Land value evaluation and circulation transaction system based on multi-source heterogeneous data fusion

Abstract

The invention relates to the technical field of land value evaluation data processing, and discloses a land value evaluation and circulation transaction system based on multi-source heterogeneous data fusion, which comprises the steps of converting policy texts, environment monitoring and market transaction data into standardized influence vectors, calculating an information unbalance degree through geographic grid projection to represent data conflict intensity, and calculating a data conflict degree; according to the method, internal contradictions of multi-source data are spatially developed, conflicts masked in traditional assessment are quantified into operable risk indexes, meanwhile, dynamic confidence degree modulation is achieved in combination with transient feature analysis, and the reliability of the system is improved. Land value judgment has trend insight and risk early warning capabilities, and a high-credibility decision-making basis is provided for circulation transaction.

Description

Technical Field

[0001] This invention relates to a land value assessment and transfer transaction system based on the fusion of multi-source heterogeneous data, belonging to the field of land value assessment data processing technology. Background Technology

[0002] In the field of land valuation, multi-source heterogeneous data fusion technology is commonly used. Existing systems usually rely on weighted averages or complex machine learning models to generate a single valuation. These methods face fundamental limitations in dynamic areas such as urban-rural fringe areas: when there is an essential conflict between heterogeneous data such as policy planning (e.g., +0.8 development orientation) and environmental monitoring data (e.g., -0.7 ecological constraints), the system is forced to unify contradictory information through algorithm fitting, resulting in two major systemic defects: first, data conflicts are blurred, making it impossible to quantify the specific risk location and intensity; second, complex models obscure the basis for value judgment, making it difficult for users to trace the root cause of the conflict.

[0003] Industry attempts to improve land value through time-series analysis or spatial interpolation have exacerbated computational complexity and failed to address the core issue: existing technologies consistently focus on the monetary value of land while neglecting the fundamental question of the reliability of value judgments. This is especially true in real-world scenarios involving fluctuations in data timeliness and sudden policy adjustments, where static valuation models prove particularly vulnerable. Therefore, the technical problem this invention aims to solve is how to construct a quantifiable and interpretable risk assessment mechanism that preserves the diversity of multi-source data and achieves a synergistic expression of land value trends and confidence levels. Summary of the Invention

[0004] This invention provides a land value assessment and transfer transaction system based on the fusion of multi-source heterogeneous data. Its main purpose is to solve the problem that existing assessment systems cannot reveal the inherent conflicts of multi-source data, resulting in the annihilation of risk information and the lack of transparency in decision-making basis.

[0005] To achieve the above objectives, the present invention provides a land value assessment and transfer transaction system based on multi-source heterogeneous data fusion, comprising:

[0006] The data standardization module is used to access multi-source heterogeneous data and transform each type of data into a standardized impact vector through pre-determined transformation rules. The transformation rules transform policy planning texts into impact vectors representing development orientation, environmental monitoring data into impact vectors representing ecological constraints, and market transaction data into impact vectors representing market activity.

[0007] The information imbalance calculation module is used to project standardized influence vectors onto a geographic information grid with a unified spatial benchmark according to the geographic scope, and calculate the degree of dispersion of all projected influence vectors within each geographic grid to generate an information imbalance degree that characterizes the degree of information conflict. The information imbalance degree reflects the level of consistency between different data sources for the same land value judgment.

[0008] The value coordinate generation module receives information imbalance and generates a confidence index based on the information imbalance using an inverse proportional function. The confidence index represents the credibility of the land value judgment, and its value is negatively correlated with the information imbalance. Furthermore, the value coordinate generation module calculates the arithmetic mean of all influence vectors within the geographic grid to generate a trend vector, which represents the development direction and intensity of land value. The system finally outputs a binary value coordinate consisting of the trend vector and the confidence index, which serves as the basis for land value assessment and transfer transactions.

[0009] Preferably, the information imbalance calculation module is specifically used to obtain standardized influence vectors and calculate the variance or standard deviation of all influence vectors within the geographic grid to obtain the information imbalance.

[0010] Preferably, in the value coordinate generation module, the set inverse proportional function for generating the confidence index is C = 1 / (1 + k × ID), where C represents the confidence index, ID represents the information imbalance degree, and k represents the system adjustment constant.

[0011] Preferably, in the data standardization module, the transformation rules generate influence vectors with corresponding positive, negative, or neutral orientations from government planning texts through keyword matching and sentiment analysis; the transformation rules generate influence vectors representing ecological constraints from PM2.5 concentration and water quality levels in environmental monitoring data after linear mapping; and the transformation rules generate influence vectors representing market activity from housing price growth rate and rent growth rate in market transaction data.

