Dynamic measurement and calculation method and system for house compensation oriented to land making mode

Abstract

The invention provides a dynamic house compensation measuring and calculating method and system oriented to a land making mode, and the method comprises the steps: carrying out the structural modeling of various compensation rules in a local policy file, extracting input variables, parameters and weights, accessing a house basic platform, completing the data standardization, constructing a symbolized tracking engine, and achieving the dynamic measurement and calculation of the house compensation. A compensation rule calculation process is abstracted into a directed graph, and dynamic propagation and tracing of influence weights among nodes are achieved; in the calculation process, different rule types are distinguished, differential weight adjustment is adopted, variable influence scores, main driving factor sorting and other structured analysis results are generated, and finally a compensation amount composition and influence chain is displayed in a visualized multi-angle mode. User decision support is optimized, and the fairness and traceability of policy execution are enhanced.

Description

Technical Field

[0001] This invention relates to the field of housing expropriation compensation calculation and result interpretability enhancement technology, and in particular to a dynamic calculation method and system for housing compensation oriented towards land development models. Background Technology

[0002] Current housing expropriation compensation calculation systems are widely used in urban renewal and land expropriation, aiming to standardize complex policies and rules and automate the calculation of housing compensation amounts using data-driven methods. Mainstream existing technologies generally employ policy rule engines or parametric formula-based modeling to automatically calculate multiple indicators such as physical attributes, ownership status, location value, and construction year of the housing. Common methods include process-driven rule calls, conditional judgments, weighting, and linear or non-linear superposition compensation calculation modes. These systems typically integrate seamlessly with basic housing data management platforms, achieving significant progress in improving calculation efficiency and ensuring compliance. With the advancement of informatization, the standardization and localization of compensation calculation systems are continuously improving, and they are gradually acquiring functions such as modular rule configuration, batch data processing, document archiving, and electronic signatures, promoting the digitalization and efficiency of expropriation management. However, existing housing compensation calculation systems still have significant shortcomings in terms of result interpretability and user interactivity: (1) The current output of compensation results is mainly based on monetary values, and lacks the automatic generation of explanatory information such as the contribution of variables that affect the composition of the amount and the path of policy rules. (2) In a complex multi-stage calculation chain, it is difficult for the system to trace the specific effects of each input variable and its corresponding rules on the final amount, which makes it difficult for policymakers and those being levied to intuitively understand the calculation logic and weight settings, thus affecting the credibility of the results; (3) Without the ability to perform sensitivity analysis and visualize variable paths, system users find it difficult to grasp the sensitive driving factors of compensation results, and it is also difficult to detect data anomalies or policy adaptation flaws in a timely manner, which in turn affects the efficiency of negotiation and expropriation decisions. (4) In order to improve interpretability, some external studies have attempted to introduce complex proxy models or statistical sampling methods to achieve local sensitivity analysis. However, these methods rely on a large amount of external data, have long training cycles, and high integration costs. They are difficult to directly embed into engineering compensation calculation systems and are not conducive to rapid policy iteration. Summary of the Invention

[0003] In order to solve the above-mentioned technical problems, the present invention provides a dynamic calculation method and system for housing compensation based on land development patterns.

[0004] The technical solution of this invention is implemented as follows: a dynamic calculation method for housing compensation based on land development patterns, comprising: S1: Construct a compensation rule model based on local policy documents, define the input variables, basic parameters and calculation type for each compensation rule, and initialize the basic impact weight value, which reflects the relative importance of the rule in the policy system; S2: Connect to the housing basic data platform and manual data entry terminal to obtain and parse housing physical attributes, ownership information and location characteristics data, and generate a standardized set of input variables including building area, property status and construction year; S3: Based on the directed graph structure, a symbolic tracking engine is built, which abstracts the compensation calculation process into a node-edge relationship model. The nodes contain input variables, rule items and intermediate outputs, and the edges record the data flow and direction of action, thus establishing a dynamic influence weight propagation channel. S4: When calculating the compensation amount, the influence weight propagation strategy is dynamically adjusted according to the calculation type of each rule: the weight superposition mode is adopted for linear addition rules, the marginal effect amplification mode is adopted for multiplicative rules, and the weight inheritance mode is adopted for condition judgment rules, generating dynamic influence weight values ​​corresponding to each intermediate result. S5: Based on the propagation path data of the symbolic tracking engine, calculate the cumulative contribution intensity, path depth and frequency of action of the input variables, and generate a dynamic influence weight propagation topology map containing the variable influence score matrix. S6: Sort the input variables according to the cumulative contribution intensity index, and generate a structured analytical information package by combining the path depth and the frequency of action. The information package includes a ranking table of main driving factors, a pie chart of positive and negative contribution ratios, and a complete transmission chain flowchart. S7: Visualize and synchronously output the compensation amount and structured parsing information package, present an expandable transmission chain flowchart on the interactive interface, and mark the numerical change of each calculation node and the corresponding impact weight percentage.

[0005] The present invention also provides a dynamic calculation system for housing compensation under the land development model, which uses the above-mentioned dynamic calculation method for housing compensation under the land development model to perform dynamic calculation of housing compensation under the land development model.

[0006] The present invention provides a dynamic calculation method and system for housing compensation based on land development patterns, which has the following beneficial effects: (1) This invention achieves full traceability and structured analysis of the compensation amount generation process by constructing a dynamic influence weight propagation mechanism. Without relying on complex uncertainty modeling or global sensitivity analysis, it uses a lightweight symbolic tracking engine to abstract the compensation calculation process into a directed graph structure, accurately records the data dependencies between input variables, rule nodes and intermediate results, and dynamically adjusts the marginal contribution weight based on the rule type to achieve chain transmission and cumulative evaluation of the influence effect. The resulting analytical information, such as the ranking of main driving factors, the topology of the influence path and the proportion of positive and negative contributions, significantly enhances the logical transparency of the compensation conclusion and the credibility of policy implementation, effectively overcoming the drawbacks of the "black box operation" of traditional systems. (2) This invention introduces a hierarchical influence weight initialization and semantic propagation mechanism, which improves the system's collaborative analysis capability for multi-source heterogeneous compensation rules while ensuring computational efficiency. Unlike machine learning interpretation methods that require a large amount of training data, this invention operates entirely based on the existing policy rule system. By setting initial influence weights for each rule and dynamically adjusting the propagation intensity according to its calculation type (such as linear addition, coefficient product, conditional judgment), it achieves refined attribution analysis from the underlying input variables to the final amount. For example, when the property rights nature adjusts the basic amount in a multiplicative form, it automatically identifies the amplification effect of this operation and improves the overall influence score of related variables, thereby more realistically reflecting their role in actual decision-making. This mechanism does not require additional parameter tuning or external simulation tools, has good engineering adaptability and policy flexibility, and can be quickly migrated and deployed under compensation standards in different regions and at different times, significantly improving the system's maintainability and cross-scenario applicability. (3) This invention significantly improves the efficiency of users' understanding of compensation results and the interactive experience by generating structured and interpretable output packages and deeply integrating them with a visual interface. It effectively promotes the consensus reached during the relocation process and can intuitively display key transmission chains such as "building area → basic amount → building age depreciation deduction → property type increase" in the form of flowcharts, etc. It marks the numerical changes of each link and their relative impact weights, so that non-professional users can clearly grasp the logic of compensation composition. Compared with traditional report-style output, this interpretation mode based on the impact path is more readable and persuasive, and significantly improves the communication quality between the expropriating party and the expropriated party. In addition, the entire interpretation generation process is embedded in the original calculation process, adding only a very low additional calculation overhead, avoiding the problem of response delay or surge in resource consumption caused by the introduction of complex algorithms, and ensuring the stable availability of the interpretation function in large-scale application scenarios. Attached Figure Description

[0007] Figure 1 This is a flowchart of a dynamic calculation method for housing compensation based on land development patterns according to the present invention. Figure 2This is a sub-flowchart of a dynamic calculation method for housing compensation based on land development patterns according to the present invention. Figure 3 This is another sub-flowchart of a dynamic calculation method for housing compensation based on land development patterns according to the present invention. Detailed Implementation

