Intelligent questioning and answering method for knowledge in education field integrated with Scoglai teaching concept

Abstract

The invention discloses an intelligent question-answering method for knowledge in the education field by fusing a Scogabi teaching concept, which comprises the following steps of: S1, establishing a Scogabi teaching question-answering instruction, and guiding a model to generate answers of Scogabi questions; s2, performing Scogabi teaching evaluation on the answers to obtain preference data and non-preference data; s3, training the original education large model according to the preference data and the non-preference data, and obtaining a Scoglai education large model; and S4, carrying out teaching questions and answers with Scogabi heuristic through the Scogabi education large model. On the basis of an original education large model, the training process and method for enhancing the Scoglai teaching ability are introduced, so that the trained model has the heuristic teaching ability and can better adapt to an education scene, the applicability of the large model in the education scene is enhanced, and the training efficiency is improved. The product involved in the invention has a wide market prospect and has a wide application range in the education field.

Description

Technical Field

[0001] The present invention belongs to the fields of artificial intelligence, large models and education industry, and specifically relates to an intelligent question-answering method for education field knowledge integrating Socratic teaching philosophy. Background Art

[0002] At present, the new generation of artificial intelligence technology is developing rapidly, and related technologies and products represented by big models are rapidly leading the development of the artificial intelligence industry. Various industry big models are rapidly integrating into various fields and links, enabling thousands of industries to upgrade their intelligence and help social productivity to leap forward. Among many industries, the field of education is also a scenario where the current artificial intelligence big models play an important role. Education big models can be widely used in the field of education. For example, in teaching scenarios for teachers, it can assist teachers in setting questions, generating lesson plans, and correcting homework, etc., effectively liberating teachers' productivity. In the learning process for students, it can provide knowledge questions and answers, wrong question analysis, and personalized learning suggestions to improve students' learning efficiency.

[0003] From general big models to education-specific big models, it is not just a simple fine-tuning and optimization based on the general big models, but it is necessary to take the "learner-centered" concept as the underlying logic of research and development, and integrate it into the entire process of algorithm model architecture design and prototype development, so as to design a big model that meets the educational scenario. Compared with big models in other fields, educational big models not only need to have a large amount of educational knowledge and be able to perform complex reasoning and teaching knowledge problems; educational big models should also understand and understand education, and have corresponding educational concepts and methods. In the teaching scenario, heuristic teaching is a common teaching ability. According to the teaching tasks and the objective laws of learning, it starts from the actual situation of students and adopts a variety of methods to inspire students' thinking as the core, so as to mobilize students' initiative and enthusiasm in learning. In heuristic education, the typical method is the Socratic teaching method. This method promotes problem solving in an interactive question-and-answer manner. During the question-and-answer process, both teachers and students are responsible for promoting the development of dialogue problems by asking questions. The teacher in the dialogue asks questions to students, constantly reveals the contradictions in the other party's answers, and guides students to summarize general conclusions. At the same time, students can also ask questions to teachers to seek answers to questions.

[0004] Existing educational big models are usually constructed in two ways. One is to obtain them based on the general domain big model by using domain knowledge enhanced training and instruction fine-tuning. Typical implementations include Educhat (Yuhao D, Zhikai L, Yiyang G, Yong L, Jianghao Y, Jiaju L, Linhao Y, Zhiyan T, Yougen Z, Yilei W, Aimin Z, Ze Z, Qin C, Jie Z, Liang H, Xipeng Q, et al. EduChat: A Large-Scale Language Model-based Chatbot System for Intelligent Education[J], CoRR, 2023, abs / 2308.02773) and CourseGPT-zh (Zheyan Q, Lu Y, Zitong Y, Wenbo W, Xing Z, etal. CourseGPT-zh: an Educational Large Language Model Based on Knowledge Distillation Incorporating Prompt Optimization[J], CoRR, 2024, abs / 2405.04781). The other is to implement it directly based on a general large model plus retrieval enhancement. It does this by sharding professional data in the field of education and storing it in a vector database. When applied, it uses retrieval enhancement to retrieve relevant context from the vector database to support the answering of questions. A typical representative in this regard is Tianjin University's Tianwen Campus Intelligent Question and Answer System Twen.ai (https: / / news.tju.edu.cn / info / 1003 / 71646.htm). Although the models trained by the above two methods have the ability to solve problems, they lack the corresponding educational understanding and teaching methods, resulting in their lack of professionalism and applicability.