[0012] Preferably, the value coordinate generation module is also used to extract the time series data of the information imbalance when the information imbalance first reaches or exceeds the set warning threshold, and calculate the change slope and duration characterizing the transient features of information conflict from the time series data; based on the change slope and duration, generate a dynamic decay coefficient through a given nonlinear function; and use the dynamic decay coefficient to modulate the confidence index to generate a final confidence index that reflects the transient nature of the conflict.

[0013] Preferably, the given nonlinear function is α = e (-k·K·T) , where α represents the dynamic attenuation coefficient, K represents the slope of change, T represents the duration, and k represents the system adjustment constant.

[0014] Preferably, it further includes: a trend vector field generation module, used to construct an original trend vector field from the trend vectors of all geographic grids; a expected trend field modeling module, used to perform Gaussian smoothing on the original trend vector field to generate an expected trend vector field; a residual vector field calculation module, used to subtract the original trend vector field from the expected trend vector field to obtain a residual vector field; and a risk vortex calculation module, used to calculate the curl of the residual vector field to generate a risk vortex index characterizing the macro-structural risk of the region.

[0015] Preferably, multi-source heterogeneous data includes government planning data, market transaction data, population flow data, and environmental remote sensing data.

[0016] Preferably, the system also includes a transaction matching module. When matching transactions, the transaction matching module receives binary value coordinates and recommends land plots with a confidence index higher than the preset threshold to the user or provides risk warnings for land plots with a confidence index lower than the preset threshold based on the comparison results between the confidence index and the preset threshold. The level of risk warning is related to the degree of decrease in the confidence index.

[0017] Preferably, the granularity of the geographic information grid can be dynamically adjusted based on the actual geographical characteristics of the land, the requirements of regional development planning, and the evaluation needs input by the user.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. By transforming multi-source heterogeneous data into standardized impact vectors and projecting them onto a unified geographic grid, the conflict of value judgments from different data sources is made explicit on a spatial benchmark. The degree of dispersion of the impact vectors within the grid is calculated to generate information imbalance, which transforms the data contradictions that are hidden in traditional assessments into quantifiable risk indicators. This mechanism does not rely on data weighted fusion, but exposes the distribution of contradictions through spatial projection, allowing decision-makers to intuitively identify the areas of fundamental disagreement in value judgments.

[0020] 2. Based on the inverse confidence index generated by information imbalance, the system maps the intensity of data conflict to the credibility level of value judgment. When the conflict first reaches the warning threshold, the system automatically extracts transient features from the time series and dynamically modulates the confidence index through a nonlinear decay function. This enables the system to distinguish between sudden risks and long-term structural contradictions, providing dynamic risk labels for land transactions that evolve over time. By calculating the spatial residual of the trend vector field and analyzing its curl characteristics, the system elevates the evaluation results of the micro-grid to macro-regional analysis. The residual vector field reveals the deviation between the actual trend and the ideal development model, while the curl calculation captures the rotational friction effect of regional value flow. The resulting risk vortex index, for the first time in the field of land assessment, achieves the visual positioning of the transmission path of systemic risks and accurately identifies the key risk points of regional development.

[0021] 3. The data standardization module establishes a unified measurement basis for spatial projection, the information imbalance calculation provides conflict characteristics for confidence modulation, and the trend vector field generation provides input for spatial risk diagnosis. The three-level processing mechanism forms a technical closed loop: the output of the front-end automatically triggers the subsequent refined analysis, enabling the system to shift from static evaluation to a dynamic response paradigm of conflict perception-confidence modulation-field diagnosis, avoiding the dependence of traditional methods on a single valuation indicator; all core calculations are implemented using mature mathematical methods, avoiding the black box risk brought by complex models, and the standardization rules adopt interpretable operations such as keyword matching and linear mapping, so that the system has low computational overhead characteristics while maintaining innovative logic and adapting to the infrastructure conditions of different regions. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the functional structure of the land value assessment and transfer transaction system based on multi-source heterogeneous data fusion, as described in this invention.

[0023] Figure 2 This is a graph showing the temporal evolution trend of information imbalance in the context of sudden risks and long-term structural contradictions.

[0024] Figure 3 This is a flowchart illustrating the interaction and calculation process of the value coordinate generation module in geographic information grid data processing according to the present invention.

[0025] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without effort are within the scope of protection of the present invention.

