Artificial intelligence-based auxiliary creation method and system for structured national natural fund

Abstract

The invention discloses an artificial intelligence-based auxiliary creation method and system for a structured national natural fund, and belongs to the field of artificial intelligence. The method comprises the steps of receiving core elements of an NSFC application form; decomposing the application form into a plurality of logic components; accessing a predefined template library containing specific prompt strategy templates for each logic component of the NSFC; selecting a corresponding template to generate a structured guiding prompt; inputting the prompt into an artificial intelligence language model to obtain a preliminary draft; and assembling the draft of each part to form a complete application form. Optionally, interactive iterative optimization and cross-part consistency check are also included. According to the method, the structured prompt strategy template and the systematic process which are designed according to the specific requirements of the NSFC are adopted, so that the creation efficiency and quality of the high-quality NSFC application form which conforms to the specification are remarkably improved, and the burden of a user is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence, and in particular relates to an artificial intelligence-based structured National Natural Science Foundation auxiliary creation method and system. Background Art

[0002] The National Natural Science Foundation of China (NSFC) is one of the main channels for supporting basic research and frontier scientific exploration in my country. The quality of its application writing is directly related to whether the project can obtain funding. NSFC applications not only require the content to be highly scientific and innovative, but also need to follow specific format specifications and clearly and logically explain the research background, scientific problems, research objectives, research content, technical routes, feasibility, innovations, and research foundations. This places extremely high demands on the applicant's comprehensive scientific research capabilities and writing skills. The traditional fund application writing process is usually time-consuming and laborious, requiring applicants to devote a lot of energy to literature research, refining ideas, organizing content, and polishing the text.

[0003] Existing general-purpose AI writing tools still face numerous challenges and limitations when directly applied to the highly structured, logical, and professionally demanding NSFC proposals. First, general-purpose AI models lack a deep understanding of the inherent logic, review key points, and specific writing paradigms of NSFC proposals. They may not accurately grasp the core requirement of being "scientifically problem-oriented" and struggle to generate content that meets the implicit expectations of the NSFC and the review focus. Second, general-purpose AI is insufficient in automatically adhering to the specific structural and content requirements of each section of the NSFC application (such as the argumentation logic of the project establishment basis, the progressive level of research content, the precise distillation of characteristics and innovations, and the targeted demonstration of feasibility analysis). The generated text is often loosely structured, poorly formatted, and lacks organic connections between sections. Third, the content generated by general-purpose AI often fails to meet the high standards of the NSFC in terms of scientific rigor, logical depth, and innovative expression, and may suffer from problems such as insufficient argumentation, vague innovation points, and superficial feasibility demonstrations. Furthermore, to obtain text that meets the requirements, users must repeatedly adjust prompts, a cumbersome and inefficient process. The resulting fragments may still require extensive revisions, which does not significantly reduce the burden on applicants. Finally, general AI struggles to adapt to the nuanced requirements of different NSFC scientific departments or project types. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides an artificial intelligence-based structured National Natural Science Foundation assisted creation method, comprising:

[0005] Receive user input on the core elements of the NSFC application to be written;

[0006] Break down the proposed application into multiple logical components according to the preset NSFC application standard structure rules;

[0007] Accessing a predefined template library storing a plurality of specific prompting strategy templates for different logical components of the NSFC application;

[0008] For at least one of the logical components, selecting a specific prompt policy template corresponding to the logical component from the template library;

[0009] Generate structured guiding prompts based on the core elements and the selected specific prompt strategy template;

[0010] Inputting the structured guiding prompts into a predetermined artificial intelligence language model;

[0011] Obtaining a preliminary draft of the logical component from the artificial intelligence language model;

[0012] Assemble said initial draft or subsequently revised drafts of each of said logical components to form a complete NSFC application.

[0013] Preferably, in the step of generating structured guiding prompts, the structured guiding prompts include at least two of role setting instructions, context information, detailed content requirement instructions, logical structure instructions, style and tone instructions, keyword injection instructions, and negative constraint instructions.

[0014] Preferably, the specific prompt strategy templates contained in the template library cover at least five logical components of the NSFC application: title, abstract, project establishment basis, research content, research objectives, key scientific problems to be solved, research plan, feasibility analysis, characteristics and innovations of the project, research basis, and working conditions.

