Large model-based graduation thesis format and grammar detection system and method

Abstract

The invention discloses a large-model-based graduation thesis format and grammar detection system and method, and the system comprises a user interaction module which is used for user login, registration, graduation thesis file uploading and task management; the analyzing and positioning module is used for analyzing and extracting the content of the graduation thesis and positioning the position of a key part; the format detection module is used for verifying the format parameters of each key part of the graduation thesis and generating a format question list; the grammar detection module is used for integrating a large language model and generating a grammar modification suggestion list; and the result output module is used for integrating the format problem or the grammar modification suggestion list and generating a document with an annotation. According to the method, the graduation thesis is submitted by the user, then the detection type is selected, the corresponding detection program can be automatically operated, the Word document with the annotation can be generated after operation is completed, the user can download the Word document, the format detection result or the grammar modification suggestion is obtained, and the thesis modification workload of students and teachers is greatly reduced.

Description

Technical Field

[0001] This invention relates to the field of computer technology, specifically to a graduation thesis format and grammar detection system and method based on a large model. Background Technology

[0002] Graduation theses are a concentrated reflection of students' research capabilities and academic literacy in higher education. Their formatting and linguistic accuracy directly affect their academic quality and review results. Currently, most universities still use traditional manual methods for formatting and grammar checking of graduation theses. However, universities have strict academic regulations regarding the format and grammatical expression of graduation theses. Due to the numerous formatting details, varying standards among universities, and complex types of grammatical errors, traditional manual checking is difficult to complete efficiently, comprehensively, and accurately. Supervisors have a heavy workload checking the formatting and grammatical issues of multiple students' theses, and manual checking is slow and struggles to quickly and accurately locate all problems. This invention integrates document structure parsing technology with large-scale model semantic understanding capabilities to achieve comprehensive, intelligent, and high-precision detection from formatting to grammar, helping to improve the quality of graduation theses and review efficiency.

[0003] Office OpenXML, also known as OpenXML or OOXML, is an XML-based document format standard. It was originally a product development technical specification for Microsoft Office 2007. In 2006, it became an ECMA international standard (ECMA-376), and in 2008, it passed the fast track procedure of the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) to become the international standard ISO / IEC 29500.

[0004] python-docx is a Python scientific toolkit for creating and updating Word documents. It belongs to the python-openxml project and provides a Pythonic API that allows developers to manipulate Word documents using Python code without relying on Microsoft or Windows operating systems.

[0005] The OpenAI SDK is an open-source Python software development kit (SDK) that helps developers quickly access various large-scale models from both domestic and international sources to build intelligent applications.

[0006] Existing technologies include some methods for detecting the format of graduation theses, but their flexibility is limited, making it difficult to adapt to the differentiated requirements of different schools, colleges, and even majors. Several shortcomings exist: 1) Some graduation thesis format detection systems rely on predefined templates or rule bases, such as dynamically matching chapter titles using regular expressions. This method only matches correctly if the chapter titles are named "Chapter *** ***". 2) Existing technologies mainly focus on format detection, while the quality of graduation theses also depends on the accuracy and professionalism of the language. Graduation thesis format detection systems lack the ability to detect grammatical errors, inappropriate academic terminology, and redundant expressions. Summary of the Invention

[0007] The purpose of this invention is to provide a graduation thesis format and grammar detection system and method based on a large model, to solve the problems of the above-mentioned prior art, and to solve the problem that there is no existing technology that can effectively and automatically detect the grammar of the thesis.

[0008] The present invention solves the above problems through the following technical solution:

[0009] A graduation thesis format and grammar detection system based on a large model, comprising:

[0010] The user interaction module is used for user login, registration, uploading graduation thesis files, and task management;

[0011] The parsing and positioning module is used to parse and extract the content of the graduation thesis and locate the key parts.

[0012] The format detection module is used to verify the format parameters of key parts of the graduation thesis and generate a list of format issues.

[0013] The syntax detection module is used to integrate a large language model and generate a list of syntax modification suggestions;

[0014] The results output module is used to integrate a list of formatting issues or grammar modification suggestions and generate an annotated document.

[0015] Furthermore, the user interaction module includes receiving graduation thesis files uploaded by users, selecting the detection type as "format detection" or "grammar detection", displaying the detection results and annotated documents for users to download, and supporting user login, registration, file management and other operations.

[0016] Furthermore, the analysis and location of the key parts of the graduation thesis specifically includes:

[0017] First, the Python-docx library is used to parse the Word document into a Document object doc. Then, doc.paragraphs is used to traverse all paragraphs in the thesis document. Each paragraph is a Paragraph object para. Then, the paragraphs are split according to the key parts contained in the standard thesis reference format.