[0027] This application provides a land value assessment and transfer transaction system based on multi-source heterogeneous data fusion. In practical applications of land value assessment, a core challenge stems from the multi-source heterogeneous nature of the input data. For example, a policy planning document guiding regional development and a set of environmental monitoring data reflecting ecological quality differ in format, content, and dimensions. Forcibly averaging these data would mask their inherent conflicts, leading to distorted assessment results. To address this challenge, the data standardization module in this invention establishes a unified metric for data from different sources. This module is configured to access multi-source heterogeneous data, including government planning data, market transaction data, population flow data, and environmental remote sensing data, and executes pre-determined transformation rules for each type of data to generate a standardized influence vector capable of mathematical operations. Specifically, for unstructured data such as government planning documents... The system uses keyword matching and sentiment analysis to identify words with clear guidance in the text, such as "encouraging development" and "ecological red line." Based on their sentiment polarity and intensity, it generates an influence vector representing the development orientation. For example, a strong positive development orientation can be transformed into a three-dimensional vector with specific components assigned positive values. For environmental monitoring data, the system normalizes key scalar indicators, such as PM2.5 concentration and water quality level, through a linear mapping function, and then generates an influence vector representing ecological constraints, whose values ​​are usually negative. Similarly, for market transaction data, the system extracts core indicators such as house price growth rate and rent growth rate, and generates an influence vector representing market activity. Through this series of transformation procedures, all raw data are reconstructed into standardized influence vectors with unified dimensions and comparability, thus laying a solid data foundation for subsequent conflict analysis on the same spatial benchmark.

[0028] After standardized influence vectors from different sources are generated, a more serious challenge is how to objectively measure the consistency of these vectors' viewpoints at specific spatial locations. Traditional methods often directly perform weighted fusion at this step, but this is precisely the root cause of risk information annihilation. Therefore, the system adopts the following procedure: all standardized influence vectors are precisely projected onto a geographic information grid with a unified spatial benchmark according to their geographical scope. The granularity of this grid can be adaptively optimized based on the actual geographical characteristics of the land, the requirements of regional development planning, and the user's input assessment needs to ensure the accuracy and efficiency of the analysis. Furthermore, the information imbalance calculation model... The core task of this module is not fusion, but quantification of contradictions. It calculates the dispersion of all influence vectors projected within each geographic grid cell. Specifically, this module obtains the set of all influence vectors within the grid and calculates their variance or standard deviation. This result is defined as the information imbalance degree, represented by the symbol ID. Thus, a high ID value directly and quantitatively reveals significant conflicts and inconsistencies in the value judgments of different data sources for a specific plot, transforming the inherent contradictions that are often obscured in traditional assessments into a clear and quantifiable risk indicator. After obtaining the quantified information imbalance degree ID, the system... The system needs to transform this into a more instructive indicator for decision-makers, namely, the credibility of value judgments. A value trend derived from highly conflicting data should have significantly lower reference value than a conclusion based on highly consistent data. To this end, the value coordinate generation module is configured to receive the information imbalance degree ID and generate an inversely proportional confidence index based on it. This confidence index is denoted by the symbol C, and its core generation logic follows a predefined inverse function C = 1 / (1 + k × ID), where k is a system adjustment constant. It should be noted that the constant k is not arbitrarily set but is determined through an offline calibration procedure that utilizes historical data... Backtracking tests were conducted, and the k value was adjusted so that the C values ​​calculated in multiple typical areas could be reasonably distributed within the range of 0 to 1, thus ensuring its universality and interpretability. At the same time, the value coordinate generation module also calculated the arithmetic mean of all influence vectors within the geographic grid in parallel, thereby generating a trend vector. The magnitude and direction of this vector intuitively represent the comprehensive development intensity and direction of the land value. Finally, the system outputs a binary value coordinate composed of the trend vector and the confidence index. It is no longer a single valuation number, but a composite decision-making basis that includes value trend and judgment credibility, thus solving the problem of decision-making black box.

[0029] Furthermore, in the face of a dynamically changing market and policy environment, a static confidence index is insufficient to fully capture the time-varying characteristics of risk. For example, a sudden and sharp increase in data conflict carries a far greater risk warning than a long-term, stable state of high conflict. To address this challenge, the value coordinate generation module integrates a transient feature analysis and confidence dynamic modulation mechanism. This mechanism sets a warning threshold, determined based on the statistical distribution of historical ID data in a region (e.g., taking its 95th percentile). When the ID value of a grid first reaches or exceeds this warning threshold, the system immediately initiates a time series analysis program, automatically extracting the sequence data of that ID value over a past period (e.g., 30 days) and calculating two key transient features: the slope K, obtained through linear regression analysis of the time series data; and the duration T, i.e., the length of time the ID value remains above the warning threshold. Subsequently, based on these two transient features, the system uses a given nonlinear function α = e (-k·K·T) A dynamic decay coefficient α is generated, where k is the system adjustment constant, and K and T are the slope and duration of change calculated above. Finally, the system uses this dynamic decay coefficient to multiply and modulate the original confidence index to generate a final confidence index that reflects the transient nature of the conflict. Through this design, the system can effectively distinguish between sudden risks and long-term structural contradictions, providing dynamic risk labels that evolve over time for land transactions, and enhancing the depth and timeliness of risk warnings.

