placed in the preceding node's generated rule fragment Rf for deduction, and the preceding node service interpretation action Ef is placed in the subsequent node's generated rule fragment Rb for deduction. By locating the forward translation articulated node Hf and the backward translation articulated node Hb, the independent judgment nodes after the bifurcation are continuously counted to obtain the forward translation path difference Dsf and the backward translation path difference Dsb. Then, the reverse parsing of the preceding rule's reverse allowable field Lf and the following rule's reverse allowable field Lb is performed to obtain the forward carrying distance Qfo and the backward carrying distance Qbo, respectively. The above process transforms the path bifurcation, rule restrictions, and directional differences generated when the preceding and following rules are mutually interpreted into computable data, so that the semantic offsets obscured by numerical consistency can be identified and quantified.
[0108] Starting from the initial node of the semantic crease marked by the CMN in the digital twin lifecycle result, subsequent lifecycle nodes that continue to reference the common field value V are selected along the digital twin lifecycle sequence. The business interpretation actions of each node are deduced using the rule fragments Rf and Rb generated by the preceding and following nodes, respectively. Crease propagation nodes are marked based on the preceding rule deduction result Xf and the following rule deduction result Xb, and the semantic propagation length G is calculated. Finally, a full lifecycle semantic trustworthy traceability result T is generated. Using the full lifecycle semantic trustworthy traceability result T, the starting node, subsequent propagation nodes, and duration of semantic changes can be identified simultaneously. This allows auditors to distinguish between "continuous field values" and "continuous field semantics," and to verify the triggering location and scope of impact of erroneous business states.
[0109] Example 2
[0110] Please refer to Figure 1 Specifically: the same-value semantic original fragment extraction module includes the same-value normalization locking unit and the rule-action oriented extraction unit;
[0111] The same value normalization locking unit receives the previous node field value Vf from the previous lifecycle node and the subsequent node field value Vb from the subsequent lifecycle node;
[0112] First, read the data type identifier, character expression form, and encoding expression form of the preceding node field value Vf and the following node field value Vb, and generate the preceding node normalized value nVf and the following node normalized value nVb according to the processing order of "data type conversion - removal of characters without business meaning - encoding value parsing".
[0113] Perform an equality operation on the normalized value nVf of the previous node and the normalized value nVb of the next node. After the normalization results are consistent, lock the current consistent result as the common field value V, and determine the assignment position of the previous node and the assignment position of the next node according to the execution record of the common field value V in the two life cycle nodes respectively.
[0114] The rule-action-oriented extraction unit reads the common field value V, the assignment position of the previous node, and the assignment position of the next node, and combines them with the existing business execution records in the two lifecycle nodes to construct a field execution dependency chain:
[0115] Starting from the current previous node's assignment position, traverse backwards along the data dependency direction, continuously reading the assignment statements, conditional statements, and intermediate operation statements that directly participate in the generation of the common field value V;
[0116] When the reverse traversal stops at the first conditional branch that can change the value of the common field V, the actual statements executed between the current conditional branch and the previous node's assignment position are compressed into the previous node's generated rule fragment Rf.
[0117] The same processing is performed on the subsequent node assignment positions to obtain the subsequent node generation rule fragment Rb;
[0118] After obtaining the previous node generation rule fragment Rf and the next node generation rule fragment Rb, the system traverses forward along the data usage direction, starting from the previous node assignment position and the next node assignment position respectively, skipping intermediate statements that only execute storage, forwarding, and format conversion, until the common field value V first enters the condition branch that can drive changes in business state. The execution action actually triggered by the current condition branch is extracted into the previous node business interpretation action Ef and the next node business interpretation action Eb.
[0119] The same-value semantic original fragment AW is formed by encapsulating the common field value V, the rule fragment generated by the previous node Rf, the rule fragment generated by the next node Rb, the business interpretation action of the previous node Ef, and the business interpretation action of the next node Eb in a fixed order.
[0120] In this embodiment, following the order of data type conversion, removal of characters without business meaning, and parsing of encoded values, a normalized value nVf for the preceding node and a normalized value nVb for the following node are generated respectively. After the equivalence operation results are consistent, the common field value V is locked. This process can gradually peel away the differences in expression caused by data type, character format, and encoding method, allowing the preceding node field value Vf and the following node field value Vb to be compared under a unified expression form. It also includes field values with different expression forms but the same actual value in subsequent analysis, overcoming the same-value identification bias that occurs when comparing only the original characters or original codes.
