Geospatial data geometric topological relation evaluation method and system based on large language model

Abstract

The invention discloses a geospatial data geometric topological relation evaluation method and system based on a large language model, and belongs to the technical field of spatio-temporal information processing, and the method comprises the following steps: S1, data preprocessing and model construction: processing a training data set containing geospatial data and a corresponding geometric topological relation thereof, and constructing a model; the geometric topological relation of the geographic space data is coded into text or vector representation which can be understood by a large language model; the coded training data is utilized to finely adjust a pre-trained large language model, so that the pre-trained large language model can evaluate a topological relation according to input geographic space data description; s2, geographic space data input and feature extraction; and S3, large language model reasoning and topological relation evaluation. According to the method, by combining the language understanding ability of the large model and the spatial topology calculation technology, intelligent and high-precision evaluation of the geographic spatial data geometrical relationship is achieved, and a brand-new and self-adaptive evaluation technology is provided for the field of geographic spatial data processing.

Description

Technical Field

[0001] This invention relates to the field of spatiotemporal information processing technology, specifically to a method and system for evaluating the geometric topological relationships of geospatial data based on a large language model. Background Technology

[0002] Geospatial topology is a core means of describing the spatial relationships between geographic features, and traditional methods mainly rely on the spatial analysis functions of Geographic Information Systems (GIS). The technical approach typically involves abstracting geographic features (such as points, lines, and polygons) into vector or raster data to construct a spatial database. Alternatively, it may be based on theories such as the DE-9IM model to clarify topological relationships such as adjacency, containment, and intersection; or it may use operations such as buffer analysis and overlay analysis to calculate the spatial relationships between features.

[0003] However, traditional methods have the following significant limitations. First, they have low automation, requiring manual definition of analysis rules and making them difficult to adapt to complex scenarios. Second, they lack semantic understanding; current traditional methods only support geometric calculations and cannot understand spatial relationships described in natural language. Finally, traditional methods suffer from efficiency bottlenecks; processing large-scale spatial data is time-consuming and cannot meet real-time requirements. Summary of the Invention

[0004] The technical objective of this invention is to address the above-mentioned shortcomings by providing a method and system for evaluating the geometric topological relationships of geospatial data based on a large language model. By combining the language understanding capabilities of the large model with spatial topological calculation technology, it achieves intelligent and high-precision evaluation of the geometric relationships of geospatial data, providing a novel and adaptive evaluation technology for the field of geospatial data processing.

[0005] The technical solution adopted by this invention to solve its technical problem is:

[0006] A method for evaluating geometric topological relationships in geospatial data based on a large language model. The implementation of this method includes the following steps:

[0007] S1: Data preprocessing and model building, including:

[0008] The training dataset containing geospatial data and its corresponding geometric topological relationships is processed to encode the geometric topological relationships of the geospatial data into text or vector representations that can be understood by the Large Language Model (LLM).

[0009] Fine-tuning a pre-trained large language model using encoded training data enables it to evaluate topological relationships based on input geospatial data descriptions.

[0010] S2: Geospatial data input and feature extraction, including:

[0011] Receive geospatial data to be evaluated and extract the geometric features of the geospatial data;

[0012] The geometric features are textualized into a natural language description and then transformed into feature vectors representing geographic elements.

[0013] S3: Large Language Model Reasoning and Topological Relationship Evaluation, including:

[0014] The feature vectors of two or more geographic elements to be evaluated are used as input, or their textual descriptions are combined into a complete sentence and input into a finely tuned large language model.

[0015] The large language model performs inference based on the input, evaluates and outputs the geometric topological relationships between the geographic elements.

[0016] Furthermore, in step S1, the geometric topological relationships of the geospatial data include: intersection, containment, and adjacency.

[0017] Furthermore, in step S2, the geometric features include at least one of the following: coordinates, bounding box, centroid, area, and perimeter.

[0018] Furthermore, step S2 is specifically implemented as follows:

[0019] S2.1 Data Input: Receive geospatial data to be evaluated, including single or multiple geographic features input in various formats (such as GeoJSON, Shapefile);

[0020] S2.2 Textual Description: The input geospatial data is described in textual form, including extracting the geometric features of geographic elements (such as coordinates, bounding boxes, centroids, areas, perimeters, etc.) and converting them into natural language text; for example, a polygon can be described as "a polygon with an area of ​​X, a centroid located at Y, and a bounding box of Z".

