System cue word generation method and system in urban and rural planning scene, terminal and medium

Abstract

The invention relates to a system cue word generation method and system in an urban and rural planning scene, a terminal and a medium, and relates to the technical field of large language models.The method comprises the steps that in response to query operation, a planning scene is extracted from an input text, and the planning scene comprises urban and rural planning, traffic planning, municipal planning and the like; acquiring role positioning and task instructions according to the planning scene; determining a planning database corresponding to the planning scene; performing type labeling and semantic annotation on the planning database to obtain a data specification; identifying an analysis task in the planning scene; setting a custom function for the analysis task; and integrating the role location, the task instruction, the data specification and the custom function to obtain a system cue word. The method has the effect of improving the processing efficiency of the system cue word.

Description

Technical Field

[0001] This application relates to the field of large language model technology, and in particular to a method, system, terminal and medium for generating system prompt words in urban and rural planning scenarios. Background Technology

[0002] System prompts are initial instructions given to the large language model to set its role, behavioral guidelines, and response style.

[0003] When generating system prompts, related technologies first need to determine the target user group and the user's core needs for the large language model. Then, based on this target group and core needs, the corresponding statements, processing flows, and visualization methods for querying the database are designed, and these statements are integrated into the system prompts.

[0004] Regarding the aforementioned technologies, the system prompts suffer from low processing efficiency in urban and rural planning scenarios. Summary of the Invention

[0005] To improve the processing efficiency of system prompts, this application provides a method, system, terminal, and medium for generating system prompts in urban and rural planning scenarios.

[0006] Firstly, this application provides a method for generating system prompts in urban and rural planning scenarios, employing the following technical solution: A method for generating system prompts in an urban and rural planning scenario includes: In response to query operations, planning scenarios are extracted from the input text, including urban and rural planning, transportation planning, and municipal planning. Based on the planned scenario, obtain the role positioning and task instructions; Determine the planning database corresponding to the planning scenario; The planning database is labeled with types and semantic annotations to obtain data specifications; Based on the input text and the planning scenario, an analysis task is generated; Set a custom function for the analysis task; By combining the aforementioned role positioning, task instructions, data specifications, and custom functions, system prompts are obtained.

[0007] By employing the aforementioned technical solution, system prompts are generated by extracting planning scenarios and obtaining corresponding role positioning, task instructions, data specifications, and custom functions. This method ensures that the prompts closely align with specific planning needs, improving the relevance and accuracy of the generated prompts. Simultaneously, it achieves automated generation, significantly reducing manual intervention and improving efficiency and reliability, making it suitable for diverse planning scenarios.

[0008] Optionally, code can be generated based on the system prompts to analyze tasks, process data, invoke tools, and visualize data. Executable code is generated based on the code analysis task, the processed data, the tool calls, and the visualization. The executable code is combined to form an executable script; The executable script is executed to obtain the output report corresponding to the input text.

[0009] By adopting the above technical solution, system prompts are converted into executable scripts, verified for execution, and finally output reports are generated. This process automates the conversion from high-level task descriptions to executable code, ensuring a high degree of consistency between the code and task requirements, improving the practicality of code generation, operational reliability, and result accuracy, while reducing development costs.

[0010] Optionally, based on the planned scenario, the professional role identity during the interaction process is set, and the role positioning is generated; According to the planning scenario, construct an instruction type outline, which includes tool priority, code style, data verification, and analysis results; According to the instruction type outline, the task instruction is obtained by searching in the preset instruction library.

[0011] By adopting the above technical solution, role positioning and task instructions are generated by setting professional role identities and searching a preset instruction library according to the instruction type outline. This method enables the system to simulate the behavior of domain experts, improves the professionalism of task execution, ensures the comprehensiveness and standardization of instructions, optimizes the generation process, and improves the system's practicality and reliability.

[0012] Optionally, according to the planned scenario, operation instructions are extracted from a preset instruction library to obtain the instruction type outline.