[0005] During the construction process, the existing educational big models mainly use pre-training, fine-tuning enhancement, and knowledge retrieval enhancement to achieve educational knowledge questions and answers and complete specific tasks in the field of education. These applications have greatly improved the efficiency of teaching and the convenience of students acquiring knowledge. However, the existing educational big models ignore the characteristics of education in the teaching process, that is, the cultivation of students' autonomous learning ability. The existing educational big models directly output answers when conducting educational knowledge questions and answers. This process is not conducive to the cultivation of students' autonomous exploration, thinking, and problem-solving abilities, and it is easy to cause students to be lazy in acquiring new knowledge. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the present invention proposes an intelligent question-answering method for educational knowledge that integrates the Socratic teaching philosophy, which solves the problem that the existing large education model ignores the cultivation of students' autonomous learning ability in the teaching process.

[0007] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: an intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy, comprising the following steps: S1. Establish Socratic teaching question-and-answer instructions to guide the model to generate answers to Socratic questions; S2, conduct Socratic teaching evaluation on the answers to obtain preference data and non-preference data; S3, training the original education model based on the preference data and non-preference data to obtain the Socratic education model; S4. Conduct Socratic-inspired teaching questions and answers through the Socratic education model.

[0008] Further: S1 specifically includes: Question-answering data are collected from public datasets as original seeds. Socratic instruction prompt templates are used according to the original seeds to input questions into the model and guide the model to generate answers to Socratic questions.

[0009] Further: S2 comprises the following sub-steps: S21, input the answer into the evaluation model, evaluate it based on the evaluation indicators of Socratic teaching through the evaluation model, and calculate the indicator value of each evaluation indicator; S22. Calculate the evaluation score according to the index values ​​of all evaluation indicators, select the answer with the highest score as the preferred data, and select the answer with the lowest score as the non-preferred data.

[0010] Further: in said S21, the evaluation indicators of Socratic teaching include question nature, relevance of the topic, helpfulness and whether the answer is revealed; The method for calculating the index value of questionability is as follows: determine whether the answer contains at least one question directed to the student. If so, the index value is 1; if not, the index value is 0; The method for calculating the index value of the relevance of the topic is specifically as follows: setting the index value according to the relevance of the answer to the ongoing conversation, and the index value ranges from 1 to 5; The method for calculating the index value of the helpfulness of the topic is as follows: the index value is set according to the degree of usefulness of the answer in providing guidance and support to students, and the index value ranges from 1 to 5; The method for calculating the index value of whether the answer is revealed is specifically as follows: determine whether the answer directly reveals the answer, if so, the index value is 0, if not, the index value is 1.

[0011] Further: In S22, the evaluation score is calculated The specific expression is:

[0012] In the formula, is the problematic indicator value, is the index value of the relevance of the topic, The helpfulness indicator value of the theme, is an indicator value indicating whether the answer is revealed.

[0013] Further: in said S3, the original educational model includes a role-playing model and a reference model, and said S3 includes the following sub-steps: S31, concatenating the preference data and non-preference data and inputting them into the role-playing model and the reference model, and respectively calculating the scores of the corresponding role-playing model and the reference model on the preference data and non-preference data; S32, calculating the reward of the role-playing model and the reward of the reference model according to the score; S33. Train the role-playing model based on the reward calculation loss function to obtain the Socrates education model.

[0014] The beneficial effect of the above further scheme is: the present invention adopts the DPO preference optimization algorithm to directly use the preference data to perform fine-tuning on the basis of the original education model, and finally makes the model generation mode meet the user preferences.