[0021] S2.3 Feature Vectorization: Input the textual description into the pre-trained embedding model to generate a feature vector representing the geographic feature; this vector captures the geometric and spatial attributes of the geographic feature.

[0022] Furthermore, in step S3, the evaluation result output by the large language model is a probability distribution or prediction result about the topological relationship, and further outputs an evaluation report containing the topological relationship type and optional confidence score.

[0023] Furthermore, if the output of the large language model is a probability distribution, then the one with the highest probability is selected as the final result.

[0024] Furthermore, the method also includes result correction and feedback optimization, specifically including:

[0025] The results of the large language model evaluation are compared with the topological relationships calculated by traditional algorithms to obtain the differences;

[0026] The discrepancies identified during the comparison are used as new training data to incrementally train or fine-tune the large language model.

[0027] This invention also claims protection for a geospatial data geometric topology relationship evaluation system based on a large language model, comprising:

[0028] The data preprocessing and model building module is used to collect and process data, encode topological relationships, and fine-tune the model using labeled training data.

[0029] The geospatial data input and feature extraction module is used to extract the geometric features of geospatial data, convert it into a natural language description, and transform it into a feature vector representing geographic elements.

[0030] The large language model reasoning and topological relationship evaluation module is used to combine the feature vectors or textual descriptions of two or more geographic elements to be evaluated and input them into the fine-tuned large language model. The large language model performs reasoning based on the input, evaluates and outputs the geometric and topological relationships between geographic elements.

[0031] The model validation module is used to compare the evaluation results of the large language model with the topological relationships calculated by traditional algorithms, and use the differences after comparison as new training data to perform incremental training or fine-tuning of the large language model.

[0032] This system can implement the methods described above.

[0033] The present invention also claims a device for evaluating the geometric topology of geospatial data based on a large language model, comprising: at least one memory and at least one processor;

[0034] The at least one memory is used to store a machine-readable program;

[0035] The at least one processor is used to call the machine-readable program to implement the above method.

[0036] The present invention also claims protection for a computer-readable medium storing computer instructions that, when executed by a processor, enable the implementation of the above-described method.

[0037] Compared with existing technologies, the method and system for evaluating the geometric topological relationships of geospatial data based on a large language model of the present invention have the following advantages:

[0038] This invention leverages the powerful natural language understanding and reasoning capabilities of large language models to transform complex geometric computation problems into semantic and feature-based reasoning problems. By converting the geometric features of geographic elements into textual descriptions, the model can "understand" spatial relationships just like it understands language, thereby achieving more flexible and intelligent evaluation. Attached Figure Description

[0039] Figure 1 This is a schematic diagram illustrating the principle of the geospatial data geometric topology relationship evaluation method based on a large language model provided in this embodiment of the invention. Detailed Implementation

[0040] This invention provides a method for evaluating the geometric topological relationships of geospatial data based on a large language model. The implementation of this method includes the following steps:

[0041] S1: Data preprocessing and model building, including:

[0042] The training dataset containing geospatial data and its corresponding geometric topological relationships is processed to encode the geometric topological relationships of the geospatial data into text or vector representations that can be understood by the Large Language Model (LLM).

[0043] The pre-trained large language model is fine-tuned using encoded training data so that it can evaluate topological relationships based on the input geospatial data description.

[0044] S2: Geospatial data input and feature extraction, including:

[0045] Receive geospatial data to be evaluated and extract the geometric features of the geospatial data;

[0046] The geometric features are textualized into a natural language description and then transformed into feature vectors representing geographic elements.

[0047] S3: Large Language Model Reasoning and Topological Relationship Evaluation, including:

[0048] The feature vectors or textual descriptions of two or more geographic elements to be evaluated are combined and input into the fine-tuned large language model.

[0049] The large language model performs inference based on the input, evaluates and outputs the geometric topological relationships between the geographic elements.

[0050] The evaluation results output by the large language model are the probability distribution or prediction results of topological relationships, and further output an evaluation report including the type of topological relationship and optional confidence scores.

[0051] S4: Result correction and feedback optimization, including:

[0052] The results of the large language model evaluation are compared with the topological relationships calculated by traditional algorithms;

[0053] The discrepancies identified during the comparison are used as new training data to incrementally train or fine-tune the large language model.