[0008] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0009] The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0010] like Figure 1 As shown, this invention provides a dynamic calculation method and system for housing compensation based on land development patterns, specifically including: S1: Construct a compensation rule model based on local policy documents, define the input variables, basic parameters and calculation type for each compensation rule, and initialize the basic impact weight value, which reflects the relative importance of the rule in the policy system; S2: Connect to the housing basic data platform and manual data entry terminal to obtain and parse housing physical attributes, ownership information and location characteristics data, and generate a standardized set of input variables including building area, property status and construction year; S3: Based on the directed graph structure, a symbolic tracking engine is built, which abstracts the compensation calculation process into a node-edge relationship model. The nodes contain input variables, rule items and intermediate outputs, and the edges record the data flow and direction of action, thus establishing a dynamic influence weight propagation channel. S4: When calculating the compensation amount, the influence weight propagation strategy is dynamically adjusted according to the calculation type of each rule: the weight superposition mode is adopted for linear addition rules, the marginal effect amplification mode is adopted for multiplicative rules, and the weight inheritance mode is adopted for condition judgment rules, generating dynamic influence weight values ​​corresponding to each intermediate result. S5: Based on the propagation path data of the symbolic tracking engine, calculate the cumulative contribution intensity, path depth and frequency of action of the input variables, and generate a dynamic influence weight propagation topology map containing the variable influence score matrix. S6: Sort the input variables according to the cumulative contribution intensity index, and generate a structured analytical information package by combining the path depth and the frequency of action. The information package includes a ranking table of main driving factors, a pie chart of positive and negative contribution ratios, and a complete transmission chain flowchart. S7: Visualize and synchronously output the compensation amount and structured parsing information package, present an expandable transmission chain flowchart on the interactive interface, and mark the numerical change of each calculation node and the corresponding impact weight percentage.

[0011] Step S1: Construct a compensation rule model based on local policy documents, define the input variables, basic parameters, and calculation type for each compensation rule, and initialize the basic influence weight value, which reflects the relative importance of the rule in the policy system. Specifically, this includes: S1.1: Perform structured analysis on local policy documents to extract compensation rule items and their applicable conditions, so as to obtain a set of rule items related to housing compensation in the policy text; Using the original electronic text or scanned image files of local policy documents as input data, and employing a text parsing engine (parameters: supports OCR recognition, Chinese word segmentation, and policy terminology dictionary), the digitization and semantic segmentation of policy documents are achieved. Furthermore, by using a rule entry extraction algorithm (parameters: dependent on a specific compensation terminology glossary and clause numbering pattern), the compensation-related clauses in the digital policy text are identified, and a preliminary set of rules containing clause numbers, clause titles, and original text content is obtained. Furthermore, by using the applicable condition parsing method (parameter: dependency parsing model based on condition sentence recognition), the condition qualifiers, applicable scope and triggering conditions of each rule text are extracted, and structured data binding rule items and condition items is generated. Furthermore, by using a compensation-theme clustering algorithm (parameters: similarity calculation method is cosine similarity, threshold is set according to historical policy clustering results), the rule set is grouped to distinguish different compensation types (such as area compensation, property rights modification, depreciation deduction) and maintain conditional correlation within each category; By using structured parsing and condition binding, the textual policy clauses from the previous step are transformed into a set of rule items that can be used for subsequent field mapping and parameter extraction, thus preparing the policy data foundation for the rule model construction process.