[0018] The first page is the cover page. The table of contents, Chinese abstract, English abstract, conclusion, acknowledgments, references, and appendix are fixed headings. The heading elements are located by matching corresponding keywords, and the content of the corresponding section follows the heading element.

[0019] In the main body, the heading styles of each chapter are extracted, that is, para.style.name matches the chapter title and locates the chapter title element. The content of each chapter is then placed after the title element.

[0020] Furthermore, the main text extracts the heading style to match the chapter headings and locates the chapter heading elements. When para.style.name == 'Heading 1', it is considered a first-level heading, and the content of each chapter follows the heading element; similarly, when para.style.name == 'Heading 2', it is considered a second-level heading.

[0021] Furthermore, the format detection module includes, based on the results of the parsing and positioning module, using the API of the python-docx library to obtain paragraph format and text attribute related information, then performing format verification according to the preset paper reference format specifications, recording the detected format error information, and finally generating a list of format problems.

[0022] Furthermore, the grammar detection module includes integrating the OpenAI SDK, accessing a large language model such as Deepseek, iFlytek Spark, and Moonshot by calling OpenAI's API, then constructing system prompts containing task instructions, clarifying the model's role and task as a grammar checking assistant, and inputting the text to be checked into the model to realize grammar and typo detection of the paper text, and finally generating corresponding modification suggestions.

[0023] Furthermore, the result output module is used to insert annotations indicating formatting issues or grammatical correction suggestions at corresponding positions in the original graduation thesis document, and simultaneously generate a Word result document with annotation information for users to download.

[0024] A method for detecting the format and grammar of graduation theses based on a large model includes the following steps:

[0025] Step S1: Enter the login page, enter your username and password, and try to log in to the system;

[0026] Step S2: Determine if the login operation was successful. If the login fails, return to the login page and re-enter the username and password; if the login is successful, enter the system main page.

[0027] Step S3: Upload the graduation thesis file to be checked on the main page and select the thesis detection type;

[0028] Step S4: The parsing and positioning module parses and locates the key parts of the graduation thesis, including the cover, table of contents, Chinese abstract, English abstract, main text, conclusion, acknowledgments, references, and appendices;

[0029] Step S5: If the detection type is format detection, proceed to step S6; if the detection type is syntax detection, proceed to step S7.

[0030] Step S6: The format detection module performs format checks on each key part of the graduation thesis parsed by the parsing and positioning module according to the preset thesis reference format specifications, and finally generates a list of format problems based on the detection results.

[0031] Step S7: The grammar detection module calls the large language model and, by executing system prompts containing task instructions, clarifies the model's role and task as a grammar checking assistant. It performs grammar checks on the text content of the Chinese abstract, English abstract, main text, conclusion, and acknowledgments parsed by the parsing and positioning module, and finally generates corresponding grammar modification suggestions based on the detection results.

[0032] Step S8: The result output module integrates the detection results and generates a Word result document with annotation information.

[0033] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0034] (1) This invention achieves full automation of the entire process from uploading, parsing, detection to generating annotations through modular design. Users only need to upload their graduation thesis and select the detection type, and the system can automatically run and generate annotated Word documents for download, which greatly reduces the manual checking burden of teachers and students and improves the efficiency and accuracy of thesis revision.

[0035] (2) This invention locates key parts of a paper by parsing the heading style and keyword matching of Word documents, supports non-standard heading structures, and improves the system's compatibility and recognition accuracy for different paper formats to a certain extent. By integrating a large language model for grammar detection, it can identify problems such as typos, grammatical errors, redundant expressions, and inappropriate academic terms. It can not only point out grammatical errors, but also explain why they are wrong and provide modification suggestions and polishing schemes, thereby further improving the language quality on the basis of format standardization and filling the functional gaps of existing technologies.

[0036] (3) This invention uses both format and grammar checks to help students systematically improve the standardization and language quality of their papers. Supervisors can focus more on the quality of the paper content and devote their energy to the core parts of the paper, such as the innovation of the research question, the rationality of the experimental design, the depth of data analysis, the rigor of the argumentation logic, and the reliability of the conclusions, without having to focus on the tedious review work of correcting format and grammatical errors. Attached Figure Description

[0037] Figure 1 This is a system architecture diagram of the present invention;

[0038] Figure 2 This is a flowchart of the process of the present invention;

[0039] Figure 3 A flowchart for locating key parts of the paper's structure; Detailed Implementation

[0040] The present invention will be further described in detail below with reference to the embodiments. The described embodiments are only some embodiments of the present invention, and the implementation of the present invention is not limited thereto.

[0041] Example 1:

[0042] like Figure 1 The diagram shown is an architecture diagram of a graduation thesis format and grammar detection system based on a large model proposed in this invention. The graduation thesis format and grammar detection system consists of the following five parts:

[0043] The user interaction module is used for user login, registration, uploading graduation thesis files, and task management;

[0044] The parsing and positioning module is used to parse and extract the content of the graduation thesis and locate the key parts.