[0121] Based on the execution records of the common field value V in the two lifecycle nodes, the assignment positions of the preceding and following nodes are determined respectively. These two assignment positions are then used as the starting point for analyzing the field execution dependency chain. Thus, the tracing entry point for the common field value V shifts from the field name or field storage location to the actual assignment event, enabling the system to distinguish between fields with the same name, duplicate fields, and multiple assignment scenarios. It also pinpoints the actual formation position of the common field value V in the preceding and following lifecycle nodes, establishing a clear starting point for analyzing the field value generation process.
[0122] After generating rule fragments, the system traverses forward along the data usage direction, skipping intermediate statements that only handle storage, forwarding, or format conversion, until the common field value V first enters a conditional branch that can drive a change in business state. Then, it extracts the preceding node's business interpretation action Ef and the following node's business interpretation action Eb. Subsequently, the common field value V, the preceding node's generated rule fragment Rf, the following node's generated rule fragment Rb, the preceding node's business interpretation action Ef, and the following node's business interpretation action Eb are encapsulated in a fixed order into a semantically identical fragment AW. With the semantically identical fragment AW, the basis for the same field value and the business usage result can be retained in the same analysis carrier. This allows the tracing process to distinguish between "consistent field values" and "consistent field business meaning," and establishes a structured analytical foundation for subsequent identification of changes in generated rules and changes in business interpretation actions.
[0123] Example 3
[0124] Please refer to Figure 1 Specifically: the counterfactual translation residual generation module includes a counterfactual action cross-inference unit and a translation bifurcation quantification and residual condensation unit;
[0125] The counterfactual action cross-deduction unit is based on the semantic original AW of the same value, and performs bidirectional cross-deduction on the previous node's generated rule fragment Rf, the subsequent node's generated rule fragment Rb, the previous node's business interpretation action Ef, and the subsequent node's business interpretation action Eb, under the condition that the common field value V remains unchanged:
[0126] Substitute the common field value V into the rule fragment Rf generated by the previous node, and expand the judgment nodes step by step according to the execution order of the conditions already formed in the rule fragment Rf generated by the previous node. Retain the judgment nodes and action nodes that are actually activated by the common field value V from the rule entry to the business interpretation action Ef of the previous node, and form the original execution path Pff of the previous node.
[0127] Substitute the common field value V into the rule fragment Rb generated by the subsequent node, and obtain the original execution path Pbb of the subsequent node from the rule entry of the subsequent node to the business interpretation action Eb of the subsequent node according to the same process;
[0128] After obtaining the native execution path Pff of the preceding node and the native execution path Pbb of the following node, a cross-side exchange is performed on the business interpretation action:
[0129] The subsequent node's business interpretation action Eb is used as the counterfactual termination action of the preceding node's generated rule fragment Rf. The conditions that can trigger the current action are reverse-analyzed from the business state contained in the subsequent node's business interpretation action Eb. The parsed conditions are then linked to the judgment nodes of the preceding node's generated rule fragment Rf level by level.
[0130] For judgment nodes that can directly pass through the common field value V, the deduction continues along the original branch. For judgment nodes that cannot directly pass through the common field value V, the common field value V remains unchanged, and the search continues along another branch of the current judgment node to find the shortest control path that can reach the subsequent node's business interpretation action Eb, thus forming the pre-rule carrying post-interpretation path Pfb.
[0131] The preceding node's business interpretation action Ef is used as the counterfactual termination action of the subsequent node's generated rule fragment Rb. The same reverse action parsing, condition attachment, and fixed value deduction processing are used to form the subsequent rule carrying the preceding interpretation path Pbf.
[0132] The mutual translation fork quantification and residual agglomeration unit is based on the original execution path Pff of the previous node, the original interpretation path Pfb of the previous rule, the original execution path Pbb of the subsequent node, and the original interpretation path Pbf of the subsequent rule. It quantifies the semantic deviation generated by bidirectional exchange based on three processing positions: "common path end position - length of independent path after fork - termination business action".
[0133] Starting from the rule entry point of the original execution path Pff of the previous node and the rule entry point of the interpretation path Pfb carried by the previous rule, compare the judgment nodes that have been passed level by level, locate the judgment node that the two paths passed together last time, and determine the current judgment node as the forward mutual translation hinge node Hf.
[0134] Starting from the forward translation hinge node Hf, proceed along the original execution path Pff of the previous node and the interpretation path Pfb of the previous rule bearer to their respective termination business actions. Count the judgment nodes passed after the fork according to the order of execution, and determine the total number of independent judgment nodes on both sides as the forward translation path difference Dsf.
[0135] Similarly, the last common judgment node is located in the original execution path Pbb of the back node and the interpretation path Pbf of the back rule, and the backward translation hinge node Hb is obtained. The number of judgment nodes that each of the two paths after the backward translation hinge node Hb passes through independently is counted, and the backward translation path difference Dsb is obtained.