[0030] Beyond the refined assessment of micro-plots, this invention also provides a mechanism for diagnosing regional macro-structural risks. While local value trends may appear reasonable, their spatial correlation patterns at the regional scale may harbor systemic risks. To reveal such risks, the system first uses a trend vector field generation module to spatially organize the trend vectors of all geographic grids, forming an original trend vector field. However, the original field contains numerous local details and noise, making it difficult to observe the macro-structure. Therefore, the expected trend field modeling module then applies Gaussian smoothing to this original trend vector field. This smoothing filters out high-frequency disturbances, generating a smoother expected trend vector field that better represents the overall development inertia of the region. Subsequently, the residual vector field calculation module compares the original trend vector field with the expected trend vector field. Vector subtraction yields a residual vector field, where each vector represents the deviation between the actual value trend of a specific land parcel and the general expected trend of its region, thus accurately revealing development anomalies. Finally, and most crucially, the risk vorticity calculation module performs curl calculation on this residual vector field. Curl, as a physical quantity measuring the rotational intensity of a vector field, is defined here as the risk vorticity index. A region with a high risk vorticity index indicates strong rotation, internal friction, or conflict in its internal value flow, which is a clear signal of regional macro-structural risk. Just as identifying the vortex core of turbulence in fluid mechanics, this technology, for the first time in the field of land assessment, enables the visual positioning of the transmission path of systemic risk, accurately identifying the eye of the storm of regional development risk.

[0031] Ultimately, the binary value coordinates and various risk indices generated by this invention play a direct decision support role in the land transfer and transaction process. The transaction matching module within the system no longer relies solely on price or a single value assessment result when matching transactions. This module receives the needs of both buyers and sellers and uses the binary value coordinates as the core filtering and sorting criterion when recommending land properties. The specific procedure is as follows: The system receives a user-preset confidence index threshold, which represents the user's risk preference. The system automatically prioritizes recommending land properties with confidence indices higher than this threshold. Simultaneously, for land properties with confidence indices lower than the preset threshold, the system proactively provides risk warnings. The warning level (e.g., level one, level two, level three risk) is positively correlated with the degree to which the confidence index is below the threshold; that is, the lower the confidence index, the higher the level of risk warning. This mechanism ensures that both parties in the transaction, especially the recipient of the land value, can make decisions in an environment of transparent information and controllable risk, providing an unprecedentedly highly reliable decision-making basis for land transfer and transactions.

[0032] The core functional constants and transformation rule models required for the operation of this system are deterministically generated through a set of offline calibration and supervised learning procedures before deployment. The specific process includes: First, for the confidence index function C = 1 / (1+k C The fundamental confidence constant k in ×ID) C With dynamic decay function The dynamic decay constant k in α Joint calibration is performed by retrieving multi-source high-frequency time-series data covering at least thirty-six complete months from historical databases of the target region. A set of historical events with known information conflict levels and risk grades is selected as a benchmark anchor set using an expert consensus annotation method. Then, a numerical optimization algorithm is employed to find the unique global optimum (k) by minimizing the Kolmogorov-Smirnov test statistic between the system's calculated confidence index and the theoretical credibility corresponding to the benchmark anchor set. C ,k α Secondly, this procedure ensures that when facing initial data environments in different cities or regions, the system's core parameters can adaptively anchor to local statistical characteristics, providing a horizontally comparable decision-making basis for risk situation projection in the city's digital twin system. The transformation rule knowledge base built into the data standardization module is constructed through a supervised learning process based on historical real development results. In this process, for each plot of land in the training set, its actual development performance within a specific time window (e.g., thirty-six months) after the transaction is completed is quantified into a multi-dimensional real development vector V. actual The vector's dimensions encompass comprehensive indicators such as the annualized growth rate of asset value, the level of industrial introduction, and changes in the value of ecosystem services; the system generates a predictive trend vector V by processing multi-source heterogeneous raw data through transformation rules. pred , with V actual The comparisons are made, and the prediction bias is quantified using the following composite loss function L: The first term is the cosine distance loss, which penalizes directional deviation, and the second term is the Euclidean distance loss, which penalizes amplitude deviation. The weight coefficients λ1 and λ2 are determined through cross-validation on the training set. Finally, the system uses the Adam optimizer to perform gradient descent on this loss function L, iteratively adjusts and finally solidifies all learnable parameters in the transformation rule, thereby establishing a transformation model that can map multi-source data to future development potential and has deterministic and high-dimensional feature capture capabilities.