[0015] Further: In said S32, the reward for playing the model and the reference model reward The specific expression is:

[0016]

[0017] In the formula, To score the role-playing model on the preference data, To score the role model on non-preferred data, Score the reference model on the preference data, Score the reference model on the non-biased data.

[0018] Further: In S33, the loss function The specific expression is:

[0019] In the formula, To take the logarithm, is the sigmoid activation function, is the weight parameter.

[0020] The beneficial effects of the present invention are: (1) The present invention provides an intelligent question-answering method for educational knowledge that integrates the Socratic teaching philosophy. On the basis of the original large educational model, the present invention introduces a training process and method for enhancing the Socratic teaching ability, so that the trained model has heuristic teaching ability and can better adapt to educational scenarios, thereby enhancing the applicability of the large model in educational scenarios. The products involved have broad market prospects and a relatively broad application range in the field of education.

[0021] (2) The method of the present invention uses a direct preference optimization algorithm to train the Socratic teaching ability of the model, and performs feedback enhancement training on the model, so that the generated Socratic education model has the Socratic heuristic teaching question-answering ability, enhances its teaching characteristics, and thus better provides intelligent teaching services for students and teachers, achieving convenience and low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The present invention is a flow chart of an intelligent question-answering method for educational knowledge that integrates Socratic teaching philosophy.

[0023] Figure 2 The process for generating answers to Socratic questions.

[0024] Figure 3 Training process for the Socratic education model.

[0025] Figure 4 The present invention is designed to provide an overall process of an intelligent question-answering method for educational knowledge that integrates Socratic teaching philosophy. DETAILED DESCRIPTION

[0026] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0027] like Figure 1 As shown, in one embodiment of the present invention, an intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy comprises the following steps: S1. Establish Socratic teaching question-and-answer instructions to guide the model to generate answers to Socratic questions; S2, conduct Socratic teaching evaluation on the answers to obtain preference data and non-preference data; S3, training the original education model based on the preference data and non-preference data to obtain the Socratic education model; S4. Conduct Socratic-inspired teaching questions and answers through the Socratic education model.

[0028] The S1 is specifically: Question-answering data are collected from public datasets as original seeds. Socratic instruction prompt templates are used according to the original seeds to input questions into the model and guide the model to generate answers to Socratic questions.

[0029] like Figure 2 As shown, in this embodiment, the present invention first randomly samples 1000 question-answer pairs from the mathematics question-answering dataset School-Math in the BELL dataset, and extracts questions from them as original seeds. Then, the Socratic instruction prompt template is input into the model to guide the model to generate questions for these questions and guide the model to generate answers to Socratic questions.

[0030] In this embodiment, in order to enhance the questioning ability of the model, based on the principles of Socratic questioning, the designed Socratic instruction prompt template is as follows: "You are a Socratic mentor. When responding to students, please follow the following principles: - Ask thought-provoking, open-ended questions that challenge students' preconceptions, - Promote open and respectful dialogue between students, create an environment where different perspectives are valued and students feel comfortable sharing their ideas. - Actively listen to students' answers, pay close attention to their underlying thought processes, and really strive to understand their perspectives. - Guide students to explore topics by encouraging them to discover answers independently rather than providing direct answers to improve their reasoning and analytical skills. Please respond to students' questions next."

[0031] In this embodiment, for each question, the present invention allows the model to sample and generate multiple different answers, and uses a referee model to evaluate these generated questions from different dimensions to evaluate whether the model's answers conform to the Socratic heuristic questioning method.

[0032] The S2 comprises the following sub-steps: S21, input the answer into the evaluation model, evaluate it based on the evaluation indicators of Socratic teaching through the evaluation model, and calculate the indicator value of each evaluation indicator; S22. Calculate the evaluation score according to the index values ​​of all evaluation indicators, select the answer with the highest score as the preferred data, and select the answer with the lowest score as the non-preferred data.