[0136] The semantic translation residual CM is calculated by combining the forward translation path difference Dsf and the backward translation path difference Dsb, using the following formula: .
[0137] The bidirectional semantic bearer offset evaluation module includes a rule-reverse allowable domain parsing unit and a bidirectional bearer distance calculation unit;
[0138] The rule reverse permission domain parsing unit takes the subsequent node business interpretation action Eb as the target action, traverses in reverse along the execution structure of the rule fragment Rf generated by the previous node, reads the judgment conditions that the subsequent node business interpretation action Eb needs to satisfy, including numerical range conditions, enumeration restriction conditions, state combination conditions and logical association conditions, and combines multiple judgment conditions according to the original rule execution order to form the previous rule reverse permission domain Lf.
[0139] The preceding node's business interpretation action Ef is used as the target action. The execution structure of the subsequent node's generated rule fragment Rb is traversed in reverse. The rule conditions that can trigger the preceding node's business interpretation action Ef are parsed in the same way, and the subsequent rule reverse permission domain Lb is formed.
[0140] The bidirectional bearer distance calculation unit receives the reverse allowable domain Lf of the preceding rule, the reverse allowable domain Lb of the following rule, the common field value V, and the semantic translation residual CM. It calculates the rule offset distance generated when the common field value V carries the other party's business interpretation action in the two rule systems, and generates the bidirectional bearer offset J. The process is as follows:
[0141] Map the common field value V to the reverse allowable domain Lf of the previous rule and the reverse allowable domain Lb of the subsequent rule, and analyze the differences between the common field value V and the judgment conditions in the two allowable domains.
[0142] For the forward rule reverse allowable domain Lf, a path search is performed along the rule decision tree from the node where the common field value V is currently located to the target branch direction that can enter the forward rule reverse allowable domain Lf. The number of decision nodes that need to be crossed is counted to obtain the forward carrying distance Qfo.
[0143] For the backward allowed domain Lb, the same path search method is used to obtain the backward bearing distance Qbo;
[0144] The forward load-bearing distance Qfo and the backward load-bearing distance Qbo are combined with the semantic translation residual CM to calculate the bidirectional load-bearing offset J, as shown in the following formula:
[0145] .
[0146] In this embodiment, assuming the common field value V remains constant, the common field value V is first substituted into the rule fragment Rf generated by the preceding node and the rule fragment Rb generated by the following node, respectively. The judgment nodes are then expanded according to the conditional execution order to obtain the original execution path Pff of the preceding node and the original execution path Pbb of the following node. Next, the business interpretation action Ef of the preceding node and the business interpretation action Eb of the following node are exchanged, forming the preceding rule carrying the subsequent interpretation path Pfb and the subsequent rule carrying the preceding interpretation path Pbf. Since the common field value V does not change during the bidirectional cross-deduction, the path differences in the deduction results can point to the semantic differences between the generated rule and the business interpretation action, allowing the tracing process to verify whether the same field value still expresses the same business meaning in the preceding and following rule systems.
[0147] Taking the subsequent node's business interpretation action Eb as the target action, the rule fragment Rf generated by the preceding node is traversed in reverse. The numerical range conditions, enumeration restrictions, state combination conditions, and logical association conditions required to trigger the subsequent node's business interpretation action Eb are read sequentially and combined according to the original rule execution order to form the reverse allowed domain Lf of the preceding rule. Simultaneously, taking the previous node's business interpretation action Ef as the target action, the same analysis is performed along the rule fragment Rb generated by the subsequent node to form the reverse allowed domain Lb of the subsequent rule. By reversing the business interpretation action into verifiable rule conditions, it is possible to identify which conditions the common field value V is required to meet when the preceding and following rule systems carry each other's business interpretation actions, thus overcoming the problem that changes in rule conditions cannot be identified when only comparing business action names or execution results.
[0148] The common field value V is mapped to the forward rule reverse allowable domain Lf and the backward rule reverse allowable domain Lb, respectively. The system searches along the rule decision tree for the decision nodes required to enter the target branch, obtaining the forward bearer distance Qfo and the backward bearer distance Qbo. These distances are then combined with the semantic translation residual CM to calculate the bidirectional bearer offset J. The bidirectional bearer offset J simultaneously retains the path branching information generated by the bidirectional exchange and the number of decision nodes that the common field value V needs to traverse to enter the other party's business interpretation conditions. This enables the system to identify situations where field values are continuous but the rule bearer capacity has changed, and provides auditors with a calculation basis for verifying the direction of semantic changes, rule positions, and the degree of deviation.