[0012] S1.2: Based on the extracted set of rule items, perform field mapping processing on each rule to identify its dependent input variables, including standardized housing attribute fields such as building area, property status, and construction year; Based on the set of rule items output in step S1.1, a field dependency resolution algorithm (parameters: set of rule items, dictionary of standardized house attribute fields) is used to resolve the mapping relationship between rule description fields and system standardized fields. Furthermore, through semantic matching and field name normalization processing algorithms (parameters: rule text keywords, attribute field alias library), the correspondence between non-standardized terms in the rule description and field names in the unified system is matched, and preliminary field matching table data is obtained; Furthermore, a multi-source field conflict detection algorithm (parameters: preliminary field matching table, historical rule mapping library) is adopted to identify and prioritize field conflicts from multiple tables in the rule items, and generate a set of field matching results after conflict resolution. Furthermore, a field type validation algorithm (parameters: field matching result set, attribute type standard table) is adopted to verify the consistency between the input variable type and data format, and to generate a standardized field mapping set that meets the requirements of the calculation module. By using the field dependency topology construction algorithm (parameter: standardized field mapping set), the results of the previous step are transformed into node-field binding relationship data, realizing the visualization and programmable representation of the dependency structure of rule items on input variables; For example, in the policy document parsing scenario of urban renewal land development projects, the input consists of three rule items extracted in step S1.1: "Building Area Compensation," "Property Status Adjustment," and "Building Age Depreciation," as well as the system's standardized attribute field dictionary {Building Area, Property Status, Construction Year, Location Code}. When using the field dependency parsing algorithm, the "House Area" field in the rule item "Building Area Compensation" is mapped to the "Building Area" field in the standardized dictionary through a semantic matching algorithm. The "House Property Category" field in the rule item "Property Status Adjustment" is mapped to the standard field "Property Status," and the "Age Range" field in the rule item "Building Age Depreciation" is mapped to the standard field "Construction Year." During multi-source field conflict detection, if "House Area" exists in both "Surveyed Area" and "Property Certificate Area," the surveyed area is determined to be the primary field based on rule priority, the property certificate area mapping is deleted, and a conflict-resolved field matching result set is generated. The field type verification algorithm verifies that "Building Area" is a floating-point number in square meters, "Property Status" is an enumeration type, and "Construction Year" is a range type. The final field-dependent topology construction algorithm produces a mapping set: {Rule Item 1 → Building Area, Rule Item 2 → Property Status, Rule Item 3 → Construction Year}, and generates a structured binding relationship for the simulation calculation module call, realizing the standardized dependency identification of the rule model on the input variables, which significantly improves the accuracy and scalability of subsequent compensation amount calculation steps; S1.3: Based on the description of the rule item and the semantic analysis results of the policy provisions, extract the basic parameter values, including compensation unit price, depreciation coefficient, and upward floating ratio, and generate a parameter configuration table; The input conditions are based on a standardized set of housing attribute fields after field mapping, including variables such as building area, property status, and construction year, combined with a set of compensation rule items obtained from policy text parsing as the processing object. A policy semantic parsing algorithm based on word segmentation and dependency parsing (parameters: rule description text, stop word list, domain dictionary) is adopted to automatically identify numerical policy parameters involved in each rule description; Furthermore, by using a regularized parameter extraction algorithm (parameters: set of numerical patterns, unit mapping table), the numerical patterns in the rule text are matched to obtain the original parameter values ​​and unit information corresponding to fields such as building area, property status, building age and depreciation. Furthermore, through a unit normalization processing algorithm (parameters: area unit standard, currency unit standard, and proportional coefficient standard), the unit conversion of the original parameter values ​​is achieved. For example, yuan / square foot is converted to yuan / square meter, and percentage proportions are converted to coefficient values ​​expressed as decimals. Furthermore, a policy parameter conflict detection algorithm (parameters: parameter source identifier, policy effective time, applicable area code) is adopted to achieve version optimization and conflict resolution for policy parameters that have multiple versions and are applicable to multiple regions, and to obtain a unique set of basic parameter values ​​that are effective in the current project area; Furthermore, through parameter classification and identifier generation algorithms (parameters: parameter value set, field mapping relationship table), the types of parameters such as compensation unit price, depreciation coefficient, and floating ratio are classified, and a unique parameter identifier and retrieval index are generated for each type of parameter to support quick reference in subsequent calculation processes; Through the above algorithm chain, the rule mapping result of the previous step is transformed into a structured basic parameter configuration table, realizing the standardization and configurability of the compensation rule model in the parameter dimension. For example, in an urban renewal project, regarding the rule of "basic compensation based on building area × unit price," the policy text identifies the compensation unit price as "calculated at 3500 yuan per square meter," the depreciation factor as "30% depreciation for houses built before 1990," and the upward adjustment ratio as "15% upward adjustment for compensation for commercial housing." A policy semantic parsing algorithm is used to analyze the numerical values ​​in this sentence and match the corresponding fields, obtaining the original parameter value set [3500 yuan / ㎡, 30%, 15%]. After unit normalization, the percentages are converted to decimal coefficients of 0.30 and 0.15, while the unit price remains unchanged at yuan / ㎡. Conflict detection confirms that only this version of the policy parameters exists in the project area, generating the following basic parameter configuration table: Compensation unit price = 3500 (yuan / ㎡), depreciation factor = 0.30, upward adjustment ratio = 0.15. The compensation amount calculation formula in this environment can be expressed as: in, For the amount of compensation, The variable represents the building area (㎡). To compensate for the unit price (yuan / ㎡) This is the depreciation factor (decimal). This represents the upward adjustment percentage (decimal). In actual implementation, for a commercial apartment with a building area of ​​120㎡, the calculation process using the formula is as follows: Calculate the compensation amount The result is output to the calculation node corresponding to the parameter configuration table. This result can be directly accessed within the system, enabling the accurate application and efficient transmission of policy parameters in the compensation calculation process for specific projects. S1.4: Based on the logical structure of the rule items, the rule type classification algorithm is used to identify the calculation type of each rule to determine whether it is linear addition, multiplication amplification, or conditional judgment, and generate a rule calculation type identifier; S1.5: Based on the authority level, scope of application, and historical implementation data of policy provisions, perform basic impact weight assignment on each rule item to generate a basic impact weight configuration matrix, which is used for initial weight allocation in the subsequent impact propagation process; S1.6: Structure the binding of rule items, input variables, basic parameters, calculation types and basic influence weights to generate a configuration data package for the compensation rule model, which can be called and executed by the simulation calculation module; Step S2: Connect to the basic housing data platform and the manual data entry terminal to obtain and parse the physical attributes, ownership information, and location characteristics of the housing, generating a standardized set of input variables including building area, property status, and construction year. Specifically, this includes: S2.1: Based on the housing basic data platform interface protocol, obtain surveying and mapping data, property registration information and geographic information system (GIS) location data to construct the original dataset of housing physical attributes and ownership information; In the data access stage of the main step S2, the sub-step S2.1 is used to construct the original dataset of the physical attributes and ownership information of the house. Its logical position is in the data collection stage before standardization, and it undertakes the function of accurately introducing external surveying data, property registration information and GIS location data into the internal processing chain. The interface protocol call method (parameters: housing basic data platform API endpoint, authentication key, data request structure) is used to achieve secure communication with the surveying and mapping data platform and obtain the two-dimensional plane coordinates, building outline polygon, and accurate building area measurement value of the house. Furthermore, by using the property registration information acquisition method (parameters: real estate registration system interface, field mapping rules, certificate number retrieval conditions), the batch extraction of fields such as property type, registration date, and right holder information is realized, and a housing ownership information dataset is obtained; Furthermore, a GIS spatial data access algorithm (parameters: GIS service port, latitude and longitude resolution, spatial reference system coding EPSG standard) is adopted to capture the latitude and longitude coordinates of the building's location and generate location code and area number data by combining the administrative division matching table; Furthermore, by utilizing data synchronization and timestamp calibration algorithms (parameters: platform server time, data collection timestamp field, allowable deviation threshold), time consistency verification of the three types of data sources is achieved, ensuring that the surveying, property rights and location information of the same house record belongs to the same time snapshot; By integrating and processing multi-source data, surveying and mapping data, property registration information and GIS location data are bound to key-value pairs according to the unique identification code of the house, and transformed into a structured raw dataset, so as to achieve traceability and accuracy from the raw input to the standardized processing stage. Exemplarily, in a certain urban renewal project, the API endpoint of the housing basic data platform is set to "http: / / dataplatform.city / housing", the authentication key is a 32-bit hash string, and the data request structure contains fields {house_id, survey_area, outline_polygon_wkt}. By calling this endpoint, the surveyed area of a certain house is obtained as 128.54㎡, and the coordinate polygon covers 5 vertices. In the real estate registration system, according to the certificate number "YueFangDengJi2023000123", the property type is extracted as "commercial housing" and the registration date is "2008-03-15". In the GIS platform, with the longitude and latitude (23.1291, 113.2644) as the input, the administrative division code "440104" and the area number "P-12" are parsed. Through the timestamp calibration algorithm, the survey data time is 1678901234 seconds, the property data time is 1678901200 seconds, and the GIS data time is 1678901250 seconds. The three match successfully within the allowable deviation threshold of 60 seconds. Finally, the original dataset is integrated: {house_id: H123456, survey_area: 128.54㎡, property_type: commercial housing, reg_date: 2008-03-15, district_code: 440104, zone_id: P-12}, and this dataset enters the S2.2 standardization processing stage, achieving a significant improvement in the accuracy and consistency of the measured input variables; S2.2: Perform format standardization processing on the building area field in the original dataset, and use a unified unit conversion algorithm to convert area data from different sources into square meters (㎡) unit representation to obtain a standardized building area variable; S2.3: Execute a property status recognition algorithm based on the property registration information field, and extract house property type labels, including categories such as 'commercial housing', 'collective property', or 'unlicensed house', etc., to generate a standardized property status variable; S2.4: Perform timestamp parsing and classification processing on the house construction year field, and use a decade interval mapping algorithm to convert specific years into standardized construction year labels such as 'before 1980', '1980 - 1999', '2000 - 2010', or 'after 2010'; S2.5: Execute a spatial coordinate parsing and regional code mapping algorithm based on the geographical information system (GIS) location data, and convert longitude and latitude information into corresponding administrative divisions and area numbers to generate a standardized location feature variable; S2.6: Standardize the building area, property status, construction year and location characteristics variables by merging and structuring them to generate a unified set of housing input variables for subsequent use by the compensation rule model and influence weight propagation.