[0015] Preferably, after the step of obtaining the preliminary draft of the logical component from the artificial intelligence language model, the following interactive iterative optimization step is also included:

[0016] providing a human-computer interaction interface requiring deep user participation to present the preliminary draft to the user;

[0017] Receiving feedback information submitted by the user through the human-computer interaction interface;

[0018] generating revision prompts based on the feedback information and the guiding prompts;

[0019] The revision prompt is input into the artificial intelligence language model to obtain a revised draft of the logical component.

[0020] Preferably, after generating drafts of multiple logical components, a step of performing content consistency check across logical components is also included, wherein the check includes at least one or more of the consistency between core scientific issues and research objectives, and the matching between research content and feasibility of research plan.

[0021] Preferably, the method further includes the step of using a preset language polishing template to perform language polishing or format adjustment on the complete NSFC application.

[0022] On the other hand, the present invention also provides a structured National Natural Science Foundation auxiliary creation system based on artificial intelligence, comprising:

[0023] A receiving unit, configured to receive the core elements of user input;

[0024] Decomposition unit, used to break down the application into its logical components;

[0025] Template library, used to store specific prompt strategy templates for different logical components of NSFC applications;

[0026] a prompt generation engine configured to select a corresponding template from the template library and generate a structured guiding prompt based on the core elements and the template;

[0027] An AI model interface unit, configured to send guiding prompts and revision prompts to the AI language model and receive preliminary drafts and revised drafts;

[0028] The assembly unit is used to assemble the drafts of each logical part to form a complete application.

[0029] Preferably, the system further comprises at least one of the following units:

[0030] An interactive processing unit, configured to provide a human-computer interaction interface for deep user engagement, receive user feedback, and generate revision prompts based on the feedback and guiding prompts;

[0031] a consistency check unit for performing content consistency checks across logical components;

[0032] The Proofreading Unit is used to polish the language or adjust the format of the completed NSFC application.

[0033] On the other hand, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the method is implemented when the processor executes the computer program.

[0034] On the other hand, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the method when executed by a processor.

[0035] Compared with the prior art, the present invention has the following advantages and technical effects:

[0036] Significantly improve the quality and standardization of NSFC application writing: the generated content is more in line with the internal logic, structural requirements and review focus of NSFC, especially in reflecting scientific problem orientation, highlighting innovations, and demonstrating feasibility.

[0037] Significantly improve writing efficiency: By precisely guiding AI to generate high-quality first drafts, users can reduce the time spent on repeated revisions and ideation. According to preliminary evaluations, efficiency can be improved by approximately 50% and the number of draft revision rounds can be reduced by approximately 30%.

[0038] Lower the threshold and burden for users: Users do not need to have advanced prompt engineering skills, and can effectively utilize AI capabilities with the help of templates, which is especially friendly to applicants with little experience.

[0039] Enhance the consistency and logic of the content: By using structured and standardized templates to guide each section, it helps to ensure the consistency of style and logical coherence between different parts of the application.

[0040] Improve targeting: The design of the template library can take into account the characteristics of different departments or project types, making AI assistance more targeted. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0042] Figure 1 Schematic diagram of the process of the artificial intelligence-based structured National Natural Science Foundation application auxiliary creation method according to an embodiment of the present invention;

[0043] Figure 2 Schematic diagram of a structured guiding prompt generation process according to an embodiment of the present invention;

[0044] Figure 3 A schematic diagram of an interactive iterative optimization process according to an embodiment of the present invention;

[0045] Figure 4 This is a functional module block diagram of an artificial intelligence-based structured National Natural Science Foundation application assistance writing system according to an embodiment of the present invention;

[0046] Figure 5Schematic diagram of the working principle of cross-part consistency check according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0049] Example 1

[0050] like Figure 1-2 As shown, this embodiment provides a structured National Natural Science Foundation assisted creation method and system based on artificial intelligence, including:

[0051] a) Receive the core elements related to the NSFC application to be written from the user;

[0052] b) breaking down the proposed application into multiple logical components according to the pre-set NSFC application standard structure rules;