[0045] The format detection module is used to verify the format parameters of key parts of the graduation thesis and generate a list of format issues.

[0046] The syntax detection module is used to integrate a large language model and generate a list of syntax modification suggestions;

[0047] The results output module is used to integrate a list of formatting issues or grammar modification suggestions and generate an annotated document.

[0048] Furthermore, the user interaction module includes receiving graduation thesis files uploaded by users, selecting the detection type as "format detection" or "grammar detection", displaying the detection results and annotated documents for users to download, and supporting user login, registration, file management and other operations.

[0049] Furthermore, combined Figure 3 As shown, the method for analyzing and locating the key parts of the paper includes the following steps:

[0050] Step (1): Load the uploaded graduation thesis Word document and use the python-docx library to parse the Word document into a Document object doc;

[0051] Step (2): Initialize the chapter structure keywords, including: table of contents, Chinese abstract, English abstract, conclusion, acknowledgments, references, and appendix. Then, use doc.paragraphs to traverse all paragraphs in the document. Each paragraph is a Paragraph object para.

[0052] Step (3): Divide the paragraphs according to the key parts included in the standard graduation thesis reference format. The first page is the cover page. The table of contents, Chinese abstract, English abstract, conclusion, acknowledgments, references, and appendix are fixed titles. Locate the title elements by matching the corresponding keywords. The content of the corresponding part is after the title element.

[0053] Step (4): Locate the content of each chapter in the main text. By extracting the heading style, i.e., para.style.name, match each chapter title to locate the chapter title element. After the title element is the content of each chapter.

[0054] Step (5): Return the location result.

[0055] Furthermore, in step (4), the main text is located by extracting the heading style and matching the headings of each chapter to locate the heading elements of each chapter. When para.style.name == 'Heading 1', it is regarded as a first-level heading, and the content of each chapter follows the heading element; similarly, when para.style.name == 'Heading 2', it is regarded as a second-level heading.

[0056] Furthermore, the format detection module includes, based on the results of the parsing and positioning module, using the API of the python-docx library to obtain paragraph format and text attribute related information, then performing format verification according to the preset paper reference format specifications, recording the detected format error information, and finally generating a list of format problems.

[0057] Furthermore, the grammar detection module includes integrating the OpenAI SDK, accessing a large language model such as Deepseek, iFlytek Spark, and Moonshot by calling OpenAI's API, then constructing system prompts containing task instructions, clarifying the model's role and task as a grammar checking assistant, and inputting the text to be checked into the model to realize grammar and typo detection of the paper text, and finally generating corresponding modification suggestions.

[0058] Furthermore, the result output module is used to insert annotations indicating formatting issues or grammatical correction suggestions at corresponding positions in the original graduation thesis document, and simultaneously generate a Word result document with annotation information for users to download.

[0059] Example 2:

[0060] like Figure 2 The diagram shows the workflow of this invention, a method for detecting the format and grammar of a graduation thesis based on a large model, comprising the following steps:

[0061] Step S1: Enter the login page, enter your username and password, and try to log in to the system;

[0062] Step S2: Determine if the login operation was successful. If the login fails, return to the login page and re-enter the username and password; if the login is successful, enter the system main page.

[0063] Step S3: Upload the graduation thesis file to be checked on the main page and select the thesis detection type;

[0064] Step S4: The parsing and positioning module parses and locates the key parts of the graduation thesis, including the cover, table of contents, Chinese abstract, English abstract, main text, conclusion, acknowledgments, references, and appendices;

[0065] Step S5: If the detection type is format detection, proceed to step S6; if the detection type is syntax detection, proceed to step S7.

[0066] Step S6: The format detection module performs format checks on each key part of the graduation thesis parsed by the parsing and positioning module according to the preset thesis reference format specifications, and finally generates a list of format problems based on the detection results.

[0067] Step S7: The grammar detection module calls the large language model and, by executing system prompts containing task instructions, clarifies the model's role and task as a grammar checking assistant. It performs grammar checks on the text content of the Chinese abstract, English abstract, main text, conclusion, and acknowledgments parsed by the parsing and positioning module, and finally generates corresponding grammar modification suggestions based on the detection results.

[0068] Step S8: The result output module integrates the detection results and generates a Word result document with annotation information.