[0013] By adopting the above technical solution, an instruction type outline is constructed by directly extracting operation instructions from a preset instruction library. This method simplifies the construction process, improves generation efficiency and accuracy, ensures the standardization and applicability of instructions, reduces user intervention, enables the system to quickly adapt to different planning scenarios, and enhances flexibility and consistency.

[0014] Optionally, data information can be extracted from the planning database, including field names, data types, and business descriptions; Based on the planning scenario, the business description is standardized to obtain the business meaning; Establish a mapping relationship between the field names, the data types, and the business meanings to form the data specifications.

[0015] By adopting the above technical solution, data information is extracted and standardized to establish mapping relationships and form data specifications. This method comprehensively captures the structure and semantics of the data, eliminates ambiguity in business terms, ensures data consistency, understandability, and accuracy in subsequent processing, and provides reliable data support for the system.

[0016] Optionally, the analysis task can be divided into complex analysis tasks and simple analysis tasks based on its complexity. Based on the complex analysis task, extract complex analysis functions from a custom analysis tool library; Based on the simple analysis task, extract simple analysis functions from the standard tool library; The complex analysis function and the simple analysis function are combined to obtain the custom function.

[0017] By adopting the above technical solution, functions are extracted and integrated from both custom and standard libraries based on the complexity of the analysis task. This method enables fine-grained task management and optimized resource allocation, allowing the system to flexibly adapt to different analysis needs, improving analytical capabilities and adaptability, while reducing unnecessary computational overhead and enhancing performance.

[0018] Optionally, the complex analysis functions include: dynamic scale OD calculation functions, proximity analysis functions for different spatial elements, and passenger flow corridor generation.

[0019] By adopting the above technical solutions and introducing dynamic-scale OD calculation functions and spatial element proximity analysis functions, the system's spatial analysis capabilities are significantly enhanced. These specialized functions can automatically adapt to different planning scales and accurately calculate spatial proximity, providing crucial decision support for spatial planning and improving the accuracy and practicality of the output results.

[0020] Secondly, this application provides a system prompt word generation system for urban and rural planning scenarios, which adopts the following technical solution, please refer to... Figure 4 : A system for generating system prompts in a planning scenario, comprising: The acquisition module is used to acquire query operations and input text; A memory for storing the program for generating system prompts in the planning scenario; The processor and the program in the memory can be loaded and executed by the processor to implement the system prompt word generation method in the planned scenario.

[0021] By employing the aforementioned technical solution, system prompts are generated by extracting planning scenarios and obtaining corresponding role positioning, task instructions, data specifications, and custom functions. This method ensures that the prompts closely align with specific planning needs, improving the relevance and accuracy of the generated prompts. Simultaneously, it achieves automated generation, significantly reducing manual intervention and improving efficiency and reliability, making it suitable for diverse planning scenarios.

[0022] Thirdly, this application provides a smart terminal, which adopts the following technical solution: A smart terminal includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute the method described in any one of the above.

[0023] Fourthly, this application provides a computer storage medium capable of storing corresponding programs, which facilitates improving the processing efficiency of system prompts, and adopts the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed by any of the above-described methods for generating system prompts, systems, terminals, and media in urban and rural planning scenarios.

[0024] In summary, this application includes at least one of the following beneficial technical effects: 1. By extracting planning scenarios and obtaining corresponding role positioning, task instructions, data specifications, and custom functions, system prompts are generated. This method ensures that the prompts closely match specific planning needs, improves the relevance and accuracy of the generated prompts, and achieves automated generation, significantly reducing manual intervention, improving efficiency and reliability, and is suitable for diverse planning scenarios; 2. Convert system prompts into executable scripts, verify their execution, and finally generate an output report. This process automates the conversion from high-level task descriptions to executable code, ensuring a high degree of consistency between the code and task requirements, improving the usability, reliability, and accuracy of the generated code, and reducing development costs. 3. By setting professional role identities and searching a preset instruction library according to the instruction type outline, role positioning and task instructions are generated. This method enables the system to simulate the behavior of domain experts, improving the professionalism of task execution, while ensuring the comprehensiveness and standardization of instructions, optimizing the generation process, and improving the system's practicality and reliability. Attached Figure Description

[0025] Figure 1 This is a schematic diagram illustrating an implementation scenario of a system prompt word generation method in an urban and rural planning context provided in this application embodiment.