[0053] c) accessing a predefined template library storing a plurality of specific prompting strategy templates for different logical components of the NSFC application;

[0054] d) for at least one of the logical components, selecting a specific prompting strategy template corresponding to the logical component from the template library; the specific prompting strategy template is pre-designed based on the writing requirements, review focus, and expert experience of the NSFC logical component, and includes a structured instruction field;

[0055] e) generating a structured guiding prompt based on the core elements received in step a) and the specific prompt strategy template selected in step d);

[0056] f) inputting the structured guiding prompts into a predetermined artificial intelligence language model;

[0057] g) obtaining a preliminary draft of the logical component from the artificial intelligence language model; and

[0058] h) assemble said initial draft or subsequently revised drafts of said logical components to form a complete NSFC application.

[0059] Step S1: receiving core elements;

[0060] The system receives user input through a user interface (e.g., a form or text box on a graphical user interface (GUI)) for core elements related to the NSFC application to be written. These core elements form the basis for subsequent content generation and may include at least: research areas, core scientific questions, scientific hypotheses, key scientific problems to be solved (e.g., broken down into question one and question two), the project's "unique" innovation strategy, the main research difficulties and coping strategies, the project's scientific attributes (e.g., free exploration or goal-oriented), an outline of the main research content, the expected research objectives, a list of key references or an abstract, etc. User input can be done by filling out a structured form or by entering free text followed by information extraction by the system.

[0061] Step S2: structured decomposition;

[0062] The system automatically breaks down the proposed proposal into several logically ordered components based on the pre-set NSFC standard proposal structure (e.g., including but not limited to: title, abstract, project rationale, research content, research objectives, key scientific questions to be addressed, research plan, feasibility analysis, project characteristics and innovations, research foundation, and working conditions). The system internally stores a list of these standard components, along with their general writing requirements and logical relationships.

[0063] Step S3: access the template library and select a template;

[0064] The system contains a predefined specific prompt strategy template library. The template library is stored in the local device and contains specific prompt strategy templates for each (or key) logical component of the NSFC application. For example, the library contains templates specifically for generating "abstracts", templates for generating "basis for project establishment", templates for generating "research content", and so on. These templates are pre-designed and constructed based on the official guidelines of NSFC, the experience of writing experts, and the analysis of a large number of successful application cases. When processing a certain logical component (such as "abstract"), the system will automatically retrieve and select the corresponding specific prompt strategy template from the template library.

[0065] Step S4: generating structured guiding prompts;

[0066] Reference Figure 2 The prompt generation engine automatically generates a structured guided prompt (GuidedPrompt) based on the specific prompt strategy template selected in step S3 and the user core elements received in step S1.

[0067] Template Application: The template itself defines the structure and instructional framework for generating prompts. For example, a template might specify that generated prompts must include "role-playing," "context," "specific content requirements," "logical structure instructions," "style and tone," and "core elements to be incorporated."

[0068] Core element injection: The system intelligently embeds the core elements input by the user (such as core scientific issues, innovation strategies, key result expectations, etc.) into the position specified by the template to form specific prompt content.

[0069] Structured Instructions: The generated guidance prompts are not just simple instructions, but contain a series of structured instructions to guide the LLM in a comprehensive and detailed manner. These instruction types** include at least two or more of the following: role-setting instructions (for example, requiring the LLM to act as an NSFC reviewer or a senior researcher in a certain field), contextual information (for example, informing the LLM that they are writing a specific part of the NSFC application), detailed content requirements (for example, explicitly requiring the generation of specific arguments, the inclusion of specific data points, and the adherence to a specific argument logic), logical structure instructions (for example, requiring the application to be generated in a specific substructure or sequence), style and tone instructions (for example, requiring objectivity, rigor, and persuasiveness), keyword injection instructions, and possible negative constraint instructions (for example, avoiding certain common errors or redundant expressions).

[0070] Step S5: AI model interaction and preliminary draft acquisition;

[0071] The system sends the structured guiding prompts generated in step S4 to a predetermined artificial intelligence language model (such as, but not limited to, GPT series models, Gemini series models, Claude series models, or other models that have excellent performance in generating Chinese scientific research texts) through the AI model interface unit via an API or other communication method. The model is instructed to generate text according to the requirements of the received structured guiding prompts, and then receives a preliminary text draft of the corresponding logical component returned by the model.

