Architectural question bank automatic generation method and system based on large model

Abstract

The invention discloses a building question bank automatic generation method and system based on a large model, and the method comprises the steps: building data analysis: analyzing characters, images, tables and formulas of building data, and carrying out the vectorization identification of the analyzed characters, images, tables and formulas through the combination of a word embedding technology, so as to form vectorized building data; analyzing the existing question bank: marking the category, keyword and knowledge point of each question in the existing building question bank through a general large model; automatic generation of a new question bank: taking questions analyzed by the existing question bank and categories, keywords, knowledge points and answers of the questions as training data of a general large model to form a question bank automatic generation large model, and taking vectorized building data as input through the question bank automatic generation large model; and automatically generating an architectural question bank containing the questions and the categories, keywords, knowledge points and answers of the questions. The building question bank can be automatically generated, and the method has the advantages of being large in knowledge point coverage range, high in efficiency and large in question setting amount.

Description

Technical Field

[0001] The present invention relates to the field of building question bank production, and in particular to a method and system for automatically generating a building question bank based on a large model. Background Art

[0002] The big model has been applied and promoted in the fields of text data retrieval and knowledge question and answer in the field of architecture, but the results obtained by the big model are one-way, and there is no standard answer to verify it in both directions. At present, the production of question banks relies heavily on experienced practitioners, who are very familiar with the knowledge of the architectural profession. And most of the questions asked by experts need to cover mainstream knowledge points and common test points. For some less popular knowledge points or knowledge points that some candidates are not familiar with, it is difficult to find answers to the targeted questions. Therefore. The questions asked by experts are limited by the barriers and restrictions of their own professional knowledge, and the coverage of knowledge points is small and the efficiency is low. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for automatically generating an architectural question bank based on a large model, so as to solve the problem that the knowledge point coverage of the architectural question bank produced by industry experts is small and the efficiency is low.

[0004] In order to solve the above technical problems, the present invention provides a method for automatically generating a building question bank based on a large model, comprising:

[0005] Building data analysis: Analyze the text, images, tables and formulas in the building data, and use word embedding technology to vectorize the parsed text, images, tables and formulas to form vectorized building data;

[0006] Existing question bank analysis: Use a general large model to mark the categories, keywords, and knowledge points of each question in the existing architectural question bank;

[0007] Automatic generation of new question bank: The questions and their categories, keywords, knowledge points and answers analyzed in the existing question bank are used as training data for the general large model to form a large model for automatic generation of question bank. The large model for automatic generation of question bank takes vectorized architectural data as input to automatically generate an architectural question bank containing questions and their categories, keywords, knowledge points and answers.

[0008] Furthermore, the method for automatically generating a large-model-based architectural question bank provided by the present invention further includes:

[0009] Automatically correct answers: A large model is automatically generated through the question bank to compare the test taker's answers to the questions with the answers in the architectural question bank for automatic correction.

[0010] Furthermore, the large-model-based automatic building question bank generation method provided by the present invention, in the step of building data parsing, uses optical character recognition technology to identify and extract text content in words, images, tables and formulas.

[0011] Furthermore, in the method for automatically generating a building question bank based on a large model provided by the present invention, in the step of automatically generating a new question bank, the categories of the questions include short-answer questions, true-or-false questions, calculation questions, definition of terms and multiple-choice questions; in the step of automatically generating a new question bank, the large model for automatically generating the question bank is used to automatically ask questions based on quantized building data, associate knowledge points and keywords to automatically generate short-answer questions, true-or-false questions, calculation questions, definition of terms and multiple-choice questions.

[0012] In order to solve the above technical problems, the present invention also provides a large-model-based automatic generation system for an architectural question bank, which adopts the above-mentioned large-model-based automatic generation method for an architectural question bank, comprising:

[0013] The building data parsing module includes a text parsing subunit, an image parsing subunit, a table parsing subunit and a formula parsing subunit, which are used for parsing text, images, tables and formulas respectively, and vectorizes the parsed text, images, tables and formulas by combining word embedding technology to form vectorized building data;

[0014] The existing question bank parsing module includes a category extraction subunit, a keyword extraction subunit and a knowledge point extraction subunit. The general large model is used to extract and mark the categories, keywords and knowledge points of each question in the existing architectural question bank through the category extraction subunit, the keyword extraction subunit and the knowledge point extraction subunit.

[0015] The new question bank automatic generation module uses the questions and their categories, keywords, knowledge points and answers of the existing question bank parsing module as training data for the general large model to form a large model for automatic question bank generation. The large model for automatic question bank generation uses the vectorized building data of the building data parsing module as input to automatically generate a building question bank containing questions and their categories, keywords, knowledge points and answers.