[0026] Figure 2This is a flowchart illustrating a method for generating system prompts in an urban and rural planning scenario, as provided in an embodiment of this application. Figure 3 This is an example of a system prompt word provided in an embodiment of this application.

[0027] Figure 4 This application provides a schematic diagram of the structure of a system prompt word generation system in an urban and rural planning scenario. Detailed Implementation

[0028] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figure 1 To be continued Figure 4 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0029] This application discloses an application scenario for a system prompt word generation method in urban and rural planning scenarios. Specifically, the planning scenario in this application refers to urban and rural planning scenarios, and this application is applied to a large language model, making the output of the large language model more consistent with the planning scenario.

[0030] For example, when the large language model is working, it generates system prompts based on the user's input text and provides these prompts to the large language model, making its output more relevant to the urban and rural planning scenario. The large language model can run on user terminal 11 or server 12. For instance, when the large language model runs on user terminal 11, after receiving the input text, user terminal 11 calls the system prompt generation method for planning scenarios provided in this application to generate system prompts and provides them to the large language model. Guided by the system prompts, the large language model generates an output report based on the input text and displays the output report on user terminal 11. Alternatively, when the large language model runs on server 12, the user inputs text into user terminal 11. User terminal 11 can transmit the input text to server 12. Server 12 calls the system prompt generation method for planning scenarios to generate system prompts and provides them to the large language model. Guided by the system prompts, server 12 generates an output report based on the input text and sends the output report to user terminal 11, where it is displayed.

[0031] In summary, this application constructs a system prompt word body composed of four parts: role positioning, data specification, task instructions, and custom function calls. This effectively integrates an understanding of spatiotemporal analysis tools (such as OD matrix calculation), thereby improving the processing efficiency and reliability of complex tasks. Simultaneously, by incorporating professional knowledge from standards, specifications, and existing planning results, structured prompt words guide the large language model to output professional answers that conform to industry standards, enhancing the readability and professionalism of the analysis results.

[0032] This application discloses a method for generating system prompts in urban and rural planning scenarios. (Refer to...) Figure 2 The method includes: Step S201: In response to the query operation, extract the planning scenario from the input text. The planning scenario includes urban and rural planning, transportation planning and municipal planning.

[0033] The query operation is used to submit input text. In this embodiment, the input text is related to urban and rural planning, for example, the input text is "What is the total number of trips in City A?".

[0034] The planning scenario refers to the external environment or internal conditions corresponding to the input text. In this embodiment, the planning scenario is limited to urban and rural planning scenarios.

[0035] In some other embodiments, the planning scenario is pre-set and does not need to be obtained by inputting text.

[0036] Step S202: Based on the planned scenario, obtain the role positioning and task instructions.

[0037] The role is defined as an expert in a specific subfield of urban and rural planning, thereby enhancing the large language model's understanding of professional terminology and industry standards. For example, the role could be defined as a transportation planner or urban designer.

[0038] In some other implementations, the role positioning is pre-set, or the role positioning can be manually selected by the user. It should be noted that the role positioning will not change during a single question-and-answer session.

[0039] It should be noted that the system prompts generated in this application are used to guide the large language model in generating executable scripts. Therefore, task instructions are used to set the operational specifications for the executable scripts. For example, task instructions include, but are not limited to, at least one of tool priority, code style, data validation, and analysis results.

[0040] Tool priority refers to the priority at which an executable script calls a tool.