[0013] like Figure 2 As shown, step S3 involves constructing a symbolic tracking engine based on a directed graph structure, abstracting the compensation calculation process into a node-edge relationship model. Nodes contain input variables, rule items, and intermediate outputs, while edges record the data flow direction and the direction of action, establishing a dynamic influence weight propagation channel. Specifically, this includes: S3.1: Based on the rule items, input variables and intermediate calculation results in the compensation rule model, define a set of nodes in a directed graph. The nodes include basic house attribute variable nodes, compensation rule nodes and intermediate amount output nodes to form the topological structure basis of the compensation calculation path. In the initial stage of loading the compensation rule model into the simulation calculation module, a node definition algorithm (parameters: rule item identifier, input variable set, intermediate result data structure) is used to initialize the configuration of the directed graph node set. Through this algorithm, the standardized basic house attribute variables obtained from step S2 are bound at the field level to the compensation rule model instance generated from step S1, ensuring that each house attribute variable node has a unique identifier and type classification information in the directed graph. A node type hierarchical mapping method (parameters: node type encoding table, attribute variable list, calculation rule list) is adopted to classify the types of basic housing attribute variable nodes, compensation rule nodes, and intermediate amount output nodes, and a hierarchical index structure is established in memory to support subsequent dynamic retrieval and traversal of the graph structure. Furthermore, by attaching a strategy to node attributes (parameters: policy scope label, calculation type identifier, initial weight value), metadata is injected into each compensation rule node, simultaneously embedding the rule's applicable conditions, calculation operation type, and basic impact weight into the node data structure, so as to facilitate the initiation and execution of subsequent impact weight propagation channels. An intermediate result node generation algorithm (parameters: rule dependency path, logic execution order) is adopted to register the intermediate output values ​​generated in each stage of the compensation amount calculation process as intermediate amount output nodes, and record the source rule identifier and dependency variable index of the result in the node, so as to provide complete traceability information for the subsequent establishment of node-edge relationship; By performing a node set consistency check (parameters: node identifier uniqueness rule, data type matching rule), the node set generated in the previous step is compared with the original definition of the compensation rule model to ensure that all input variables, rule items and intermediate outputs are accurately mapped in the directed graph node set, avoiding the breakage of the propagation path due to missing nodes or incorrect types during the calculation process. By structurally binding node sets, the definition results of node sets are transformed into a topological structure basis that can be recognized by the symbolic tracing engine, thereby providing technical support for data dependency modeling, action type marking, runtime extension, and influence weight propagation in subsequent steps S3.2 to S3.6. For example, in an urban renewal project, the building area variable node is configured with type code A01 and attribute value of 93.5㎡; the property status variable node is configured with type code A02 and attribute value of "commercial housing"; and the construction year variable node is configured with type code A03 and attribute value of "2000-2010". The compensation rule node "Basic Amount Calculation" is configured with type code R11, calculation type identifier is linear addition, and basic influence weight is 0.35; the compensation rule node "Building Age Depreciation Deduction" is configured with type code R21, calculation type identifier is multiplicative amplification, and basic influence weight is 0.15. The intermediate amount output node "O_BASE" records the source rule as R11, the dependent variable set is {building area, unit price}, and the node attribute includes the basic amount calculated initially. The above set of nodes is defined and stored in the memory structure of the symbolic tracing engine through a hierarchical indexing method. This ensures that the directed edge relationships between nodes are established in the order of rule calls in step S3.2, so that the dependency between the building area node and the basic amount node can be accurately parsed and transmitted in the influence weight propagation channel, and finally realizes the full traceability modeling of the compensation amount composition path. S3.2: Based on the data dependencies between compensation rules, construct a set of edges in a directed graph, where each edge represents the data flow direction and action direction from one node to another, used to record the data propagation path during the execution of compensation rules; S3.3: Based on the calculation type of the compensation rule, assign an action type label to each edge. The action type includes linear addition, multiplicative amplification, and conditional inheritance, which are used for the differentiated processing of the subsequent dynamic influence weight propagation strategy. S3.4: Construct the runtime graph structure of the symbolic tracing engine, mapping all node and edge relationships in the compensation amount calculation process to a graph data structure in memory. The graph structure supports dynamic node expansion and edge weight updates to track the compensation calculation path in real time. During the initialization phase of the symbolic tracing engine, a node-edge mapping method (parameters: node set, edge set, action type label) is used to store all nodes and their associated edge relationships defined during the compensation amount calculation process into a runtime memory graph data structure, thereby achieving an abstract representation of the overall calculation path. Furthermore, a dynamic node management algorithm (parameters: node type identifier, topology index) is used to provide a real-time node expansion mechanism for the runtime graph data structure. Based on newly introduced house attribute variables or policy rule items, corresponding new nodes are automatically generated and inserted into the appropriate positions in the topology. Furthermore, an edge weight update algorithm (parameters: initial impact weight value, rule calculation type, edge action type) is adopted to adjust the weight values ​​of existing edges in real time, ensuring that weight changes can be reflected in the data flow model in real time during the calculation of compensation amount, and synchronously affect the input state of subsequent calculation nodes. Furthermore, through a bidirectional index mapping method (parameters: unique node identifier, unique edge identifier), the fast retrieval and positioning of nodes and edges in the memory graph structure are realized, supporting the symbolic tracing engine to efficiently access and dynamically update any path segment; Furthermore, by combining the node state synchronization algorithm (parameters: calculation progress identifier, data dependency matrix), a state refresh operation is performed on each node in the runtime graph structure. New intermediate amount values ​​and incoming dynamic influence weights are written in real time as the calculation process progresses, so as to achieve strict synchronization between the calculation process and the propagation path. Through the above graph structure construction and update algorithm, the static topology structure generated in the previous step is transformed into runtime graph data that can support dynamic expansion of nodes and real-time update of edge weights, so as to achieve the expected technical effect of full state recording and visual tracking of the compensation amount calculation process. For example, during the calculation of housing compensation for a land development project, the initial runtime graph structure contains a total of 120 nodes, including 40 housing input variable nodes, 60 compensation rule nodes, and 20 intermediate output nodes. During the calculation, when a newly introduced "seismic resistance level bonus" rule is detected, the node management algorithm immediately generates the corresponding rule node N121 and establishes a directed edge E200 between it and the "basic amount node" in the topology, with an initial weight value set to 0.05. Subsequently, the edge weight update algorithm updates the weight value of the edge according to the rule's multiplication coefficient of 1.2. The system then synchronizes the changes to the input weight set of the affected downstream nodes. Using a bidirectional index mapping method, the system can locate node N50 and all its incoming and outgoing edges within an average retrieval time of 0.002 seconds, achieving fast access. During node state synchronization, the output amount of the new rule node is changed from... Multiplying the original by the coefficient 1.2 yields... The system updates the status of connected intermediate output nodes in real time, thereby ensuring that the runtime graph structure of the entire compensation path has significantly improved real-time traceability and interpretability after the introduction of new rules. S3.5: Initialize the dynamic influence weight propagation channel in the graph structure, and assign an initial influence weight value to each node. The initial influence weight value is assigned based on the basic influence weight value of the rule in the policy system, and serves as the starting point for subsequent propagation calculations. S3.6: Based on the execution order and data flow of the compensation rules, dynamically update the node state and edge weight values ​​in the graph structure to achieve synchronous modeling of the compensation calculation process and the influence propagation path, and generate a complete compensation amount influence propagation graph; Based on the execution order of the compensation rules and the data flow of the input variables, a directed graph node state update algorithm (parameters: node type identifier, initial influence weight, execution time index) is adopted to realize the dynamic state maintenance of nodes during the calculation process. Furthermore, by using a dynamic adjustment method for edge weights (parameters: edge action type label, rule calculation coefficient, propagation attenuation factor), the edge weight values ​​are responsively modified to the rule execution process, and an updated list of edge weights is obtained. Furthermore, a synchronous modeling triggering mechanism (parameters: rule execution event queue, node dependency matrix) is adopted to achieve temporal consistency processing of node state updates and edge weight adjustments, and to generate graph structure snapshot data at each time step; Furthermore, by using a graph structure snapshot merging algorithm (parameters: snapshot time series, node weight difference threshold), the graph structure at different time points is synthesized to obtain the change trajectory mapping matrix; Furthermore, by using the influence propagation path generation method (parameters: node state matrix, edge weight matrix), the full path relationship for compensation amount calculation is extracted from the synthetic graph structure, and a complete compensation amount influence propagation graph is formed. The above dynamic update algorithm transforms the runtime graph structure result of the previous step into full-path propagation data containing real-time node states and edge weights, achieving synchronous modeling of the compensation calculation process and the impact propagation path. For example, in an urban renewal project, there are 300 basic housing attribute variable nodes, 50 compensation rule nodes, and 20 intermediate amount nodes. In the initial influence weight configuration matrix, the initial weight of rule nodes with higher policy authority is set to 0.8, and that of ordinary rule nodes is 0.5. The rule execution order is defined by an event queue with a length of 370 events. Edge action type labels are divided into three categories: linear addition, multiplicative amplification, and conditional inheritance, with the multiplicative amplification rule coefficient ranging from 1.1 to 1.5. The dynamic adjustment formula for edge weights is: in, The edge weight value. Index for the current execution time. Calculate coefficients for the rule. When the rule type is multiplicative amplification and... When the edge weight is 1.3, if the edge weight in the previous step was 0.5, then the updated edge weight will be... The node status update employs a dual maintenance mechanism of Boolean status identifiers and numerical weights to ensure the traceability of the path generation process. The resulting compensation amount impact propagation graph contains 370 valid paths, and the node weights maintain a controllable fluctuation range throughout the execution process, significantly improving the system's ability to interpret the compensation results and demonstrate the paths to success.