[0033] Example 1: In a specific industrial application scenario, the operation of the technical solution of the present invention is as follows: In a rapidly urbanizing urban-rural fringe area, the local government has issued a plan to build a high-tech industrial park. This area happens to include a wetland zone designated as an ecological protection red line. Simultaneously, the surrounding real estate market is experiencing drastic price fluctuations driven by short-term speculative capital. Faced with this complex decision-making environment, after deployment, the data standardization module of the present invention first accesses and processes three key heterogeneous data types. Based on preset transformation rules, it transforms the government planning text pointing to the high-tech industrial park into a development-oriented influence vector with a strong positive amplitude, and generates a significant wetland ecological monitoring index from environmental remote sensing data. The negative magnitude of the ecological constraint impact vector is used to generate an impact vector representing market heat from the abnormal price growth rate in market transaction data. This step establishes a data foundation for subsequent conflict analysis on the same spatial benchmark. Subsequently, the information imbalance calculation module projects these three types of impact vectors acting on the same geographic space onto a geographic information grid with a unified spatial benchmark. For each grid cell covering the urban-rural fringe area, the system performs its core discreteness calculation, that is, by calculating the variance of all impact vectors within the grid, it generates an information imbalance ID that quantifies the degree of information conflict. Given the huge differences in the vector directions of development orientation, ecological constraints, and market heat, the ID value of this area is calculated as a significantly high value.

[0034] The value coordinate generation module receives the high ID value and uses it as a key input to generate the confidence index. It generates the confidence index C using a set inverse proportional function C = 1 / (1 + k × ID). Due to the significantly high ID value, the final generated C value is correspondingly at an extremely low level. Simultaneously, although the trend vector calculated by the system may point to a positive increase due to the strength of the development guidance vector, the system ultimately outputs a binary value coordinate system to the decision-maker, consisting of this trend vector and the extremely low confidence index C. This process demonstrates that the evaluation paradigm of this scheme does not seek an absolute, single value anchor point, but rather transforms into a direct measurement of the reliability of value judgments. Its inherent technical logic is that the unified vector format generated by the data standardization module is a direct prerequisite for the information imbalance calculation module to perform discreteness analysis. The information imbalance calculation... The module transforms the vector set output by the standardized module into a new information dimension representing risk. Functionally, the two form a close coupling relationship of preconditions and value enhancement. When this binary value coordinate is submitted to the transaction matching module, its application value is realized. For a land parcel with a good trend vector but a confidence index C far below the user's preset threshold, the system automatically triggers a high-level risk warning according to built-in procedures, clearly pointing out to the user that the value judgment of the parcel is based on highly conflicting data. In this way, the inherent risks that were previously dissolved in the traditional evaluation model are accurately captured and transmitted to the final decision-maker. Whether it is a bank conducting credit approval or an investor making asset allocation, they can re-examine their decisions based on this objective risk indication, thereby effectively avoiding significant economic losses that may be caused by information inconsistency.

[0035] Example 2: This example aims to conduct a controlled simulation experiment in a high-fidelity simulation environment. This environment constructs a 10×10 virtual geographic information grid and deploys the aforementioned complete evaluation and trading system. A programmable data injection script is used to inject a set of standardized influence vectors into the grid's central cells daily for 60 consecutive simulation days. To ensure the engineering rationality of the experimental parameters, the key is establishing a warning threshold. The technical consideration lies in balancing the sensitivity of risk perception with the system's false alarm rate. Too low a threshold will introduce excessive noise, while too high a threshold may delay warnings. Therefore, the setting procedure is determined to be based on the statistical distribution of benchmark data, specifically taking the 95th percentile of the historical information imbalance ID dataset as the threshold. In an initial benchmark test, the system stably operated, and the measured mean ID was 0.15 with a standard deviation of 0.1. Based on this, the warning threshold was set to 0.15 + 1.645. ×0.1≈0.315 serves as the condition for triggering transient feature analysis. The experimental process includes two parallel test scenarios to simulate two typical risk evolution patterns. In scenario A, sudden risk, the data injection script maintains low data conflict for the first 30 days, keeping the ID value below 0.315. On the 31st day, the script instantaneously injects a set of strong influence vectors that significantly conflict with the existing data direction, causing the ID value to jump from approximately 0.2 to 0.8 within 24 hours and maintain that level. In scenario B, long-term structural contradiction, the script also maintains low data conflict for the first 30 days. Starting from the 31st day, the influence vector is fine-tuned daily, causing the ID value to increase linearly at a gradual rate, slowly climbing from 0.2 to 0.45 over 15 days and maintaining that level. The system runs continuously under both scenarios, and its key performance indicators are continuously recorded. Table 1 lists system state snapshots at multiple key time points for both scenarios, including the baseline state, for comparative analysis.