[0041] Code style refers to the mandatory requirements for the writing format, naming conventions, and structural organization of executable scripts. For example, using a consistent naming format for variables within an executable script, and adding concise comments to key steps or complex logic sections.

[0042] Data validation refers to checking relevant data before and after the execution of an executable script. For example, before the executable script is executed, it checks whether the relevant data exists; during the execution of the executable script, it determines whether the values ​​of the output data are reasonable; and after the executable script is executed, it checks whether the structure is complete.

[0043] The analysis results refer to the mandatory specifications for the output of the executable script. Optionally, the analysis results can specify the format, structure, and content. For example, the analysis results can restrict the output format of the executable script to a structured dictionary; or, the analysis results can restrict the output of the executable script to include both charts and text.

[0044] Optionally, according to the planned scenario, operation instructions can be extracted from a preset instruction library to obtain an instruction type outline.

[0045] In some embodiments, the effectiveness of role positioning in narrowing the problem-solving space can also be measured, then H_role = -Σ(P(a_i|R,S)*logP(a_i|R,S)) + Σ(P(a_i|S)*logP(a_i|S)), where H_role is the information entropy reduction brought about by role positioning, P(a_i|S) is the prior probability of the system taking an action a_i (such as calling a function or adopting a certain analysis approach) when only the planning scenario S is known, and P(a_i|R,S) is the posterior probability of the system taking action a_i when the role positioning R and the planning scenario S are known. The larger the value of the information entropy reduction, the greater the contribution of the role positioning to focusing on the problem and reducing decision-making confusion. Step S203: Determine the planning database corresponding to the planning scenario.

[0046] In this embodiment of the application, there are at least two planning databases corresponding to different planning scenarios.

[0047] Planning databases are multi-source heterogeneous datasets. Planning datasets include datasets from different sources and in different formats. For example, planning dataset A comes from website 1, and planning dataset B comes from website 2; another example is that planning dataset C stores data in text format, and fixed dataset D stores data in numerical format.

[0048] Step S204: Perform type labeling and semantic annotation on the planning database to obtain data specifications.

[0049] Optionally, data information is extracted from the planning database, including field names, data types, and business descriptions. Based on the planning scenario, the business descriptions are standardized to obtain their business meanings. A mapping relationship is established between field names, data types, and business meanings to form data specifications.

[0050] To ensure the standardization and professionalism of the business descriptions in the planning scenario, the business descriptions need to be standardized to obtain business descriptions that are more in line with the planning scenario.

[0051] In some embodiments, after obtaining the data specifications, the completeness of the data specifications can be evaluated. For example, using DSI to represent the completeness index of the data specifications, we have DSI = w1 * (N_annotated / N_total) + w2 * C_standardization, where w1 and w2 are weighting coefficients, N_annotated is the number of fields with completed type annotation and semantic annotation, N_total is the total number of fields, and C_standardization is the degree of standardization of business descriptions, which is the number of fields of standard terms / N_annotated.

[0052] Step S205: Generate an analysis task based on the input text and the planned scenario.

[0053] For example, a set of analysis tasks corresponding to the planning scenarios is determined, with each analysis task set corresponding one-to-one with a planning scenario. The semantics of the input text are obtained. Analysis tasks are then determined from the set of analysis tasks based on the semantics.

[0054] For example, in an urban and rural planning scenario where the input text is "What is the total number of trips in City A?", the analysis task could include counting the total number of trips in City A, calculating the changes in the total number of trips in City A, and calculating the changes in the age of trippers in City A.

[0055] Step S206: Set up a custom function for the analysis task.

[0056] Optionally, based on the complexity of the analysis task, the analysis task can be divided into complex analysis tasks and simple analysis tasks. For complex analysis tasks, complex analysis functions are extracted from a custom analysis tool library. For simple analysis tasks, simple analysis functions are extracted from a standard tool library. The complex and simple analysis functions are then integrated to obtain the custom function.