[0054] Combined with appendix Figure 1 As shown, the specific implementation process of this method is as follows:

[0055] S1. Data preprocessing and model building.

[0056] S1.1 Data Collection and Cleaning: Collect a training dataset containing geospatial data (such as point, line, and polygon features) and their corresponding geometric topological relationships. Clean the data to remove noise, null values, and inconsistent records.

[0057] S1.2 Topological Relationship Encoding: The geometric topological relationships of geospatial data (such as "intersection", "containment", "adjacency", etc.) are labeled and converted into text or vector representations that can be understood by a Large Language Model (LLM). For example, "A contains B" can be encoded as the text "Feature A contains Feature B", or converted into a specific vector.

[0058] S1.3 Model Fine-tuning: Select a pre-trained large language model, such as DeepSeek or Qwen, as the base model. Fine-tune the model using labeled training data to enable it to understand the geometric features and topological relationships of geospatial data and learn how to evaluate these relationships based on input data. During fine-tuning, the model's input is a description of the geospatial data, and its output is the corresponding topological relationship evaluation results.

[0059] S2. Geospatial data input and feature extraction.

[0060] S2.1 Data Input: Receive geospatial data to be evaluated, which can be single or multiple geographic features input in various formats (such as GeoJSON, Shapefile).

[0061] S2.2 Textual Description: This involves providing a textual description of the input geospatial data. This includes extracting the geometric features of geographic elements (such as coordinates, bounding boxes, centroids, areas, perimeters, etc.) and converting them into natural language text. For example, a polygon can be described as "a polygon with an area of ​​X, a centroid at Y, and a bounding box at Z."

[0062] S2.3 Feature Vectorization: The textual description is input into a pre-trained embedding model to generate a feature vector representing the geographic feature. This vector captures the geometric and spatial attributes of the geographic feature.

[0063] S3, Large Language Model Reasoning and Topological Relationship Evaluation.

[0064] S3.1 Input Combination: Take the feature vectors of two or more geographic elements to be evaluated as input, or combine their textual descriptions into a complete sentence, and input them into the fine-tuned large language model.

[0065] S3.2 Topological Relationship Inference: Based on its learned knowledge, the large language model analyzes the input feature vectors or text descriptions, infers and evaluates the geometric topological relationships between these geographic elements. The model outputs a probability distribution of the topological relationships or directly provides a prediction result, such as "intersecting," "containing," or "separating."

[0066] S3.3 Result Validation and Output: Validate the evaluation results output by the large language model. If the model outputs a probability distribution, select the highest probability as the final result. The final output is an evaluation report, which includes the evaluated topological relationship type and an optional confidence score.

[0067] S4. Model Validation.

[0068] S4.1 Result Feedback: Compare the results of the large language model evaluation with the actual topological relationships calculated by traditional algorithms (such as the nine-way intersection model) to obtain the differences.

[0069] S4.2 Error Analysis: Analyze cases of model evaluation errors to identify the causes of the errors, such as inaccurate data descriptions or insufficient understanding of specific complex situations by the model.

[0070] S4.3 Model Iterative Optimization: Using erroneous cases as new training data, the large language model is incrementally trained or fine-tuned again to improve its evaluation accuracy and generalization ability. This is a continuous optimization process aimed at making the model perform better when dealing with more complex and diverse geospatial data.

[0071] This method integrates natural language processing capabilities with the nine-intersection model algorithm to achieve efficient and accurate assessment of topological relationships between geospatial objects. It can be widely applied in fields such as land spatial planning, smart city management, and geographic information system (GIS) development.

[0072] This invention also provides a geospatial data geometric topology relationship evaluation system based on a large language model, comprising:

[0073] The data preprocessing and model building module is used to collect and process data, encode topological relationships, and fine-tune the model using labeled training data.

[0074] The geospatial data input and feature extraction module is used to extract the geometric features of geospatial data, convert it into a natural language description, and transform it into a feature vector representing geographic elements.

[0075] The large language model reasoning and topological relationship evaluation module is used to combine the feature vectors or textual descriptions of two or more geographic elements to be evaluated and input them into the fine-tuned large language model. The large language model performs reasoning based on the input, evaluates and outputs the geometric and topological relationships between geographic elements.