[0036] Table 1: Comparison of dynamic attenuation performance data.

[0037]

[0038] In scenario A, due to the rapid change in ID, the system calculates a significantly high slope K, causing the dynamic decay coefficient α to increase according to the formula α = e (-k·K·T)The first scenario shows a rapid exponential decline, decaying to 0.13 by the 10th day after exceeding the threshold, causing the final confidence index to be significantly modulated from the base value of 0.56 to 0.07. In contrast, in scenario B, the gradual ID growth rate corresponds to a smaller K value, making the decay process of α much more gradual. After the same 10 days, the α value remains at 0.69, and the final confidence index only mildly decreases from 0.69 to 0.48. The underlying mechanism of this significant difference in data results lies in the exponential sensitivity of the system's built-in nonlinear function to the slope K of change. This mechanism imposes a stronger confidence penalty on sudden events with a higher slope of change in ID values.

[0039] Example 3: This example combines Figures 1 to 3 This document describes a land value assessment and transfer transaction system based on the fusion of multi-source heterogeneous data. Figure 1 As shown, within the geographic information grid, the information imbalance calculation module is responsible for calculating the vector dispersion and generating an information imbalance ID, which characterizes the intensity of conflict between multi-source data. The value coordinate generation module uses this ID to calculate the confidence index C, employing the function C = 1 / (1 + k × ID), where k is the system adjustment constant. Simultaneously, it calculates the vector average of the influence vectors within the geographic grid to generate a trend vector representing the direction of land value development. This module ultimately outputs a binary value coordinate trend vector + confidence index as the core output indicator for land value assessment. The system also integrates a transient feature analysis module that triggers when the ID exceeds a threshold, used to address the issue of information imbalance. When the information imbalance ID reaches a preset warning threshold, the system automatically extracts time-series features to modulate the final confidence index, enhancing dynamic risk warning capabilities. The figure further shows that the system can be equipped with optional expansion modules, including a trend vector field generation module for constructing regional trend flow fields; an expected trend field modeling module for performing Gaussian smoothing on the original trend field to generate an expected pattern; a residual vector field calculation module for capturing the deviation between the actual trend and the expected trend; a risk vortex calculation module for calculating the curl of the residual vector field to extract regional system risk characteristics; and a transaction matching module for screening and recommending transaction targets based on the confidence index, realizing intelligent matching of land transactions.

[0040] like Figure 2As shown in the figure, the dynamic changes of two typical scenarios are illustrated: First, scenario A, sudden risk, represented by a solid line, where the ID remains stable at around 0.2 for the first 30 days. From day 31 onwards, due to the injection of conflicting external data, the ID rises sharply to over 0.8 in a short period of time, forming a significant jump. Second, scenario B, long-term structural contradiction, represented by a dashed line, where the ID value rises slowly and linearly from day 31, stabilizing at around 0.45 around day 45, showing a gentle but continuous risk accumulation trend. In addition, the figure also uses a dashed line to represent the warning threshold, which is 0.315. This threshold is the critical standard set by the system to determine whether transient feature analysis is triggered. When the information imbalance (ID) in any scenario exceeds this warning threshold for the first time, the system will automatically start the corresponding transient feature analysis module to extract key parameters such as the slope of change and duration, which are used to modulate the final confidence index to enhance the system's ability to identify and respond to sudden risks and long-term structural contradictions.