[0016] Furthermore, the large model-based automatic generation system for architectural question banks provided by the present invention further includes:

[0017] The automatic answer correction module automatically generates a large model through the question bank, compares the answer of the respondent to the question with the answer of the architectural question bank, and automatically corrects it.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The method and system for automatically generating a building question bank based on a large model provided by the present invention analyzes and vectorizes building data to form vectorized building data for understanding, querying and comparison by a general large model; analyzes and marks the category, keyword and knowledge point of each question by analyzing the existing question bank; uses the questions and their categories, keywords, knowledge points and answers analyzed by the existing question bank as training data for the general large model to form a large model for automatically generating a question bank, and uses the vectorized building data as input to automatically generate a building question bank containing questions and their categories, keywords, knowledge points and answers. Compared with the traditional building question bank produced by industry experts, it has the advantages of a wide coverage of knowledge points, high efficiency and a large number of questions. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of the method for automatically generating a building question bank based on a large model;

[0021] Figure 2 It is a structural composition diagram of the automatic generation system of the architectural question bank based on the large model;

[0022] Figure 3 It is a structural composition diagram of the building data analysis model;

[0023] Figure 4 It is a structural composition diagram of the existing question bank parsing module;

[0024] As shown in the figure:

[0025] 100. Automatic generation system of architectural question bank based on large model, 110. Architectural data parsing module, 111. Text parsing unit, 112. Image parsing unit, 113. Table parsing unit, 114. Formula parsing unit, 120. Existing question bank parsing module, 121. Category extraction unit, 122. Keyword extraction unit, 123. Knowledge point extraction unit, 130. New question bank automatic generation module, 140. Answer automatic correction module. DETAILED DESCRIPTION

[0026] The present invention is described in detail below in conjunction with the accompanying drawings: The advantages and features of the present invention will become more apparent from the following description. It should be noted that the accompanying drawings are in very simplified form and in non-precise proportions, and are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.

[0027] Please refer to Figure 1 The embodiment of the present invention provides a method for automatically generating a building question bank based on a large model, which may include the following steps:

[0028] Step S1, parsing of building data. Specifically: parsing the text, images, tables and formulas of the building data, combining the word embedding technology to vectorize the parsed text, images, tables and formulas to form vectorized building data. Vectorized building data can facilitate the understanding, query and comparison of general large models. The text content in text, images, tables and formulas can be identified and extracted by optical character recognition technology. The understanding of image content can be parsed according to the image large model; when parsing the table, the positional relationship between each cell needs to be retained to facilitate the subsequent general large model to establish the logical relationship between each cell; the parsing of the formula, including the formula body, the introduction of the formula and the description of the formula variables.

[0029] Step S2, analyzing the existing question bank, specifically, marking the category, keywords, and knowledge points of each question in the existing architectural question bank through a general large model.

[0030] Step S3, automatic generation of a new question bank. Specifically, the questions and their categories, keywords, knowledge points and answers parsed from the existing question bank are used as training data for the general large model to form a large model for automatic generation of question banks. The large model for automatic generation of question banks takes vectorized architectural data as input to automatically generate an architectural question bank containing questions and their categories, keywords, knowledge points and answers.

[0031] In order to generate different types of questions, in step S2, the categories of the questions include short-answer questions, true-or-false questions, calculation questions, definition of terms and multiple-choice questions; in step S3, the question bank automatically generates a large model for automatically asking questions based on vectorized building data, associating knowledge points and keywords to automatically generate short-answer questions, true-or-false questions, calculation questions, definition of terms and multiple-choice questions.

[0032] Please refer to Figure 1 In order to improve the efficiency of correcting the answers of the answerers, the method for automatically generating an architectural question bank based on a large model provided by an embodiment of the present invention may also include:

[0033] Step S4, automatic correction of answers. The large model is automatically generated by the question bank to compare the answer of the examinee to the question with the answer of the architectural question bank for automatic correction.

[0034] Please refer to Figures 2 to 4 The embodiment of the present invention provides a large-model-based automatic generation system 100 for an architectural question bank, which adopts the large-model-based automatic generation method for an architectural question bank, and includes an architectural data parsing module 110, an existing question bank parsing module 120, and a new question bank automatic generation module 130, wherein:

[0035] The building data parsing module 110 includes a text parsing subunit 111, an image parsing subunit 112, a table parsing subunit 113 and a formula parsing subunit 114, which are respectively used for parsing text, images, tables and formulas, and vectorizes the parsed text, images, tables and formulas in combination with word embedding technology to form vectorized building data.

[0036] The existing question bank parsing module 120 includes a category extraction subunit 121, a keyword extraction subunit 122 and a knowledge point extraction subunit 123. A general large model is used to extract and mark the categories, keywords and knowledge points of each question in the existing architectural question bank through the category extraction subunit 121, the keyword extraction subunit 122 and the knowledge point extraction subunit 123.

[0037] The new question bank automatic generation module 130 uses the questions and their categories, keywords, knowledge points and answers of the existing question bank analysis module 120 as training data for the general large model to form a large model for automatic generation of question banks. The large model for automatic generation of question banks uses the vectorized building data of the building data analysis module 110 as input to automatically generate a building question bank containing questions and their categories, keywords, knowledge points and answers.