[0149] Example 4
[0150] Please refer to Figure 1 Specifically: the twin semantic crease value generation module includes a mutual translation direction difference calculation unit and a crease value fusion generation unit;
[0151] The mutual translation direction difference calculation unit is based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb, and analyzes the degree of rule interpretation and transfer to calculate and obtain the mutual translation asymmetry quantity KU.
[0152] The mutual translation asymmetric quantity KU is obtained through the following formula:
[0153] ;
[0154] When the path changes of the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb are close, it indicates that the forward and backward rule systems have mutual interpretability.
[0155] When the change in one direction is greater than that in another direction, it indicates that the common field value V has generated a directional business interpretation offset during the lifecycle transmission process.
[0156] The crease value fusion generation unit uses the bidirectional load offset J as the basic offset and uses the mutual translation asymmetric quantity KU to correct the direction of the basic offset so that the calculation results can reflect the degree of imbalance in the transfer of rule interpretation before and after.
[0157] Finally, the bidirectional common offset formed by the forward bearing distance Qfo and the backward bearing distance Qbo is added to the calculation to obtain the twin semantic crease value F;
[0158] The twin semantic crease value F is obtained using the following formula:
[0159] ;
[0160] Based on the twin semantic crease value F, the lifecycle node number that currently generates the crease value FF, the common field value V, the forward translation path difference Dsf, and the backward translation path difference Dsb are associated and fitted into the semantic crease result CMN.
[0161] The twin crease propagation tracing module includes a crease node propagation verification unit and a full lifecycle semantic trustworthy result generation unit;
[0162] The crease node propagation verification unit extracts the semantic crease node positions in the semantic crease result CMN, and determines the starting node where the semantics of the current field changes based on the semantic crease node positions, and takes the node as the crease starting node;
[0163] Read the field usage records in subsequent nodes sequentially along the digital twin's lifecycle, filter out lifecycle nodes that still reference the common field value VV, and form a sequence of subsequent lifecycle nodes. ;
[0164] In the formula, nm represents the m-th subsequent node used in chronological order.
[0165] For each subsequent lifecycle node ni, read the actual business interpretation action executed by the node, and call the previous node generation rule fragment Rf saved before the crease start node and the subsequent node generation rule fragment Rb saved after the crease start node respectively to perform dual rule deduction for the same business interpretation action;
[0166] The subsequent node's business interpretation action is input into the previous node's generated rule fragment Rf, and matching is performed along the previous node's rule judgment path to obtain the previous rule deduction result Xf; at the same time, the same business interpretation action is input into the subsequent node's generated rule fragment Rb, and matching is performed along the subsequent node's rule judgment path to obtain the subsequent rule deduction result Xb.
[0167] Perform path consistency analysis on the derivation results Xf of the previous rule and Xb of the subsequent rule:
[0168] When both the current rule derivation result Xf and the subsequent rule derivation result Xb can reach the same business interpretation action, the current node is marked as a semantically unpropagated node.
[0169] If the current rule derivation result Xb can maintain the current business interpretation action, while Xf cannot generate the current business interpretation action along the original rule path, then the current node is marked as a crease propagation node.
[0170] The semantic propagation length G is generated based on the number of consecutively occurring crease propagation nodes, using the following formula: ;
[0171] In the formula, This represents the propagation identifier of the i-th subsequent lifecycle node, which is 1 when the node belongs to the crease propagation node, and 0 otherwise.
[0172] The full lifecycle semantic trustworthy result generation unit performs lifecycle association on the common field value VV, twin semantic crease value FF, crease start node position and semantic propagation length GG, takes the common field value VV as the tracking object, takes the crease start node as the semantic change location point, and takes the semantic propagation length GG as the influence range description quantity.
[0173] The full lifecycle semantic trustworthy traceability result T is generated according to the digital twin lifecycle sequence, as follows:
[0174] ;
[0175] In the formula, This represents the crease propagation path, used to record the nodes that generate semantic changes and subsequent propagation nodes.
[0176] In this embodiment, the forward translation path difference Dsf and the backward translation path difference Dsb are used as the basis for calculation. The translation asymmetry quantity KU is calculated by comparing the path changes in the two translation directions. When the path changes on both sides are close, it can be determined that the rule fragment Rf generated by the forward node and the rule fragment Rb generated by the backward node have mutual interpretation capabilities. When the path change on one side is greater than that on the other side, it can be determined that the common field value V has a directional business interpretation offset during the lifecycle transmission. This processing no longer merges the bidirectional translation results into a non-directional difference result, but retains the directional difference between the forward rule interpretation of the backward node's business interpretation action Eb and the backward rule interpretation of the forward node's business interpretation action Ef, so that the unbalanced state of rule interpretation migration can be identified.