[0076] The model validation module is used to compare the evaluation results of the large language model with the topological relationships calculated by traditional algorithms, and use the differences after comparison as new training data to perform incremental training or fine-tuning of the large language model.

[0077] This system can implement the geospatial data geometric topology relationship evaluation method based on a large language model as described in the above embodiments.

[0078] The data preprocessing and model building module includes:

[0079] Data collection and cleaning: Collect a training dataset containing geospatial data (such as point, line, and polygon features) and their corresponding geometric topological relationships. Clean the data to remove noise, null values, and inconsistent records.

[0080] Topological relationship encoding: This involves labeling the geometric topological relationships (such as "intersection", "containment", "adjacency", etc.) of geospatial data and converting them into text or vector representations that can be understood by a Large Language Model (LLM). For example, "A contains B" can be encoded as the text "Feature A contains Feature B", or converted into a specific vector.

[0081] Model fine-tuning: Select a pre-trained large language model, such as DeepSeek or Qwen, as the base model. Fine-tune the model using labeled training data to enable it to understand the geometric features and topological relationships of geospatial data and learn how to evaluate these relationships based on input data. During fine-tuning, the model's input is a description of the geospatial data, and its output is the corresponding topological relationship evaluation results.

[0082] The geospatial data input and feature extraction module includes:

[0083] Data Input: Receive geospatial data to be evaluated, which can be single or multiple geographic features input in various formats (such as GeoJSON, Shapefile).

[0084] Textual description: This involves providing a textual description of the input geospatial data. This includes extracting the geometric features of geographic elements (such as coordinates, bounding boxes, centroids, areas, perimeters, etc.) and converting them into natural language text. For example, a polygon can be described as "a polygon with an area of ​​X, a centroid at Y, and a bounding box at Z".

[0085] Feature vectorization: The textual description is input into a pre-trained embedding model to generate a feature vector representing the geographic feature. This vector captures the geometric and spatial attributes of the geographic feature.

[0086] The large language model reasoning and topological relation evaluation module includes:

[0087] Input combination: Take the feature vectors of two or more geographic features to be evaluated as input, or combine their textual descriptions into a complete sentence, and input it into a fine-tuned large language model.

[0088] Topological relationship inference: Based on its learned knowledge, the large language model analyzes the input feature vectors or text descriptions to infer and evaluate the geometric topological relationships between these geographic elements. The model outputs a probability distribution of the topological relationships or directly provides a prediction result, such as "intersecting," "containing," or "separating."

[0089] Results Validation and Output: Validate the evaluation results output by the large language model. If the model output is a probability distribution, select the highest probability as the final result. The final output is an evaluation report, which includes the evaluated topological relationship type and an optional confidence score.

[0090] The model validation module includes:

[0091] Results feedback: The results of the large language model evaluation are compared with the actual topological relationships calculated by traditional algorithms (such as the nine-way intersection model) to obtain the differences.

[0092] Error analysis: Analyze cases where the model evaluation is incorrect to identify the causes of the errors, such as inaccurate data descriptions or insufficient understanding of specific complex situations by the model.

[0093] Model iterative optimization: Error cases are used as new training data to incrementally train or fine-tune the large language model to improve its evaluation accuracy and generalization ability. This is a continuous optimization process aimed at making the model perform better when dealing with more complex and diverse geospatial data.

[0094] This invention also provides a device for evaluating the geometric topology of geospatial data based on a large language model, comprising: at least one memory and at least one processor;

[0095] The at least one memory is used to store a machine-readable program;

[0096] The at least one processor is used to call the machine-readable program to implement the geospatial data geometric topology relationship evaluation method based on a large language model as described in the above embodiments.

[0097] This invention also provides a computer-readable medium storing computer instructions. When executed by a processor, the computer instructions cause the processor to perform the geospatial data geometric topology relationship evaluation method based on a large language model as described in the above embodiments. Specifically, a system or apparatus equipped with a storage medium storing software program code that implements the functions of any of the embodiments described above, and enabling the computer (or CPU or MPU) of the system or apparatus to read and execute the program code stored in the storage medium.

[0098] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0099] Examples of storage media used to provide program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0100] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0101] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion unit connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion unit execute some and all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0102] The present invention has been shown and described in detail above with reference to the accompanying drawings and preferred embodiments. However, the present invention is not limited to these disclosed embodiments. Based on the above multiple embodiments, those skilled in the art will know that more embodiments of the present invention can be obtained by combining the code review methods in the different embodiments above. These embodiments are also within the protection scope of the present invention.