[0072] Step S6: assembling;

[0073] When the preliminary drafts of each (or multiple selected by the user) logical component (or the final draft after optimization in subsequent step S7) are generated, the assembly unit assembles these drafts according to the standard order of the NSFC application determined in step S2 to form a relatively complete draft of the NSFC application.

[0074] Step S7: interactive iterative optimization;

[0075] Reference Figure 3,In order to further improve the quality of the draft and incorporate ,the user’s specific modification intentions, the system can provide ,interactive iterative optimization function.

[0076] Presentation and feedback: The interactive processing unit presents the preliminary draft obtained in step S5 to the user through a human-computer interaction interface (such as a text editor combined with a sidebar tool).

[0077] Deep user engagement: Users review the draft and can submit feedback on specific sentences, paragraphs, or entire sections. This feedback requires deep user engagement, allowing users to highlight text, select preset editing actions (such as "rewrite," "expand," "simplify," "needs more support," "add reference suggestions," etc.), and enter specific revisions, additional information, or new requirements.

[0078] Revision Prompt Generation: The interactive processing unit receives and analyzes the user's feedback information. Combining the feedback with the original structured guiding prompts for generating the preliminary draft (from step S4), the system automatically generates one or more Revision Prompts.

[0079] Draft revision: The AI model interface unit sends the revision prompt to the artificial intelligence language model again to obtain the draft revised based on user feedback.

[0080] Multiple rounds of iteration: This interactive optimization process (presentation -> feedback -> revision prompts -> revised draft) can be repeated multiple times until the user is satisfied with the draft of the part.

[0081] Step S8: cross-part consistency check;

[0082] Reference Figure 5 ,After the drafts of multiple logical components (e.g., project basis, research objectives, research content, research plan, feasibility analysis, etc.) are generated to a certain extent or are all preliminarily completed, the consistency check unit can perform cross-component ,content consistency checks.

[0083] Check content: This check can be performed using a rule engine or by re-calling the AI model (via specific consistency check prompts). The check content should include at least one or more of the following: whether the core scientific question and research objectives are consistent and mutually supportive; whether the research content is feasible and compatible with the methods and techniques described in the research plan; whether the issues identified in the project establishment basis are addressed in the research content and objectives; and whether the use of key terms throughout the paper is consistent.

[0084] Result processing: If the check finds inconsistencies or logical breaks, the system can highlight the problem to the user and suggest modifications, or guide the user to return to step S7 to adjust the relevant parts.

[0085] Step S9: Language polishing and format adjustment;

[0086] After the main content of the application is basically determined, the system can provide final language polishing and format adjustment functions.

[0087] like Figure 4 As shown, on the other hand, this embodiment also provides a structured National Natural Science Foundation auxiliary creation system based on artificial intelligence, including:

[0088] A receiving unit, configured to receive the core elements of user input;

[0089] Decomposition unit, used to break down the application into its logical components;

[0090] A template library is used to store specific prompt strategy templates for different logical components of NSFC applications; the specific prompt strategy templates contained in the template library cover at least five or more logical components of NSFC applications selected from the following: title, abstract, project establishment basis, research content, research objectives, key scientific problems to be solved, research plan, feasibility analysis, characteristics and innovations of this project, research foundation, and working conditions.

[0091] A prompt generation engine is configured to select a corresponding template from the template library and generate a structured guiding prompt based on the core elements and the template. The generated structured guiding prompt includes at least two or more instruction types selected from the following: role setting instructions, context information, detailed content requirement instructions, logical structure instructions, style and tone instructions, keyword injection instructions, and negative constraint instructions.

[0092] An AI model interface unit, configured to send guiding prompts and revision prompts to the AI language model and receive preliminary drafts and revised drafts;

[0093] The assembly unit is used to assemble the drafts of each logical part to form a complete application.