[0177] Using the bidirectional bearing offset J as the base offset, directional correction is performed using the mutual translation asymmetry KU. Then, the bidirectional common offset formed by the forward bearing distance Qfo and the backward bearing distance Qbo is incorporated to calculate the twin semantic crease value F. Subsequently, the twin semantic crease value F, lifecycle node number, common field value V, forward mutual translation path difference Dsf, and backward mutual translation path difference Dsb are combined according to a unified structure to form the semantic crease result CMN. Thus, the semantic crease result CMN can both identify nodes where field semantics have changed and retain the calculation basis such as directional differences, bearing distances, and path bifurcations, providing verifiable node evidence for situations where the field value remains unchanged but the business meaning has changed.
[0178] The node where the previous rule derivation result Xf cannot generate the current business interpretation action along the original rule path, while the subsequent rule derivation result Xb can maintain the current business interpretation action, is marked as a crease propagation node. The semantic propagation length G is calculated based on the number of consecutively occurring crease propagation nodes. Then, following the lifecycle sequence, using the common field value V as the tracking object, the crease start node as the semantic change location point, and the semantic propagation length G as the influence range descriptor, a full lifecycle semantic reliable traceability result T is generated. With the full lifecycle semantic reliable traceability result T, auditors can identify which node the semantic change originated from, which subsequent nodes it passed through, and its duration. Based on this, they can verify the location where business status errors were triggered, resolving the problem of only having data source records but being unable to identify the starting point and propagation range of semantic changes.
[0179] Example 5
[0180] A trusted data lifecycle traceability method based on digital twins, please refer to... Figure 2 Specifically, it includes the following steps:
[0181] Step 1: The same-value semantic original fragment extraction module performs a normalized comparison between the field value Vf of the preceding node and the field value Vb of the following node to obtain the common field value V. Then, along the field assignment path, the preceding node generates a rule fragment Rf, the following node generates a rule fragment Rb, the preceding node generates a business interpretation action Ef, and the following node generates a business interpretation action Eb, and forms the same-value semantic original fragment AW.
[0182] Step 2: The counterfactual translation residual generation module uses the same-value semantic original AW as a basis to perform bidirectional exchange and deduction of the preceding and following rules and business interpretation actions, obtain the forward translation path difference Dsf and the backward translation path difference Dsb, and generate the semantic translation residual CM.
[0183] Step 3: The bidirectional semantic bearer offset evaluation module, based on the semantic translation residual CM, solves the reverse allowable domain of the common field value V in the preceding and following rules, obtains the forward bearer distance Qfo and the backward bearer distance Qbo, and generates the bidirectional bearer offset J.
[0184] Step 4: The twin semantic crease value generation module generates the mutual translation asymmetry quantity KU based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb, and combines it with the bidirectional bearing offset J, the forward bearing distance Qfo and the backward bearing distance Qbo to generate the twin semantic crease value F, and form the semantic crease result CMN.
[0185] Step 5: The twin crease propagation tracing module performs dual-rule interpretation on subsequent lifecycle nodes based on the semantic crease result CMN, obtains the semantic propagation length G, and generates the full lifecycle semantic reliable tracing result T based on the common field value V, the twin semantic crease value F, and the semantic propagation length G.
[0186] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.
Claims
1. A data lifecycle-wide trusted traceability system based on digital twins, characterized in that: It includes a semantic original extraction module, a counterfactual translation residual generation module, a bidirectional semantic load offset evaluation module, a twin semantic crease value generation module, and a twin crease propagation tracing module; The same-value semantic original fragment extraction module performs a normalized comparison between the field value Vf of the preceding node and the field value Vb of the following node to obtain the common field value V. It then generates a rule fragment Rf for the preceding node, a rule fragment Rb for the following node, a business interpretation action Ef for the preceding node, and a business interpretation action Eb for the following node along the field assignment path, and forms the same-value semantic original fragment AW. The counterfactual translation residual generation module is based on the same-value semantic original AW. It performs bidirectional exchange and deduction of the rules and business interpretation actions before and after the translation, obtains the forward translation path difference Dsf and the backward translation path difference Dsb, and generates the semantic translation residual CM. The bidirectional semantic load offset evaluation module is based on the semantic translation residual CM. It solves the reverse allowable domain of the common field value V in the preceding and following rules, obtains the forward load distance Qfo and the backward load distance Qbo, and generates the bidirectional load offset J. The twin semantic crease value generation module generates a translation asymmetry quantity KU based on the forward translation path difference Dsf and the backward translation path difference Dsb, and combines it with the bidirectional load offset J, the forward load distance Qfo and the backward load distance Qbo to generate the twin semantic crease value F, and form the semantic crease result CMN. The twin crease propagation tracing module interprets subsequent lifecycle nodes using dual rules based on the semantic crease result CMN, obtains the semantic propagation length G, and generates a full lifecycle semantic reliable tracing result T based on the common field value V, the twin semantic crease value F, and the semantic propagation length G.