Claims

1. A method for evaluating geometric topological relationships in geospatial data based on a large language model, characterized in that, The implementation of this method includes the following steps: S1: Data preprocessing and model building, including: The training dataset containing geospatial data and its corresponding geometric topological relationships is processed to encode the geometric topological relationships of the geospatial data into text or vector representations that can be understood by a large language model. Fine-tuning a pre-trained large language model using encoded training data enables it to evaluate topological relationships based on input geospatial data descriptions. S2: Geospatial data input and feature extraction, including: Receive geospatial data to be evaluated and extract the geometric features of the geospatial data; The geometric features are textualized into a natural language description and then transformed into feature vectors representing geographic elements. S3: Large Language Model Reasoning and Topological Relationship Evaluation, including: The feature vectors of two or more geographic elements to be evaluated are used as input, or their textual descriptions are combined into a complete sentence and input into a finely tuned large language model. The large language model performs inference based on the input, evaluates and outputs the geometric topological relationships between the geographic elements.

2. The method for evaluating the geometric topological relationships of geospatial data based on a large language model according to claim 1, characterized in that, In step S1, the geometric topological relationships of the geospatial data include: intersection, containment, and adjacency.

3. The method for evaluating the geometric topological relationships of geospatial data based on a large language model according to claim 1, characterized in that, In step S2, the geometric features include at least one of the following: coordinates, bounding box, centroid, area, and perimeter.

4. The method for evaluating the geometric topological relationships of geospatial data based on a large language model according to claim 1, characterized in that, The specific implementation process of step S2 is as follows: S2.1 Data Input: Receive geospatial data to be evaluated, including single or multiple geographic features entered in various formats; S2.2 Textual Description: The input geospatial data is described in textual form, including extracting the geometric features of geographic elements and converting them into natural language text; S2.3 Feature Vectorization: Input the textual description into the pre-trained embedding model to generate a feature vector representing the geographic feature; this vector captures the geometric and spatial attributes of the geographic feature.

5. The method for evaluating the geometric topological relationships of geospatial data based on a large language model according to claim 1, characterized in that, In step S3, the evaluation result output by the large language model is a probability distribution or prediction result of the topological relationship, and further outputs an evaluation report containing the topological relationship type and optional confidence score.

6. The method for evaluating the geometric topological relationships of geospatial data based on a large language model according to claim 5, characterized in that, If the output of the large language model is a probability distribution, then the one with the highest probability is selected as the final result.

7. The method for evaluating the geometric topological relationships of geospatial data based on a large language model according to claim 1, characterized in that, This method also includes result correction and feedback optimization, specifically including: The results of the large language model evaluation are compared with the topological relationships calculated by traditional algorithms to obtain the differences; The discrepancies identified during the comparison are used as new training data to incrementally train or fine-tune the large language model.

8. A geospatial data geometric topological relationship evaluation system based on a large language model, characterized in that, include: The data preprocessing and model building module is used to collect and process data, encode topological relationships, and fine-tune the model using labeled training data. The geospatial data input and feature extraction module is used to extract the geometric features of geospatial data, convert it into a natural language description, and transform it into a feature vector representing geographic elements. The large language model reasoning and topological relationship evaluation module is used to combine the feature vectors or textual descriptions of two or more geographic elements to be evaluated and input them into the fine-tuned large language model. The large language model performs reasoning based on the input, evaluates and outputs the geometric and topological relationships between geographic elements. The model validation module is used to compare the evaluation results of the large language model with the topological relationships calculated by traditional algorithms, and use the differences after comparison as new training data to perform incremental training or fine-tuning of the large language model. This system can implement the method described in any one of claims 1 to 7.

9. A device for evaluating the geometric topological relationships of geospatial data based on a large language model, characterized in that, include: At least one memory and at least one processor; The at least one memory is used to store a machine-readable program; The at least one processor is configured to invoke the machine-readable program to implement the method according to any one of claims 1 to 7.

10. A computer-readable medium, characterized in that, The computer-readable medium stores computer instructions that, when executed by a processor, enable the implementation of the method described in any one of claims 1 to 7.

Images