[0014] like Figure 3 As shown, in step S4, when calculating the compensation amount, the influence weight propagation strategy is dynamically adjusted according to the calculation type of each rule: a weight superposition mode is used for linear addition rules, a marginal effect amplification mode is used for multiplicative rules, and a weight inheritance mode is used for conditional judgment rules, generating dynamic influence weight values ​​corresponding to each intermediate result. Specifically, this includes: S4.1: Based on the rule calculation type in the predefined compensation rule model, classify and identify each rule. The calculation type includes linear addition, multiplication, and condition judgment to determine the corresponding influence weight propagation mode. Based on the configuration data package of the node-edge relationship model and compensation rule model established by the symbolic tracing engine, a rule type identification algorithm (input parameters: rule node identifier, calculation type label set) is used to initially determine the calculation type of each compensation rule. The identification algorithm generates a calculation type classification label for the rule by parsing the calculation type identifier field in the rule model and comparing it with a predefined type mapping table. Furthermore, by employing a computational semantic parsing method (parameters: regular expression, operator precedence table), the operator structure of the regular expression is parsed, yielding a set of syntactic features for linear addition, multiplication, and conditional rules. This process decomposes the computational expression into operator nodes and operand nodes, forming an operator tree structure for subsequent propagation pattern binding. Furthermore, by using a rule dependency parameter analysis algorithm (parameters: set of input variables, dependency matrix), the structured extraction of the set of input variables for each rule is achieved, and a dependency mapping table from variables to rules is generated, providing a basis for the correlation of variable influence for optimizing the propagation strategy; Furthermore, a rule type matching and propagation pattern binding algorithm (parameters: calculation type label, propagation pattern library) is adopted to accurately map the identified calculation type to the corresponding influence weight propagation pattern, including the linear addition weight superposition pattern, the multiplicative marginal effect amplification pattern, and the conditional judgment weight inheritance pattern. This binding process ensures the consistency of weight update strategies for different calculation types during the propagation process; Through the above algorithm processing method, the rule node information in the previous symbolic tracking engine is transformed into rule classification results containing calculation type labels and propagation mode binding, thereby achieving the effect of mode initialization technology that dynamically affects weight propagation. For example, in an urban renewal project, the rule model includes a rule that reads "basic area multiplied by unit price and adjusted according to property type". The rule expression is "building area × unit price × property adjustment coefficient". In the input data, the building area variable is 120, the unit price variable is 5500, and the property adjustment coefficient is 1.05. The rule expression is parsed using a rule type recognition algorithm. The parsing result of the operator structure is two multiplication operators and three operand nodes. According to the type mapping table, this rule is identified as a multiplicative rule, and the propagation mode is bound to the marginal effect amplification mode. In the dependency parameter analysis, the building area, unit price, and property status variables are extracted into the dependency mapping table, and their input positions in the rule are marked respectively. After mapping to the pattern library, the rule propagation strategy is as follows: when calculating the influence weight of the current node, the weight values ​​of each input variable in the previous node are combined with the multiplication coefficient to calculate the amplified node weight, as shown in the following formula: ,in Output node weights for the rules. Input weights for building area. Input a weight for the unit price. Input weights for property ownership status. In this example, it is assumed that the weight of the previous node's building area is 0.42, the weight of the unit price is 0.37, and the weight of the property ownership status is 0.15. The calculation result is as follows: The application of this propagation model enables the accurate quantification of the marginal effect of the multiplication rule during the weight transfer process, significantly improving the interpretability and stability of the final compensation amount composition path; S4.2: Perform weighted summation processing on the linear additive rule. Based on the original influence weight values ​​of the input variables and the linear additive coefficient of the rule, use a weighted summation algorithm to calculate the comprehensive influence weight value of the rule's output node, so as to reflect the combined effect of multiple variables on the final compensation amount. S4.3: Perform marginal effect amplification mode processing on multiplicative rules. Utilize the current influence weight value of the input variable and the multiplication coefficient of the rule to generate the amplified influence weight value of the rule's output node based on the multiplicative synthesis algorithm, so as to reflect the amplification effect of multiplicative rules on the influence of variables in the compensation amount composition path. S4.4: Perform weight inheritance mode processing on the condition judgment rule. Based on the original influence weight value of the input variable and the logical branch result of the condition judgment rule, the branch weight inheritance algorithm is used to determine the inherited influence weight value of the rule output node, so as to realize the logical selection and transmission of the influence path of the variable by the condition rule. S4.5: The output influence weight value of each rule node is used as the input influence weight value of the next calculation node. Based on the directed graph structure in the symbolic tracking engine, the weight is propagated layer by layer to form a dynamic influence weight propagation sequence of each variable in the compensation amount composition path. Receive the dynamic influence weight values ​​of each rule node output by S4.2, S4.3, and S4.4, and use them as the input weight set for the next computation node in the directed graph structure of the symbolic tracing engine; A node-by-node weight transfer algorithm (parameters: current node output weight value, edge action type label) is used to realize the influence weight transfer process from the rule node to the next calculation node; Furthermore, by using a directed graph traversal method (parameters: starting node ID, topology path set), the weight values ​​are propagated layer by layer along the compensation calculation path, and the cumulative input weight vector of the target node is automatically updated. Furthermore, a weight fusion calculation algorithm (parameters: cumulative input weight vector of the target node, weight superposition coefficient matrix) is adopted to synthesize the weight values ​​from multiple upstream nodes into a single current weight value of the target node, and generate the time series weight record of the node during the dynamic propagation process; Furthermore, by using a weight normalization method (parameters: current weight value of the node, total weight within the path), the weight values ​​of each node are expressed proportionally to eliminate the interference of different path lengths and weight scales on subsequent analysis; By using a dynamic weight recording and caching mechanism, the updated node weight values ​​are written into the propagation sequence storage module of the symbolic tracking engine, enabling full traceability recording of the dynamic influence weight propagation sequence of each variable in the compensation amount composition path. For example, in the calculation of housing compensation under a certain land development model, the initial influence weight value of the building area variable is set to... The initial influence weight of the property rights state variable is set to After applying the linear addition rule of "building area × unit price", the weighted value of the building area output is calculated using a weighted summation algorithm as follows: This is then passed to the next node, the "Property Rights Modification" multiplicative rule. The multiplier coefficient for the multiplicative rule is... The new output weight values ​​are generated using a multiplicative synthesis algorithm. This is then passed to the "Building Age Depreciation Deduction" condition rule branch. In this condition rule, the weight inheritance algorithm directly passes the weight value of the building area along the branches that meet the conditions. To the deduction operation node. Property state variables, within the same multiplicative rule, are determined by initial weights. Obtained by multiplicative synthesis The conditional branch is inherited by the deduction node. Throughout the complete calculation path, the symbolic tracing engine records the propagation sequence of the building area weight values ​​as follows: The property rights state weight sequence is After normalization, the weighting of the building area of ​​the deducted nodes is as follows: The weighting of property rights status is as follows: This enables the quantitative recording and verification of the dynamic influence weights of the two variables in the compensation amount composition path. S4.6: Based on the dynamic influence weight propagation sequence, generate a set of dynamic influence weight values ​​for each intermediate calculation node, which serves as the input basis for the subsequent construction of the variable influence scoring matrix and the generation of the influence path topology map, so as to support the multidimensional interpretable output of the compensation amount. Based on the dynamic influence weight propagation sequence output by the symbolic tracking engine, the node state parsing method (parameters: node identifier, current weight value, propagation path index) is used to extract the weight and bind the identifier of each intermediate calculation node in the sequence. Furthermore, through a weight aggregation algorithm (parameters: rule type identifier, input variable weight set, node execution coefficient), the weight values ​​of the same node under multiple path input scenarios are synthesized, and a set of node comprehensive dynamic influence weight values ​​is obtained. Furthermore, a weight normalization processing method (parameters: set of node weight values, normalization ratio coefficient) is adopted to achieve consistent adjustment of the weight dimensions of different nodes and generate a standardized dataset of intermediate node influence weight values. Furthermore, through a weight matrix construction algorithm (parameters: standardized weight set, node topology index), the set of weight values ​​dynamically influenced by intermediate nodes is transformed into a matrix format, and a node-weight comparison matrix structure is generated. By using matrix index mapping, the node-weight comparison matrix from the previous step is transformed into the input data for generating the variable influence scoring matrix and the influence path topology map, thereby achieving a multi-dimensional interpretive output of the compensation amount. For example, in a certain urban renewal project, based on the dynamic impact weight propagation sequence of the pre-compensation rules, node A (building area × unit price) receives weight values ​​of 0.35, 0.40, and 0.30 in the three propagation paths, respectively. A weight aggregation algorithm is used, with the path weighted average formula as follows: The overall weight value of node A is 0.35. After normalization (the normalization ratio is taken as the maximum node weight value of 0.45), the weight transformation formula for node A is as follows: The normalized result is 0.777. The weights of node A and the remaining nodes are combined to form a node-weight comparison matrix, which is then mapped to the variable influence scoring matrix input. This matrix is ​​used in subsequent steps to generate an influence path topology map, clearly indicating the cumulative contribution and hierarchical effect of the building area variable in multiple monetary contribution chains, thereby significantly improving the readability of the explanation output and the decision support effect.