[0057] Optional, complex analysis functions include: dynamic scale OD calculation functions, proximity analysis functions for different spatial elements, and generation of passenger flow corridors, etc.

[0058] Dynamic-scale OD calculation functions are algorithms or functions used to adaptively calculate and aggregate origin-end traffic at different spatial levels, such as communities, streets, and administrative districts.

[0059] Proximity analysis functions for different spatial elements are geographic computation methods used to quantify the proximity relationships and accessibility between different types of spatial objects (points, lines, and polygons). Through buffer analysis, nearest neighbor search, cost path analysis, and other techniques, they transform qualitative "locational relationships" into quantitative "spatial indicators," serving as a fundamental tool for assessing facility service range, resource accessibility, and spatial interactions.

[0060] Passenger flow corridors are identified through density calculation, pattern recognition, and spatial clustering, extracting major passenger flow channels with high flow rates and continuity. Passenger flow corridors can visually reveal the dominant direction and intensity of pedestrian movement in a city or region.

[0061] Step S207: Combine role positioning, task instructions, data specifications, and custom functions to obtain system prompts.

[0062] Furthermore, after receiving the system prompts, the system generates code analysis tasks, processes data, invokes tools, and visualizes the results. Based on these prompts, executable code is generated. This executable code is then combined to form an executable script. Executing the executable script yields an output report corresponding to the input text.

[0063] In some implementations, guided by system prompts, the code analysis process is planned based on the code analysis task and data processing to obtain executable code. This includes generating executable code related to visualization requirements and executable code related to tool invocation requirements. Since executable code is composed of multiple individual lines of code, it needs to be combined into an executable script to ensure its executability.

[0064] By employing the aforementioned technical solution, system prompts are generated by extracting planning scenarios and obtaining corresponding role positioning, task instructions, data specifications, and custom functions. This method ensures that the prompts closely align with specific planning needs, improving the relevance and accuracy of the generated prompts. Simultaneously, it achieves automated generation, significantly reducing manual intervention and improving efficiency and reliability, making it suitable for diverse planning scenarios.

[0065] This application discloses a schematic diagram of a system prompt word, please refer to... Figure 3 The system prompts consist of a combination of role positioning, data structure, task instructions, and custom functions.

[0066] The role is defined as "You are a resident travel survey data analysis expert, specializing in helping to analyze resident travel survey data and writing Python code to process travel survey datasets. You can provide plain text explanations or code solutions to answer questions."

[0067] The data structure records that "travel survey information is divided into multiple data tables, such as households (family information), persons (personal information), travels (travel records), and activities (activity records), and these tables are stored in a MongoDB database."

[0068] The field types and explanations are as follows: households:{ hos id:integer, #family number dist:string, #administrative region members: integer, #total number of family members ...... Family expansion coefficient: float#family expansion coefficient }".

[0069] The task instructions were recorded as "Use Python, along with the libraries numpy (np), pandas (pd), shapely, geopandas (gpd), geopy, networkx, keplergl, and thefuzz". Avoid writing code that involves saving, reading, or writing to disk. The results are stored in the `result` dictionary, containing the keys `answer` and `additonal_info`. ...... "Address potential errors and missing data in travel survey data."

[0070] The custom function is recorded as "Description: To statistically analyze and generate 0D data across different analysis units and then visualize it; parameter: travel_cata: Travel record form anajysis_unit: The analysis unit data that needs to be aggregated, such as streets, districts, traffic areas, etc.; Output: Returns a 0D connection containing the total number of trips in GeoDataFrame format. Example: The source uses the predefined function Great_OD(tavel_data analysis_unit), and the output data includes... [O, D, OName, DName, trips, geometry]".

[0071] Based on the same inventive concept, embodiments of this application provide a system prompt word generation system for urban and rural planning scenarios, including: Module 401 is used to acquire query operations and input text; The memory 402 is used to store the program for the system prompt word generation method in the planning scenario; The processor 403 can load and execute the program in the memory to implement the system prompt word generation method in the planned scenario.