[0038] Among them, the keyword extraction subunit 122 extracts entity words in the title and assigns attributes to the title with the entity words. The professional language generation ability of the general large model is strengthened through the entity words to improve professionalism. By extracting knowledge points, the hitting ability of the knowledge points of the large model automatically generated by the question bank can be improved.

[0039] Please refer to Figure 2 The large-model-based automatic generation system 100 of the architectural question bank provided by the embodiment of the present invention may further include:

[0040] The automatic answer correction module 140 automatically generates a large model through the question bank and compares the answer of the respondent to the question with the answer of the architectural question bank for automatic correction.

[0041] The method and system for automatically generating an architectural question bank based on a large model provided by the embodiment of the present invention analyzes architectural data and performs vectorized marking to form vectorized architectural data for understanding, querying and comparison by a general large model; analyzes the existing question bank to mark the category, keyword, and knowledge point of each question; uses the questions and their categories, keywords, knowledge points, and answers analyzed by the existing question bank as training data for the general large model to form a large model for automatically generating a question bank, and uses the vectorized architectural data as input to automatically generate an architectural question bank containing questions and their categories, keywords, knowledge points, and answers. Compared with the traditional architectural question bank produced by industry experts, it has the advantages of a wide coverage of knowledge points, high efficiency, and a large number of questions.

[0042] The method and system for automatically generating an architecture question bank based on a large model provided in an embodiment of the present invention can simultaneously provide reference answers to various questions in the architecture question bank, and use multimodal analysis, vectorized retrieval and large model technology to form a large model for automatically generating questions to produce a new architecture question bank, which can realize fast, large-scale and automatic generation of architecture question banks and reference answers, and can generate reference answers based on questions.

[0043] The present invention is not limited to the above-mentioned specific implementation modes. Obviously, the above-mentioned embodiments are only some embodiments of the embodiments of the present invention, but not all embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention. Those skilled in the art can make other levels of modifications and changes to the present invention. In this way, if these modifications and changes of the present invention fall within the scope of the claims of the present invention, the present invention is also intended to include these changes and changes.

Claims

1. A method for automatically generating a building question bank based on a large model, characterized in that: include: Building data analysis: Analyze the text, images, tables and formulas in the building data, and use word embedding technology to vectorize the parsed text, images, tables and formulas to form vectorized building data; Existing question bank analysis: Use a general large model to mark the categories, keywords, and knowledge points of each question in the existing architectural question bank; Automatic generation of new question bank: The questions and their categories, keywords, knowledge points and answers analyzed in the existing question bank are used as training data for the general large model to form a large model for automatic generation of question bank. The large model for automatic generation of question bank takes vectorized architectural data as input to automatically generate an architectural question bank containing questions and their categories, keywords, knowledge points and answers.

2. The method for automatically generating a large model-based architectural question bank according to claim 1 is characterized in that: Also includes: Automatically correct answers: A large model is automatically generated through the question bank to compare the test taker's answers to the questions with the answers in the architectural question bank for automatic correction.

3. The method for automatically generating a large model-based architectural question bank according to claim 1, characterized in that: In the step of parsing building data, optical character recognition technology is used to identify and extract text content in words, images, tables and formulas.

4. The method for automatically generating a large model-based architectural question bank according to claim 1, characterized in that: In the step of automatically generating a new question bank, the categories of the questions include short-answer questions, true-or-false questions, calculation questions, definition of terms and multiple-choice questions; in the step of automatically generating a new question bank, a large model for automatically generating a question bank is used to automatically ask questions based on vectorized building data, associate knowledge points and keywords to automatically generate short-answer questions, true-or-false questions, calculation questions, definition of terms and multiple-choice questions.

5. A large-model-based automatic generation system for architectural question banks, characterized in that: The method for automatically generating a large model-based architectural question bank according to any one of claims 1 to 4 comprises: The building data parsing module includes a text parsing subunit, an image parsing subunit, a table parsing subunit and a formula parsing subunit, which are used for parsing text, images, tables and formulas respectively, and vectorizes the parsed text, images, tables and formulas by combining word embedding technology to form vectorized building data; The existing question bank parsing module includes a category extraction subunit, a keyword extraction subunit and a knowledge point extraction subunit. The general large model is used to extract and mark the categories, keywords and knowledge points of each question in the existing architectural question bank through the category extraction subunit, the keyword extraction subunit and the knowledge point extraction subunit. The new question bank automatic generation module uses the questions and their categories, keywords, knowledge points and answers of the existing question bank parsing module as training data for the general large model to form a large model for automatic question bank generation. The large model for automatic question bank generation uses the vectorized building data of the building data parsing module as input to automatically generate a building question bank containing questions and their categories, keywords, knowledge points and answers.

6. The automatic generation system of architectural question bank based on large model according to claim 5 is characterized in that: Also includes: The automatic answer correction module automatically generates a large model through the question bank, compares the answer of the respondent to the question with the answer of the architectural question bank, and automatically corrects it.

Images