[0015] Step S5: Based on the propagation path data from the symbolic tracing engine, calculate the cumulative contribution strength, path depth, and frequency of action of the input variables, and generate a dynamic influence weight propagation topology graph containing a variable influence scoring matrix. Specifically, this includes: S5.1: Perform topological analysis on the node-edge relationship model output by the symbolic tracing engine, extract the set of data flow paths between input variables and rule items, and construct a complete set of variable propagation paths; For the node-edge relationship model output by the symbolic tracing engine, a topology parsing algorithm (parameters: node type identifier, edge action type label, data flow direction attribute) is used to realize the hierarchical deconstruction of the compensation calculation graph structure in order to identify the direct or indirect relationship between input variable nodes and rule item nodes. Furthermore, by using the path enumeration algorithm (parameters: starting node set = all input variable nodes, ending node set = final compensation amount output node), a full path traversal starting from each input variable is achieved, and a path data set containing node sequences and edge action types is obtained; Furthermore, by using a data flow filtering algorithm (parameters: edge direction = from input variable to rule item or intermediate output node, function type filtering condition = linear addition, multiplicative amplification, conditional inheritance), irrelevant path branches are eliminated, and a set of effective propagation paths that conform to the compensation amount composition logic is generated; Furthermore, a path uniqueness processing algorithm (parameter: path node sequence hash value) is adopted to remove duplicate paths and obtain a complete and non-redundant set of variable propagation paths to support subsequent path depth and contribution intensity calculations. Furthermore, by combining edge action type labels and rule calculation types, and using the path structure annotation method (parameters: node number, edge type symbol, initial node weight), the calculation attribute markers and weight information of each node are embedded in the variable propagation path set to ensure the traceability of subsequent calculations; By using topology analysis and path filtering to uniquely process the output of the symbolic tracing engine in the previous step, the results are transformed into structured data containing a set of variable propagation paths, thus enabling a complete description and computable modeling of the data flow between input variables and rule items. For example, in a compensation rule model containing three types of input variables—building area, property status, and construction year—the symbolic tracking engine outputs a node-edge relationship model. This model contains 12 nodes and 15 directed edges. Node types include input variable nodes, rule nodes, and intermediate output nodes. Edge action types include linear addition, multiplicative amplification, and conditional inheritance. A topology parsing algorithm reads the node type identifier and edge action direction, identifying a direct linear addition relationship between the building area node and the compensation unit price rule item; a multiplicative amplification relationship between the property status node and the property rights correction rule item; and a conditional inheritance relationship between the construction year node and the depreciation coefficient rule item. A path enumeration algorithm starts at the building area node and terminates at the project total cost node, enumerating the path "building area → compensation unit price rule → basic amount output → property rights correction rule → corrected amount → depreciation rule → final compensation amount," and labeling each edge with its action type. A data flow filtering algorithm removes intermediate branches that do not affect the final amount, obtaining a unique set of valid paths. The path uniqueness processing algorithm calculates the hash value of the path sequence, removes redundant sequences, and finally generates three complete variable propagation paths. Combined with initial node weights (e.g., building area weight 0.5, property status weight 0.3, construction year weight 0.2) and edge action type labels, structured data of the variable propagation path set is formed for subsequent path depth and contribution intensity calculations. In this scenario, this effectively improves the traceability and structural clarity of the causal chain between input variables and compensation amounts. S5.2: Based on each path in the propagation path set, calculate the path depth value of each input variable on that path. The path depth value represents the level of the variable's role in the compensation calculation process, so as to reflect the degree of indirectness of its impact on the final amount. S5.3: Perform dynamic weight accumulation calculation on the nodes in each propagation path, and combine the dynamic influence weight value of the rule nodes to calculate the cumulative contribution intensity of the input variable on the entire path, so as to quantify the overall influence of the variable on the compensation amount. S5.4: Based on the frequency of occurrence of input variables in the propagation path set, count the frequency of their role in different paths to evaluate the activity level and global influence of the variable in the entire compensation calculation model; S5.5: Based on the three indicators of cumulative contribution intensity, path depth and frequency of action, construct the input variable influence scoring matrix and normalize each variable to generate standardized variable influence scoring results. Based on the three quantitative indicators of cumulative contribution intensity, path depth and frequency of action obtained from the previous steps, a vectorized construction method (parameter: indicator set data structure) is used to initialize and structure the input variable influence scoring matrix. Furthermore, through a normalization algorithm (parameters: minimum value, maximum value, normalization interval [0,1]), a unified scale conversion of indicators with different dimensions is achieved, and the standardized cumulative contribution intensity value, standardized path depth value, and standardized action frequency value of each input variable are generated. Furthermore, through a weighted combination algorithm (parameter: weight coefficient vector) This allows for the comprehensive scoring calculation of three types of standardized indicators, and the comprehensive global impact score of each input variable is obtained. The weighting coefficients are set based on policy adaptability and historical measurement accuracy. Furthermore, the influence scoring matrix is ​​numerically filled using a matrix filling method (parameters: variable index, comprehensive scoring result), generating a complete matrix-form two-dimensional structure of <variable × score>. Furthermore, the structural and numerical correctness of the rating matrix is ​​verified by a matrix consistency verification algorithm (parameters: matrix dimension, rating range), and the standardized variable influence rating matrix that passes the verification is output. The cumulative contribution intensity value is linearly normalized using a normalization algorithm. The formula is as follows: in, This represents the original cumulative contribution intensity value. This is the minimum value of the indicator. This is the maximum value of the indicator; The same formula is used to perform uniform normalization on the path depth and frequency of action indices; The weighted aggregation formula is used: in, Let i be the overall score value of the i-th variable. , and These represent the normalized cumulative contribution intensity, path depth, and frequency of action, respectively. , , These are the corresponding weighting coefficients; By using vectorized calculation, the above formula is applied to the entire variable index set to achieve batch generation and storage of the scoring matrix; By verifying the consistency of the matrix, we ensure that each scoring result is in the range of [0,1] and that the matrix structure has no missing columns or rows, so that the final scoring matrix can be directly used for topology generation and visualization analysis. For example, in a land development project, the set of input variables includes three standardized variables: building area, property status, and construction year. Their cumulative contribution intensity values ​​are 0.82, 0.65, and 0.47, respectively; path depth values ​​are 2, 3, and 4; and action frequency values ​​are 8, 12, and 5. A weighted coefficient vector is then defined. =0.5, =0.3, =0.2. The path depth was normalized to 0.33, 0.66, and 1.00, and the frequency of action was normalized to 0.67, 1.00, and 0.42. Substituting these values ​​into the weighted composite formula, the building area score was... =0.64, property rights status score is =0.73, construction date rating is =0.66. After the scoring matrix was generated and verified, all values ​​were within the range of 0-1 and the matrix structure was complete. It can be directly used by step S5.6 to generate the dynamic influence weight propagation topology diagram, which significantly improves the interpretability and policy adaptability of the compensation calculation results. S5.6: Based on the variable influence scoring matrix and propagation path structure, generate a dynamic influence weight propagation topology graph. The topology graph includes node influence weight annotations, path contribution intensity annotations, and visual connection lines for variable action paths to support the generation of subsequent structured parsing information packages.