[0041] like Figure 3 As shown, firstly, the geographic information grid transmits the ID values ​​of grid cells and the set of influence vectors to the value coordinate generation module. The value coordinate generation module then enters the processing stage, processing two dimensions in parallel. On one hand, it calculates the confidence index by calling the inverse proportional function C = 1 / (1 + k × ID). The calculation engine performs the inverse proportional calculation based on the ID value and returns the basic confidence index C. On the other hand, it performs trend vector calculation by summing and normalizing the set of influence vectors, calculating the arithmetic mean of the influence vectors, and then returning the trend vector magnitude direction. Subsequently, the system constructs a binary value coordinate trend vector and a confidence index. Under specific conditions, the system can enter the transient feature modulation process, including extracting the auxiliary sequence before the warning trigger, applying the dynamic attenuation coefficient α, and generating the final confidence index C × α through the formula. This achieves dynamic response modulation to sudden data conflicts. The final output is a binary value coordinate, composed of the trend vector and the dynamically modulated confidence index. The output is formatted and delivered as a decision-making basis through the output interface. The information output by the system includes a composite judgment basis of value trend and credibility, providing structured support for subsequent land transactions or planning.

[0042] Example 4: The technical solution of this invention is applied to the systemic risk assessment of a large heterogeneous new area. The new area is bordered by a high-density commercial core area on the west, a low-density residential area in the middle, and a large area of ​​homogeneous ecological green space on the east. To address this technical constraint, before the assessment, this invention first executes a dynamic adjustment procedure for the granularity of the geographic information grid. The initial input of this procedure is the vector boundary map of the new area and a feature fluctuation layer representing the complexity of the land surface. This layer is generated by calculating the road network density and land use type entropy value per unit area. The operation flow of the adjustment procedure is as follows: The system first performs an initial grid division of the entire new area with an initial granularity of 500 meters × 500 meters. Subsequently, for each initial grid unit... The system calculates the standard deviation of the feature fluctuation layer values ​​covered within it and compares this standard deviation with a feature stability threshold. This feature stability threshold is determined by taking the 80th percentile of the fluctuation standard deviation within a group after statistical analysis of samples from different urban functional areas. If the internal standard deviation of a certain grid cell is greater than the threshold, it indicates that the internal complexity of the cell is high. The system then uniformly splits it into four sub-grids for first-level refinement. This splitting process continues recursively until the internal feature fluctuation standard deviation of all grid cells is lower than the threshold or reaches the maximum refinement level of 25 meters × 25 meters set by the system. Through this procedure, the system finally generates a non-uniform quadtree grid that matches the internal complexity of the region.

[0043] Based on this adaptive grid, the system further performs risk vorticity calculations to reveal structural risks. After the system generates a trend vector for each grid cell and then calculates the residual vector field, the risk vorticity calculation module initiates its core operation process, with the risk vorticity index ω... z The curl of the residual vector field is obtained by calculation. In one implementation, this calculation is achieved by numerically approximating the partial differential using the central difference method, and the formula is as follows: Among them, R x With R y Let ω represent the components of the residual vector in the x and y directions, (i,j) be the grid index, and Δx and Δy be the grid size at that location; after calculation, the system applies ω to the entire new region. zThe values ​​are normalized and mapped to the range of -1 to 1. The final assessment report includes, in addition to the binary value coordinates of each plot, a visualized heatmap rendered with the normalized risk vortex index. Positive values ​​represent a counter-clockwise rotation tendency in the value flow, while negative values ​​represent a clockwise rotation tendency. In the final output of this assessment, the visualized heatmap shows a strong positive vortex core at the boundary between the commercial core area and the residential area. Based on the value flow conflict indicated by this vortex core, the urban planning department adjusted the regional traffic management strategy and the layout of public service facilities to mitigate this structural risk. This result demonstrates that the risk vortex index generated by the present invention can provide a macro-level decision-making basis for land planning and management.

[0044] Example 5: The data standardization module of the present invention has a built-in transformation rule knowledge base constructed through an offline calibration and data filling procedure. This procedure first involves a group of experts in urban planning, environmental science, and land economics defining a seed dictionary containing core keywords, basic influence vectors, and sentiment modifiers. Subsequently, the system applies this seed dictionary to a large-scale historical data corpus containing land planning texts and market reports with known development results. When processing each historical document, the system uses its built-in optimization algorithm to compare the generated standardized influence vector with the actual development results of the land parcel. Based on the deviation between the two, the gradient descent method is used to iteratively fine-tune the influence vector values ​​corresponding to the keywords in the seed dictionary and the multiplier factors of the modifiers until the prediction deviation of the system on the entire corpus converges to below a preset minimum threshold, thereby completing the construction of the transformation rule knowledge base.