2. The data lifecycle trusted traceability system based on digital twins according to claim 1, characterized in that: The same-value semantic original fragment extraction module includes a same-value normalization locking unit and a rule-action oriented extraction unit; The same value normalization locking unit receives the previous node field value Vf from the previous lifecycle node and the subsequent node field value Vb from the subsequent lifecycle node; First, read the data type identifier, character expression form, and encoding expression form of the preceding node field value Vf and the following node field value Vb, and generate the preceding node normalized value nVf and the following node normalized value nVb according to the processing order of "data type conversion - removal of characters without business meaning - parsing of encoding value". Perform an equality operation on the normalized value nVf of the previous node and the normalized value nVb of the next node. After the normalization results are consistent, lock the current consistent result as the common field value V, and determine the assignment position of the previous node and the assignment position of the next node according to the execution record of the common field value V in the two life cycle nodes respectively. The rule-action-oriented extraction unit reads the common field value V, the assignment position of the previous node, and the assignment position of the next node, and combines them with the existing business execution records in the two lifecycle nodes to construct a field execution dependency chain: Starting from the current previous node's assignment position, traverse backwards along the data dependency direction, continuously reading the assignment statements, conditional statements, and intermediate operation statements that directly participate in the generation of the common field value V; When the reverse traversal stops at the first conditional branch that can change the value of the common field V, the actual statements executed between the current conditional branch and the previous node's assignment position are compressed into the previous node's generated rule fragment Rf. The same processing is performed on the subsequent node assignment positions to obtain the subsequent node generation rule fragment Rb; After obtaining the previous node generation rule fragment Rf and the next node generation rule fragment Rb, the system traverses forward along the data usage direction, starting from the previous node assignment position and the next node assignment position respectively, skipping intermediate statements that only execute storage, forwarding, and format conversion, until the common field value V first enters the condition branch that can drive changes in business state. The execution action actually triggered by the current condition branch is extracted into the previous node business interpretation action Ef and the next node business interpretation action Eb. The same-value semantic original fragment AW is formed by encapsulating the common field value V, the rule fragment generated by the previous node Rf, the rule fragment generated by the next node Rb, the business interpretation action of the previous node Ef, and the business interpretation action of the next node Eb in a fixed order.
3. A data lifecycle trusted traceability system based on digital twins according to claim 2, characterized in that: The counterfactual translation residual generation module includes a counterfactual action cross-inference unit and a translation bifurcation quantification and residual condensation unit; The counterfactual action cross-deduction unit is based on the semantic original AW of the same value, and performs bidirectional cross-deduction on the previous node's generated rule fragment Rf, the subsequent node's generated rule fragment Rb, the previous node's business interpretation action Ef, and the subsequent node's business interpretation action Eb, under the condition that the common field value V remains unchanged: Substitute the common field value V into the rule fragment Rf generated by the previous node, and expand the judgment nodes step by step according to the execution order of the conditions already formed in the rule fragment Rf generated by the previous node. Retain the judgment nodes and action nodes that are actually activated by the common field value V from the rule entry to the business interpretation action Ef of the previous node, and form the original execution path Pff of the previous node. Substitute the common field value V into the rule fragment Rb generated by the subsequent node, and obtain the original execution path Pbb of the subsequent node from the rule entry of the subsequent node to the business interpretation action Eb of the subsequent node according to the same process; After obtaining the native execution path Pff of the preceding node and the native execution path Pbb of the following node, a cross-side exchange is performed on the business interpretation action: The subsequent node's business interpretation action Eb is used as the counterfactual termination action of the preceding node's generated rule fragment Rf. The conditions that can trigger the current action are reverse-analyzed from the business state contained in the subsequent node's business interpretation action Eb. The parsed conditions are then linked to the judgment nodes of the preceding node's generated rule fragment Rf level by level. For judgment nodes that can directly pass through the common field value V, the deduction continues along the original branch. For judgment nodes that cannot directly pass through the common field value V, the common field value V remains unchanged, and the search continues along another branch of the current judgment node to find the shortest control path that can reach the subsequent node's business interpretation action Eb, thus forming the pre-rule carrying post-interpretation path Pfb. The preceding node's business interpretation action Ef is used as the counterfactual termination action of the subsequent node's generated rule fragment Rb. The same reverse action parsing, condition attachment, and fixed value deduction processing are used to form the subsequent rule carrying the preceding interpretation path Pbf.