[0094] Preferably, the system further comprises at least one of the following units:

[0095] An interactive processing unit, configured to provide a human-computer interaction interface for deep user engagement, receive user feedback, and generate revision prompts based on the feedback and guiding prompts;

[0096] like Figure 5As shown, the consistency check unit is used to perform content consistency checks across logical components. The check content can be performed using a rule engine or by re-calling the AI model (through specific consistency check prompts). The check content includes at least one or more of the following: whether the core scientific question and the research objective are consistent and mutually supportive; whether the research content is feasible and matches the methods and techniques described in the research plan; whether the issues identified in the project establishment basis are addressed in the research content and objectives; and whether the use of key terms throughout the text is consistent.

[0097] The Proofreading Unit is used to polish the language or adjust the format of the completed NSFC application. Using pre-set language polishing templates in the template library, specifically designed for language style optimization or grammar proofreading, the Proofreading Unit can invoke AI models to comprehensively improve the language style, fluency, grammatical errors, and spelling of the assembled NSFC application. It may also integrate formatting adjustments to assist users in adjusting to the latest NSFC formatting requirements.

[0098] On the other hand, this embodiment further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the method is implemented when the processor executes the computer program.

[0099] On the other hand, this embodiment further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the method when executed by a processor.

[0100] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A structured National Natural Science Foundation assisted creation method based on artificial intelligence, characterized by: include: Receive user input on the core elements of the NSFC application to be written; Break down the proposed application into multiple logical components according to the preset NSFC application standard structure rules; Accessing a predefined template library storing a plurality of specific prompting strategy templates for different logical components of the NSFC application; For at least one of the logical components, selecting a specific prompt policy template corresponding to the logical component from the template library; Generate structured guiding prompts based on the core elements and the selected specific prompt strategy template; Inputting the structured guiding prompts into a predetermined artificial intelligence language model; Obtaining a preliminary draft of the logical component from the artificial intelligence language model; Assemble said initial draft or subsequently revised drafts of each of said logical components to form a complete NSFC application.

2. The method according to claim 1, characterized in that In the step of generating structured guiding prompts, the structured guiding prompts include at least two of role setting instructions, context information, detailed content requirement instructions, logical structure instructions, style and tone instructions, keyword injection instructions, and negative constraint instructions.

3. The method according to claim 1, characterized in that The specific prompt strategy templates included in the template library cover at least five logical components of the NSFC application: title, abstract, project establishment basis, research content, research objectives, key scientific problems to be solved, research plan, feasibility analysis, characteristics and innovations of the project, research basis, and working conditions.

4. The method according to claim 1, wherein After the step of obtaining the preliminary draft of the logical component from the artificial intelligence language model, the following interactive iterative optimization step is also included: providing a human-computer interaction interface requiring deep user participation to present the preliminary draft to the user; Receiving feedback information submitted by the user through the human-computer interaction interface; generating revision prompts based on the feedback information and the guiding prompts; The revision prompt is input into the artificial intelligence language model to obtain a revised draft of the logical component.

5. The method according to claim 1, characterized in that After generating drafts of multiple logical components, the step of performing a content consistency check across the logical components is also included, wherein the check includes at least one or more of the consistency between the core scientific issues and the research objectives, and the matching between the research content and the feasibility of the research plan.

6. The method according to claim 1, characterized in that The method further includes the step of using a preset language polishing template to perform language polishing or format adjustment on the complete NSFC application.

7. A structured National Natural Science Foundation auxiliary creation system based on artificial intelligence, characterized by: include: A receiving unit, configured to receive the core elements of user input; Decomposition unit, used to break down the application into its logical components; Template library, used to store specific prompt strategy templates for different logical components of NSFC applications; a prompt generation engine configured to select a corresponding template from the template library and generate a structured guiding prompt based on the core elements and the template; An AI model interface unit, configured to send guiding prompts and revision prompts to the AI language model and receive preliminary drafts and revised drafts; The assembly unit is used to assemble the drafts of each logical part to form a complete application.

8. The system according to claim 7, characterized in that The system further comprises at least one of the following units: An interactive processing unit, configured to provide a human-computer interaction interface for deep user engagement, receive user feedback, and generate revision prompts based on the feedback and guiding prompts; a consistency check unit for performing content consistency checks across logical components; The Proofreading Unit is used to polish the language or adjust the format of the completed NSFC application.

9. An electronic device comprising a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images