[0016] Step S6: The input variables are sorted according to the cumulative contribution intensity index, and a structured analytical information package is generated by combining path depth and frequency of action. This information package includes a ranking table of main driving factors, a pie chart of positive and negative contribution ratios, and a complete transmission chain flowchart. Specifically, it includes: S6.1: Based on the cumulative contribution intensity index in the dynamic influence weight propagation topology graph, the input variables are normalized to eliminate the influence of dimensional differences on the ranking results; S6.2: Sort the normalized cumulative contribution intensity values ​​to generate a ranking table of main driving factors, in which each input variable is arranged from high to low according to its overall impact weight on the compensation amount, in order to identify key influencing factors; S6.3: Based on the propagation path data recorded by the symbolic tracking engine, extract the path depth information of each input variable in the directed graph structure, calculate its role level in the formation of compensation amount, and evaluate the indirectness and directness of the variable's influence. S6.4: Statistical analysis is performed on the frequency of the input variables in the propagation path. Based on the weighted combination of the number of times the variables act and the path depth, a variable influence scoring matrix is ​​constructed to quantify their stability contribution in different compensation rule chains. The propagation path set generated by the symbolic tracing engine is subjected to frequency statistical processing. The path enumeration algorithm (parameters: node set, edge set, path length limit) is used to calculate the occurrence frequency of each input variable in different compensation rule links. Furthermore, by using a path depth weighting algorithm (parameters: number of times the variable appears, path depth value set, weighting coefficient setting rules), the number of actions is weighted hierarchically, and a weighted set of action frequency values ​​is obtained to take into account the influence distribution of the variable in shallow and deep paths. Furthermore, a normalization algorithm (parameters: weighted frequency value set, normalization range [0,1]) is adopted to unify the dimensions of the variable frequency values ​​and generate a standardized frequency sequence. Furthermore, the influence score of the variables is calculated using a weighted combined scoring formula, as follows: in, Score the influence of variables. These are the weighting coefficients. This is the normalized frequency of action value. This is a standardized path depth value; Furthermore, through a matrix assembly algorithm (parameters: variable set, influence score sequence, cumulative contribution intensity value set, path depth value set), the variable influence score matrix is ​​constructed, and each indicator is arranged by column and bound by variable index row to generate a score matrix data structure for subsequent structured parsing information package generation; By combining frequency of action statistics with path depth weighted processing, the occurrence frequency and hierarchical information of the previous step are transformed into a quantitative stability contribution score, thereby achieving an objective description of the global stability and influence of variables in different compensation rule chains. For example, in a real-world land development project with 20 compensation rules and 50 input variables, the node set has 70 nodes, the edge set has 120 edges, and the path length limit is set to 6. The variable "building area" appears 45 times in all paths, with an average path depth of 3. After weighting by a factor of 0.6, the weighted frequency of action is... After normalization to the range [0,1], it becomes The variable property rights status appeared 30 times, with an average path depth of 4. After weighting by a coefficient of 0.6, the weighted frequency of action was [value missing]. After normalization, it becomes Take the weighting coefficients. Substituting into the weighted combination scoring formula, the impact score of building area is: = The impact score of property rights status is = The final result is a 50×3 variable influence scoring matrix, where the columns are cumulative contribution intensity, normalized path depth, and weighted action frequency, respectively. This matrix is ​​visualized to identify highly stable contribution variables, significantly improving the accuracy and reliability of the analysis of compensation amount composition. S6.5: Based on the ranking table of main driving factors, the variable influence scoring matrix and the information on the direction of the propagation path, generate a pie chart of the positive / negative contribution ratio to graphically show the enhancing or inhibiting effect of each variable on the compensation amount and its relative proportion. S6.6: Based on the node-edge relationship model in the directed graph structure, extract the complete transmission chain path, generate an interactive flowchart structure, and label the numerical changes of variables and rule nodes in each path and the corresponding influence weight percentage, so as to support users in tracing and understanding the process of compensation amount formation. Based on the node-edge relationship model in the directed graph structure, a path extraction algorithm (parameters: node type = input variable, rule node, intermediate amount node; edge attribute = data flow direction and action direction) is used to extract and serialize the complete transmission chain path. Furthermore, by using a path serialization parsing algorithm (parameters: node order = rule execution order, edge weight source = dynamic influence weight value), the ordered arrangement of path nodes and the mapping of their interconnections are realized, and a path information dataset is obtained. Furthermore, a numerical change calculation method (parameters: node calculated value = intermediate result of compensation amount, predecessor node calculated value = result of previous rule execution) is adopted to calculate the difference in amount between adjacent nodes and generate numerical change indicators for each node in the path. Furthermore, the relative influence of nodes is labeled using a weight percentage calculation formula, as follows: Among them, the influence weight of the current node is the dynamic weight of the node recorded by the symbolic tracking engine, and the final sum of the influence weight of the compensation amount is the sum of the weight values ​​of all nodes in the compensation amount constitute the path. Furthermore, an interactive flowchart generation algorithm (parameters: graph rendering engine = supports directed graph layout optimization, interactive events = click, hover, drag; layout strategy = hierarchical layout) is adopted to realize the visual layout drawing of nodes and edges, and to bind numerical change and weight percentage labels to each node; Furthermore, by using a node label rendering method (parameters: font size and color dynamically matched with weight values, position anchored to the geometric center of the node), the label information is directly overlaid and displayed in the flowchart structure, generating an interactive and complete transmission chain flowchart; Through the above graph structuring process, the variable influence scoring matrix and dynamic propagation path results from the previous step are transformed into an interactive flowchart containing numerical changes and weight percentage labels, thereby achieving full traceability and multi-dimensional interpretation of the compensation amount composition path. For example, in a housing compensation calculation task for an urban renewal land development model, the input node set includes variables such as building area (120㎡), property status (commercial housing), construction year (2005), and location characteristics (area number A12). The rule node set includes the basic amount calculation rule (unit price 8000 yuan / ㎡), building age depreciation rule (depreciation coefficient 0.95), and property rights correction rule (increase ratio 1.05). The node sequence obtained by the path extraction algorithm is: building area → basic amount → building age depreciation → property rights correction → final compensation amount. The formula for calculating the numerical change is applied to adjacent nodes. For example, the difference between the basic amount node (calculated value = 960,000 yuan) and the building age depreciation node (calculated value = 912,000 yuan) is -48,000 yuan, and the difference between the property rights correction node (calculated value = 957,600 yuan) and the building age depreciation node is +45,600 yuan. When applying the weight percentage calculation formula, it is assumed that the total weight of the final compensation amount is... The dynamic impact weight of building age and depreciation nodes is: Then its weight percentage is After the flowchart is generated, users can click on the "Building Age Depreciation" node on the interface to view detailed information about the reduction of 48,000 yuan and its impact weight of 32%, and can also view the change curve of the nodes before and after it by hovering over it, thus realizing the transparency and enhanced interpretability of the compensation amount formation process; S6.7: Integrate the ranking table of main driving factors, the pie chart of positive and negative contribution ratios, and the flowchart of the complete transmission chain to generate a structured analytical information package as a multi-dimensional explanatory supplementary information for the compensation amount output, so as to improve the visualization and interpretation capabilities and decision support value of the calculation results.

[0017] Step S7: The compensation amount and the structured parsed information package are simultaneously visualized and output, presenting an expandable transmission chain flowchart on the interactive interface, annotating the numerical changes of each calculation node and the corresponding percentage of influence weight. Specifically, this includes: S7.1: Based on the compensation amount calculation results and structured parsing information package, the visualization output content is integrated and processed to generate a unified format visualization dataset. The visualization dataset includes compensation amount values, a ranking table of main driving factors, a pie chart of positive and negative contribution ratios, and a complete transmission chain flowchart. S7.2: Based on the visualization dataset, perform data mapping operations of the front-end rendering engine, use the front-end graphics engine to highlight the compensation amount value, and draw the pie chart of positive and negative contribution ratios to generate interactive visualization components. S7.3: Construct a dynamic, expandable, interactive flowchart model based on complete transmission chain flowchart data, and use a graph structure rendering engine to visualize the nodes and edges of the flowchart to form an interactive interface for displaying the compensation amount transmission path. S7.4: Label the numerical changes and influence weight percentages of each calculation node in the flowchart, and use a dynamic labeling engine to overlay numerical information on nodes and edges to improve the readability and interpretability of the compensation amount path. S7.5: Based on user interaction commands, the flowchart is dynamically expanded and collapsed. The front-end event listening mechanism responds to the user's click and hover behavior, so as to realize the partial expansion and focused display of any path segment in the transmission chain flowchart. S7.6: The compensation amount, pie chart, and transmission chain flowchart are integrated into the interface layout. A multi-screen adapted interactive interface is built based on a responsive front-end framework to ensure that the compensation results and analysis information can be displayed and interacted with synchronously on devices with different resolutions.