4. A data lifecycle trusted traceability system based on digital twins according to claim 3, characterized in that: The mutual translation fork quantification and residual agglomeration unit is based on the original execution path Pff of the previous node, the original interpretation path Pfb of the previous rule, the original execution path Pbb of the subsequent node, and the original interpretation path Pbf of the subsequent rule. It quantifies the semantic deviation generated by bidirectional exchange based on three processing positions: "common path end position - length of independent path after fork - termination business action". Starting from the rule entry point of the original execution path Pff of the previous node and the rule entry point of the interpretation path Pfb carried by the previous rule, compare the judgment nodes that have been passed level by level, locate the judgment node that the two paths passed together last time, and determine the current judgment node as the forward mutual translation hinge node Hf. Starting from the forward translation hinge node Hf, proceed along the original execution path Pff of the previous node and the interpretation path Pfb of the previous rule bearer to their respective termination business actions. Count the judgment nodes passed after the fork according to the order of execution, and determine the total number of independent judgment nodes on both sides as the forward translation path difference Dsf. Similarly, the last common judgment node is located in the original execution path Pbb of the back node and the interpretation path Pbf of the back rule, and the backward translation hinge node Hb is obtained. The number of judgment nodes that each of the two paths after the backward translation hinge node Hb passes through independently is counted, and the backward translation path difference Dsb is obtained. The semantic translation residual CM is calculated by combining the forward translation path difference Dsf and the backward translation path difference Dsb, using the following formula: .
5. A data lifecycle trusted traceability system based on digital twins according to claim 4, characterized in that: The bidirectional semantic bearer offset evaluation module includes a rule-reverse allowable domain parsing unit and a bidirectional bearer distance calculation unit; The rule reverse permission domain parsing unit takes the subsequent node business interpretation action Eb as the target action, traverses in reverse along the execution structure of the rule fragment Rf generated by the previous node, reads the judgment conditions that the subsequent node business interpretation action Eb needs to satisfy, including numerical range conditions, enumeration restriction conditions, state combination conditions and logical association conditions, and combines multiple judgment conditions according to the original rule execution order to form the previous rule reverse permission domain Lf. The preceding node's business interpretation action Ef is used as the target action. The execution structure of the subsequent node's generated rule fragment Rb is traversed in reverse. The rule conditions that can trigger the preceding node's business interpretation action Ef are parsed in the same way, and the subsequent rule reverse permission domain Lb is formed.
6. A data lifecycle trusted traceability system based on digital twins according to claim 5, characterized in that: The bidirectional bearer distance calculation unit receives the reverse allowable domain Lf of the preceding rule, the reverse allowable domain Lb of the following rule, the common field value V, and the semantic translation residual CM. It calculates the rule offset distance generated when the common field value V carries the other party's business interpretation action in the two rule systems, and generates the bidirectional bearer offset J. The process is as follows: Map the common field value V to the reverse allowable domain Lf of the previous rule and the reverse allowable domain Lb of the subsequent rule, and analyze the differences between the common field value V and the judgment conditions in the two allowable domains. For the forward rule reverse allowable domain Lf, a path search is performed along the rule decision tree from the node where the common field value V is currently located to the target branch direction that can enter the forward rule reverse allowable domain Lf. The number of decision nodes that need to be crossed is counted to obtain the forward carrying distance Qfo. For the backward allowed domain Lb, the same path search method is used to obtain the backward bearing distance Qbo; The forward load-bearing distance Qfo and the backward load-bearing distance Qbo are combined with the semantic translation residual CM to calculate the bidirectional load-bearing offset J, as shown in the following formula: 。 7. A data lifecycle trusted traceability system based on digital twins according to claim 6, characterized in that: The twin semantic crease value generation module includes a mutual translation direction difference calculation unit and a crease value fusion generation unit; The mutual translation direction difference calculation unit is based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb, and analyzes the degree of rule interpretation and transfer to calculate and obtain the mutual translation asymmetry quantity KU. The mutual translation asymmetric quantity KU is obtained through the following formula: ; When the path changes of the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb are close, it indicates that the forward and backward rule systems have mutual interpretability. When the change in one direction is greater than that in another direction, it indicates that the common field value V has generated a directional business interpretation offset during the lifecycle transmission process.