[0069] One specific method for implementing syntax detection using a large language model in step S7 is, for example, deepseek:

[0070] First, configure the access address (BASE_URL) and key (API_KEY) of the model service. Create a communication channel between the client and the model using OpenAI (api_key = API_KEY, base_url = BASE_URL). Then, construct a system prompt that requires the model to act as a "grammar checker" to find typos and grammatical errors. After encapsulating the system prompt and the text to be checked into messages, call client.chat.completions.create(model = "deepseek-chat", messages = messages, stream = False) to send a request to the large language model. The model performs inference based on the system prompt and the input text and generates the check results. Finally, the grammar check can be achieved by extracting the model output.

Claims

1. A large model-based graduation thesis format and grammar detection system, characterized by, The application relates to a thesis review system and method. The user interaction module is used for user login, registration, uploading of graduation thesis files and task management. The parsing and positioning module is used for parsing and extracting the content of the graduation thesis and positioning the positions of key parts. The format detection module is used for checking the format parameters of the key parts of the graduation thesis and generating a format problem list. The syntax detection module is used for integrating a large language model and generating a syntax modification suggestion list. The result output module is used for integrating the format problem list or the syntax modification suggestion list and generating an annotated document.

2. The dissertation format and grammar detection system based on a large model according to claim 1, wherein, The user interaction module receives the user-uploaded graduation thesis files, selects the detection type as "format detection" or "syntax detection", displays the detection results and annotated documents for the user to download, and supports user login, registration, file management and other operations.

3. The dissertation format and grammar detection system based on a large model according to claim 1, wherein, The content of the graduation thesis includes the following key parts: cover, table of contents, Chinese abstract, English abstract, main text, conclusion, thanks, references and appendix.

4. The dissertation format and grammar detection system based on a large model according to claim 3, wherein, The parsing and positioning of the key parts of the graduation thesis specifically include the following steps: First, the python-docx library is used to parse the Word document into a Document object doc, all paragraphs in the graduation thesis document are traversed through doc.paragraphs, each paragraph is a Paragraph object para, and then the paragraphs are segmented according to the key parts contained in the standard graduation thesis reference format; The first page is the cover page, the table of contents, the Chinese abstract, the English abstract, the conclusion, the thanks, the references and the appendix are fixed titles, the title elements are positioned through the matching of corresponding keywords, and the content of the corresponding part is positioned after the title elements; The chapters in the main text are positioned through the extraction of the title Heading style, that is, para.style.name matches the chapter title, and the content of each chapter is positioned after the title element.

5. The dissertation format and grammar detection system based on a large model according to claim 4, wherein, The main text is positioned through the extraction of the title Heading style, that is, para.style.name matches the chapter title, and the content of each chapter is positioned after the title element.

6. The dissertation format and grammar detection system based on a large model according to claim 1, wherein, The format detection module includes, according to the results of the parsing and positioning module, using the API of the python-docx library to obtain paragraph format and text attribute related information, then performing format checking according to the preset thesis reference format specification, recording the detected format error information, and finally generating a format problem list.

7. The dissertation format and grammar detection system based on a large model according to claim 1, wherein, The syntax detection module includes, integrating the OpenAI SDK, calling the API of OpenAI to access a large language model such as but not limited to deepseek, Xunfei Spark and Moonshot, then constructing a system prompt containing a task instruction, clearly defining the role and task of the model as a grammar checking assistant, inputting the to-be-detected text into the model as input to realize syntax and error detection of the thesis text, and finally generating corresponding modification suggestions.

8. The dissertation format and grammar detection system based on a large model according to claim 1, wherein, The result output module is configured to insert the comment mark of the format problem or the grammar modification suggestion into the corresponding position of the original graduation thesis document, and generate a Word result document with the comment information for the user to download. 9.A method for detecting graduation thesis format and grammar based on a large model, characterized in that, The method comprises the following steps: Step S1: enter the login page, input the user account and password, and try to log in the system; Step S2: determine whether the login operation is successful, if the login fails, return to the login page to re-input the account and password; If the login is successful, enter the main page of the system; Step S3: upload the graduation thesis file to be detected on the main page, and select the thesis detection type; Step S4: the parsing and positioning module parses and positions each key part of the graduation thesis, including the cover, table of contents, Chinese abstract, English abstract, main text, conclusion, thanks, reference, appendix; Step S5: if the detection type is format detection, proceed to step S6; if the detection type is grammar detection, jump to step S7; Step S6: the format detection module performs format detection on each key part of the graduation thesis parsed by the parsing and positioning module according to the preset thesis reference format specification, and finally generates a format problem list according to the detection result; Step S7: the grammar detection module calls a large language model, clearly defines the role and task of the model as a grammar checking assistant by executing a system prompt containing a task instruction, performs grammar detection on the text content of the Chinese abstract, English abstract, main text, conclusion, and thanks part parsed by the parsing and positioning module, and finally generates corresponding grammar modification suggestions according to the detection result; Step S8: the result output module integrates the detection results to generate a Word result document with comment information.

Images