[0018] The present invention also provides a dynamic calculation system for housing compensation under the land development model, which uses the above-mentioned dynamic calculation method for housing compensation under the land development model to perform dynamic calculation of housing compensation under the land development model.

[0019] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0020] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and rules of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic calculation method for housing compensation based on land development patterns, characterized in that, Includes the following steps: S1: Construct a compensation rule model based on local policy documents, define the input variables, basic parameters and calculation type for each compensation rule, and initialize the basic impact weight value; S2: Connect to the housing basic data platform and manual data entry terminal to obtain and parse housing physical attributes, ownership information and location characteristics data, and generate a standardized set of input variables including building area, property status and construction year; S3: Based on the directed graph structure, a symbolic tracking engine is built, which abstracts the compensation calculation process into a node-edge relationship model. The nodes contain input variables, rule items and intermediate outputs, and the edges record the data flow and direction of action, thus establishing a dynamic influence weight propagation channel. S4: When performing compensation amount calculation, dynamically adjust the influence weight propagation strategy according to the calculation type of each rule, and generate dynamic influence weight values ​​corresponding to each intermediate result; S5: Based on the propagation path data of the symbolic tracking engine, calculate the cumulative contribution strength, path depth, and frequency of action of the input variables, and generate a dynamic influence weight propagation topology graph. S6: Sort the input variables according to the cumulative contribution intensity, and generate a structured parsing information packet by combining the path depth and the frequency of action; S7: Visualize and synchronously output the compensation amount and the structured parsing information package, and present an expandable transmission chain flowchart on the interactive interface, marking the numerical change of each calculation node and the corresponding influence weight percentage.

2. The dynamic calculation method for housing compensation based on land development patterns according to claim 1, characterized in that, Step S1 specifically includes: Structured analysis of local policy documents is performed to extract compensation rule items and their applicable conditions, thereby obtaining a set of rule items related to housing compensation in the policy text; Based on the set of rule items, field mapping processing is performed on each rule to identify its dependent input variables; Based on the description of the rule item and the semantic analysis results of the policy provisions, basic parameter values ​​are extracted. Based on the logical structure of the rule items, a rule type classification algorithm is used to identify the calculation type of each rule and generate a rule calculation type identifier. Based on the authority level, scope of application, and historical implementation data of the policy provisions, a basic impact weight assignment is performed on each rule item to generate a basic impact weight configuration matrix; The set of rule items, the input variables, the basic parameter values, the rule calculation type identifier, and the basic influence weight configuration matrix are structurally bound to generate a configuration data package for the compensation rule model.

3. The dynamic calculation method for housing compensation based on land development patterns according to claim 2, characterized in that, The compensation rule model uses a text parsing engine to perform OCR recognition and semantic segmentation on policy documents in electronic text or scanned images, extracts compensation rule items and their applicable conditions, and generates a set of rule items.

4. The dynamic calculation method for housing compensation based on land development patterns according to claim 1, characterized in that, Step S2 specifically includes: Based on the housing basic data platform interface protocol, we acquire surveying and mapping data, property registration information and geographic information system location data to construct the original dataset of housing physical attributes and ownership information. The building area field in the original dataset is format-standardized, and a unified unit conversion algorithm is used to convert area data from different sources into square meter units to obtain standardized building area variables. Based on the property registration information, a property status identification algorithm is executed to extract the property ownership type label and generate standardized property status variables; The construction year field of the building is parsed and classified, and a year interval mapping algorithm is used to convert specific years into standardized construction year labels. Based on the geographic information system location data, a spatial coordinate analysis and regional coding mapping algorithm is executed to convert latitude and longitude information into corresponding administrative divisions and area numbers, generating standardized location feature variables. The standardized building area variable, the standardized property status variable, the standardized construction year label, and the standardized location feature variable are merged and structurally encapsulated to generate a unified set of housing input variables.

5. The dynamic calculation method for housing compensation based on land development patterns according to claim 1, characterized in that, Step S3 specifically includes: Based on the rule items, input variables, and intermediate calculation results in the compensation rule model, a set of nodes in a directed graph is defined to form the topological structure basis of the compensation calculation path; Based on the data dependencies between compensation rules, a set of edges is constructed in a directed graph, where each edge represents the data flow direction and action direction from one node to another. Based on the calculation type of the compensation rule, each edge is assigned an action type label, which includes linear addition, multiplicative amplification, and conditional inheritance. A runtime graph structure for a symbolic tracing engine is constructed, mapping all node and edge relationships in the compensation amount calculation process to a graph data structure in memory. The graph structure supports dynamic node expansion and edge weight updates, and tracks the compensation calculation path in real time. In the graph structure, a dynamic influence weight propagation channel is initialized, and an initial influence weight value is assigned to each node. The initial influence weight value is assigned based on the basic influence weight value of the rule in the policy system. Based on the execution order and data flow of the compensation rules, the node states and edge weights in the graph structure are dynamically updated to generate a complete graph of the impact of compensation amount.

6. The dynamic calculation method for housing compensation based on land development patterns according to claim 5, characterized in that, The node set includes basic housing attribute variable nodes, compensation rule nodes, and intermediate amount output nodes.

7. The dynamic calculation method for housing compensation based on land development patterns according to claim 1, characterized in that, Step S4 specifically includes: Based on the rule calculation types in the predefined compensation rule model, each rule is classified and identified. The rule calculation types include linear addition rules, multiplication rules, and condition judgment rules. The linear addition rule is processed using a weighted summation mode. Based on the original influence weight values ​​of the input variables and the linear addition coefficient of the rule, a weighted summation algorithm is used to calculate the comprehensive influence weight value of the rule's output node. The multiplicative rule is processed using a marginal effect amplification mode. The amplified influence weight value of the rule's output node is generated based on a multiplicative synthesis algorithm by using the current influence weight value of the input variable and the multiplication coefficient of the rule. The condition judgment rule is processed using a weight inheritance mode. Based on the original influence weight values ​​of the input variables and the logical branch results of the condition judgment rule, the inherited influence weight value of the rule output node is determined by a branch weight inheritance algorithm. The output influence weight value of each rule node is used as the input influence weight value of the next calculation node. Based on the directed graph structure in the symbolic tracking engine, the weight is propagated layer by layer to form a dynamic influence weight propagation sequence of each variable in the compensation amount composition path. Based on the dynamic influence weight propagation sequence, a set of dynamic influence weight values ​​for each intermediate computing node is generated.

8. The dynamic calculation method for housing compensation based on land development patterns according to claim 1, characterized in that, Step S5 specifically includes: The node-edge relationship model output by the symbolic tracing engine is analyzed for topology, and the set of data flow paths between input variables and rule items is extracted to construct a set of variable propagation paths. Based on each path in the set of variable propagation paths, calculate the path depth value of each input variable on each path; Dynamic weight accumulation calculation is performed on the nodes in each path, and the cumulative contribution intensity of the input variable on the entire path is calculated by combining the dynamic influence weight value of the rule node; Based on the frequency of occurrence of input variables in the set of variable propagation paths, the frequency of their role in different paths is statistically analyzed. Based on the path depth value, the cumulative contribution intensity, and the frequency of action, an input variable influence scoring matrix is ​​constructed, and each variable is normalized to generate a standardized variable influence scoring result. Based on the variable influence scoring matrix and propagation path structure, a dynamic influence weight propagation topology graph is generated.

9. A dynamic calculation method for housing compensation based on land development patterns according to claim 8, characterized in that, The dynamic influence weight propagation topology graph includes node influence weight annotation, path contribution intensity annotation, and visual connection lines for variable action paths.

10. A dynamic calculation system for housing compensation based on land development patterns, characterized in that: The dynamic calculation method for housing compensation under the land development mode, as described in any one of claims 1-9, is used to perform dynamic calculation of housing compensation under the land development mode.

Images