8. A data lifecycle trusted traceability system based on digital twins according to claim 7, characterized in that: The crease value fusion generation unit uses the bidirectional load offset J as the basic offset and uses the mutual translation asymmetric quantity KU to correct the direction of the basic offset so that the calculation results can reflect the degree of imbalance in the transfer of rule interpretation before and after. Finally, the bidirectional common offset formed by the forward bearing distance Qfo and the backward bearing distance Qbo is added to the calculation to obtain the twin semantic crease value F; The twin semantic crease value F is obtained using the following formula: ; Based on the twin semantic crease value F, the lifecycle node number that currently generates the crease value FF, the common field value V, the forward translation path difference Dsf, and the backward translation path difference Dsb are associated and fitted into the semantic crease result CMN.
9. A data lifecycle trusted traceability system based on digital twins according to claim 8, characterized in that: The twin crease propagation tracing module includes a crease node propagation verification unit and a full lifecycle semantic trustworthy result generation unit; The crease node propagation verification unit extracts the semantic crease node positions in the semantic crease result CMN, and determines the starting node where the semantics of the current field changes based on the semantic crease node positions, and takes the node as the crease starting node; Read the field usage records in subsequent nodes sequentially along the digital twin's lifecycle, filter out lifecycle nodes that still reference the common field value VV, and form a sequence of subsequent lifecycle nodes. ; In the formula, nm represents the m-th subsequent node used in chronological order. For each subsequent lifecycle node ni, read the actual business interpretation action executed by the node, and call the previous node generation rule fragment Rf saved before the crease start node and the subsequent node generation rule fragment Rb saved after the crease start node respectively to perform dual rule deduction for the same business interpretation action; The subsequent node's business interpretation action is input into the previous node's generated rule fragment Rf, and matching is performed along the previous node's rule judgment path to obtain the previous rule deduction result Xf; at the same time, the same business interpretation action is input into the subsequent node's generated rule fragment Rb, and matching is performed along the subsequent node's rule judgment path to obtain the subsequent rule deduction result Xb. Perform path consistency analysis on the derivation results Xf of the previous rule and Xb of the subsequent rule: When both the current rule derivation result Xf and the subsequent rule derivation result Xb can reach the same business interpretation action, the current node is marked as a semantically unpropagated node. If the current rule derivation result Xb can maintain the current business interpretation action, while Xf cannot generate the current business interpretation action along the original rule path, then the current node is marked as a crease propagation node. The semantic propagation length G is generated based on the number of consecutively occurring crease propagation nodes, using the following formula: ; In the formula, This represents the propagation identifier of the i-th subsequent lifecycle node, which is 1 when the node belongs to the crease propagation node, and 0 otherwise. The full lifecycle semantic trustworthy result generation unit performs lifecycle association on the common field value VV, twin semantic crease value FF, crease start node position and semantic propagation length GG, takes the common field value VV as the tracking object, takes the crease start node as the semantic change location point, and takes the semantic propagation length GG as the influence range description quantity. The full lifecycle semantic trustworthy traceability result T is generated according to the digital twin's lifecycle sequence, as follows: ; In the formula, This represents the crease propagation path, used to record the nodes that generate semantic changes and subsequent propagation nodes.
10. A data lifecycle trusted traceability method based on digital twins, applied to a data lifecycle trusted traceability system based on digital twins as described in any one of claims 1 to 9, characterized in that: Includes the following steps: Step 1: The same-value semantic original fragment extraction module performs a normalized comparison between the field value Vf of the preceding node and the field value Vb of the following node to obtain the common field value V. Then, along the field assignment path, the preceding node generates a rule fragment Rf, the following node generates a rule fragment Rb, the preceding node generates a business interpretation action Ef, and the following node generates a business interpretation action Eb, and forms the same-value semantic original fragment AW. Step 2: The counterfactual translation residual generation module uses the same-value semantic original AW as a basis to perform bidirectional exchange and deduction of the preceding and following rules and business interpretation actions, obtain the forward translation path difference Dsf and the backward translation path difference Dsb, and generate the semantic translation residual CM. Step 3: The bidirectional semantic bearer offset evaluation module, based on the semantic translation residual CM, solves the reverse allowable domain of the common field value V in the preceding and following rules, obtains the forward bearer distance Qfo and the backward bearer distance Qbo, and generates the bidirectional bearer offset J. Step 4: The twin semantic crease value generation module generates the mutual translation asymmetry quantity KU based on the forward mutual translation path difference Dsf and the backward mutual translation path difference Dsb, and combines it with the bidirectional bearing offset J, the forward bearing distance Qfo and the backward bearing distance Qbo to generate the twin semantic crease value F, and form the semantic crease result CMN. Step 5: The twin crease propagation tracing module performs dual-rule interpretation on subsequent lifecycle nodes based on the semantic crease result CMN, obtains the semantic propagation length G, and generates the full lifecycle semantic reliable tracing result T based on the common field value V, the twin semantic crease value F, and the semantic propagation length G.