[0045] For deployments in different geographical regions, a pre-deployment calibration procedure is performed before the system is officially launched to determine the system adjustment constant k adapted to the local data environment. This procedure first collects multi-source heterogeneous data from the past three to five years in the deployment area as a calibration dataset. Subsequently, the system runs in calibration mode, repeatedly processing the calibration dataset and adjusting the value of the constant k in each iteration. This allows the system to fit a preset target distribution curve based on the overall statistical distribution of the confidence index C calculated from the entire calibration dataset. The target distribution curve is characterized by a median of 0.75 and a 5th percentile of 0.1. When the chi-square test value between the actual distribution and the target distribution is less than a predetermined tolerance, the iteration process terminates, and the currently used k value is determined as the final system adjustment constant for the deployment in that region.

[0046] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion, characterized in that, The system includes: The data standardization module is used to access multi-source heterogeneous data and transform each type of data into a standardized impact vector through pre-determined transformation rules. The transformation rules transform policy planning texts into impact vectors representing development orientation, environmental monitoring data into impact vectors representing ecological constraints, and market transaction data into impact vectors representing market activity. The information imbalance calculation module is used to project standardized influence vectors onto a geographic information grid with a unified spatial benchmark according to the geographic scope, and calculate the degree of dispersion of all projected influence vectors within each geographic grid to generate an information imbalance degree that characterizes the degree of information conflict. The information imbalance degree reflects the level of consistency between different data sources for the same land value judgment. The value coordinate generation module receives information imbalance and generates a confidence index based on the information imbalance using an inverse proportional function. The confidence index represents the credibility of the land value judgment, and its value is negatively correlated with the information imbalance. Furthermore, the value coordinate generation module calculates the arithmetic mean of all influence vectors within the geographic grid to generate a trend vector, which represents the development direction and intensity of land value. The system finally outputs a binary value coordinate consisting of the trend vector and the confidence index, which serves as the basis for land value assessment and transfer transactions.

2. The land value assessment and transfer transaction system based on multi-source heterogeneous data fusion according to claim 1, characterized in that, The information imbalance calculation module is specifically used to obtain standardized influence vectors and calculate the variance or standard deviation of all influence vectors within the geographic grid to obtain the information imbalance.

3. The land value assessment and transfer transaction system based on multi-source heterogeneous data fusion according to claim 1, characterized in that, In the value coordinate generation module, the inverse function for generating the confidence index is set as C = 1 / (1 + k × ID), where C represents the confidence index, ID represents the information imbalance degree, and k represents the system adjustment constant.

4. The land value assessment and transfer transaction system based on multi-source heterogeneous data fusion according to claim 1, characterized in that, In the data standardization module, the transformation rules generate positive, negative, or neutral influence vectors from government planning texts through keyword matching and sentiment analysis; the transformation rules generate influence vectors representing ecological constraints from PM2.5 concentration and water quality levels in environmental monitoring data after linear mapping; and the transformation rules generate influence vectors representing market activity from house price growth rate and rent growth rate in market transaction data.

5. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion as described in claim 1, characterized in that, The value coordinate generation module is also used to extract the time series data of the information imbalance when the information imbalance first reaches or exceeds the set warning threshold, and to calculate the change slope and duration that characterize the transient features of information conflict from the time series data; based on the change slope and duration, a dynamic decay coefficient is generated by a given nonlinear function; and the confidence index is modulated using the dynamic decay coefficient.

6. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion as described in claim 5, characterized in that, Given a nonlinear function α = e (-k·K·T) , where α represents the dynamic attenuation coefficient, K represents the slope of change, T represents the duration, and k represents the system adjustment constant.

7. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion as described in claim 1, characterized in that, It also includes: a trend vector field generation module, used to construct the original trend vector field from the trend vectors of all geographic grids; a expected trend field modeling module, used to perform Gaussian smoothing on the original trend vector field to generate the expected trend vector field; a residual vector field calculation module, used to subtract the expected trend vector field from the original trend vector field to obtain the residual vector field; and a risk vortex calculation module, used to calculate the curl of the residual vector field to generate a risk vortex index that characterizes the macro-structural risk of the region.

8. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion as described in claim 1, characterized in that, Multi-source heterogeneous data includes government planning data, market transaction data, population flow data, and environmental remote sensing data.

9. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion as described in claim 1, characterized in that, The system also includes a transaction matching module. When matching transactions, the transaction matching module receives binary value coordinates and recommends land plots with a confidence index higher than the preset threshold to users based on the comparison results between the confidence index and the preset threshold, or provides risk warnings for land plots with a confidence index lower than the preset threshold. The level of risk warning is related to the degree of decrease in the confidence index.

10. A land value assessment and transfer transaction system based on multi-source heterogeneous data fusion according to claim 1, characterized in that, The granularity of geographic information grid division can be dynamically adjusted based on the actual geographical characteristics of the land, the requirements of regional development planning, and the evaluation needs input by users.

Images