[0072] By employing the aforementioned technical solution, system prompts are generated by extracting planning scenarios and obtaining corresponding role positioning, task instructions, data specifications, and custom functions. This method ensures that the prompts closely align with specific planning needs, improving the relevance and accuracy of the generated prompts. Simultaneously, it achieves automated generation, significantly reducing manual intervention and improving efficiency and reliability, making it suitable for diverse planning scenarios.

[0073] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0074] This application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as a method for generating system prompts in a planning scenario.

[0075] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.

[0076] Based on the same inventive concept, embodiments of this application provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as a method for generating system prompts in a planned scenario.

[0077] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0078] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for generating system prompts in an urban and rural planning scenario, characterized in that, The method includes: In response to a query operation, planning scenarios are extracted from the input text, including urban and rural planning, transportation planning, and municipal planning. Based on the planned scenario, obtain the role positioning and task instructions; Determine the planning database corresponding to the planning scenario; The planning database is labeled with types and semantic annotations to obtain data specifications; Based on the input text and the planning scenario, an analysis task is generated; Set a custom function for the analysis task; By combining the aforementioned role positioning, task instructions, data specifications, and custom functions, system prompts are obtained.

2. The method for generating system prompts in urban and rural planning scenarios according to claim 1, characterized in that, The method further includes: Based on the system prompts, code is generated to analyze tasks, process data, invoke tools, and visualize data. Executable code is generated based on the code analysis task, the processed data, the tool calls, and the visualization. The executable code is combined to form an executable script; The executable script is executed to obtain the output report corresponding to the input text.

3. The method for generating system prompts in urban and rural planning scenarios according to claim 2, characterized in that, The step of obtaining role positioning and task instructions based on the planned scenario includes: Based on the planned scenario, set the professional role identity in the interaction process and generate the role positioning; According to the planning scenario, construct an instruction type outline, which includes tool priority, code style, data verification, and analysis results; According to the instruction type outline, the task instruction is obtained by searching in the preset instruction library.

4. The method for generating system prompts in urban and rural planning scenarios according to claim 3, characterized in that, The step of constructing an instruction type outline according to the planned scenario includes: According to the planned scenario, operation instructions are extracted from the preset instruction library to obtain the instruction type outline.

5. The method for generating system prompts in urban and rural planning scenarios according to claim 1, characterized in that, The process of performing type labeling and semantic annotation on the planning database to obtain data specifications includes: Extract data information from the planning database, including field names, data types, and business descriptions; Based on the planning scenario, the business description is standardized to obtain the business meaning; Establish a mapping relationship between the field names, the data types, and the business meanings to form the data specifications.

6. The method for generating system prompts in urban and rural planning scenarios according to claim 1, characterized in that, Setting a custom function for the analysis task includes: Based on the complexity of the analysis task, the analysis task is divided into complex analysis tasks and simple analysis tasks; Based on the complex analysis task, extract complex analysis functions from a custom analysis tool library; Based on the simple analysis task, extract simple analysis functions from the standard tool library; The complex analysis function and the simple analysis function are combined to obtain the custom function.

7. The method for generating system prompts in urban and rural planning scenarios according to claim 6, characterized in that, The complex analysis functions include: dynamic scale OD calculation function, proximity analysis function for different spatial elements, and generation of passenger flow corridors.

8. A system for generating system prompts in urban and rural planning scenarios, characterized in that, The system is used to execute the system prompt word generation method in the urban and rural planning scenario as described in any one of claims 1 to 7, the system comprising: The acquisition module is used to acquire query operations and input text; A memory is used to store the program for the system prompt word generation method in the urban and rural planning scenario; The processor and the program in the memory can be loaded and executed by the processor to implement the system prompt word generation method in the urban and rural planning scenario.

9. A smart terminal, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and execute the method as described in any one of claims 1 to 7.

Images