[0033] In Socratic heuristic teaching, the core is to let the model learn to ask Socratic heuristic questions to the questions raised by students, and guide students to think for themselves and then discover the essence of the problem. Therefore, in this embodiment, it is crucial to evaluate whether the answers generated by the model have Socratic heuristic questions. The present invention extracts a comprehensive evaluation framework to evaluate whether it is suitable for subsequent training models. Specifically, in the present invention, a third-party closed-source model GPT4o is used as an evaluation model to evaluate the orders generated in the first step of the model. In terms of evaluation indicators, the present invention evaluates from four evaluation criteria, namely: questions, relevance of topics (on-topics), helpfulness, and whether the answer is revealed (reveal answer). The specific indicator description and indicator score are shown in the following table.

[0034] Table 1 Description of the Socratic teaching evaluation dimensions and corresponding indicator values

[0035] In S21, the evaluation indicators of Socratic teaching include question nature, relevance of the topic, helpfulness, and whether the answer is revealed; The method for calculating the index value of questionability is as follows: determine whether the answer contains at least one question directed to the student. If so, the index value is 1; if not, the index value is 0; The method for calculating the index value of the relevance of the topic is specifically as follows: the index value is set according to the relevance of the answer to the ongoing conversation, and the index value ranges from 1 to 5. In this embodiment, a value of 1 indicates that the answer is completely irrelevant to the conversation, and a value of 5 indicates that the answer is completely relevant to the topic; The method for calculating the index value of the helpfulness of the topic is specifically as follows: the index value is set according to the usefulness of the answer in providing guidance and support to the students, and the index value ranges from 1 to 5. In this embodiment, a value of 1 indicates that the answer is completely unhelpful to the students, and a value of 5 indicates that the answer is completely useful in guiding the students; The method for calculating the index value of whether the answer is revealed is specifically as follows: determine whether the answer directly reveals the answer, if so, the index value is 0, if not, the index value is 1.

[0036] In this embodiment, the present invention uses the GPT4o model as a referee for evaluation. During the evaluation, the present invention designs a specific prompt word model to let the referee model output specific indicator values ​​for each data of the model according to 4 dimensions. The prompt words designed by the present invention are as follows: "You are a professional dialogue evaluator. Your goal is to evaluate whether the professor allows students to reflect more deeply and think critically. The professors are instructed to apply the Socratic method, in which they should not immediately reveal the answers to students' questions, but should guide them to discover the answers themselves. You will see excerpts of conversations between professors and students. You must evaluate the professor's answers based on the following criteria: 1) Does the professor ask questions? Choose between "yes" or "no". If there is no question mark in the answer, choose "no". 2) Is the answer about the same topic of the conversation? The score range is 1 to 5, where 1 means that the professor's answer deviates from the original topic and 5 means it is completely in line. 3) Is the answer helpful to students? The score range is 1 to 5, where 1 means that the professor's answer may mislead students' thinking and 5 means that the answer can be perfect." 4) Does it reveal the answer immediately? Choose between "yes" or "no". Select "Yes" if the professor's response reveals the answer, so it doesn't require any thinking on the part of the student. Be very strict when performing assessments. You should only assess the answers. Finally, provide the final assessment according to the following JSON format: {"questions":Result Criteria 1, "on-topic":Result Criteria 2, "helpful":Result Criteria 3, "reveal-answer":Result Criteria 4} Do not generate any start or end notes. ” In S22, the evaluation score is calculated. The specific expression is:

[0037] In the formula, is the problematic indicator value, is the index value of the relevance of the topic, The helpfulness indicator value of the theme, is an indicator value indicating whether the answer is revealed.

[0038] In this embodiment, after calculating the evaluation scores for the five different answers generated for each question, all the answers are further sorted according to the scores, and then the answer with the highest score is selected as a good answer and marked as selected, that is, the data that meets Socrates' preference; the answer with the lowest score is regarded as a bad answer and marked as rejected. After marking according to the scores, each question has two answers, namely, preference data and non-preference data; finally, these labeled preference data are trained on the model using the preference optimization algorithm to obtain a large model with Socrates' teaching ability.

[0039] In S3, the original educational model includes a role-playing model and a reference model, and S3 includes the following sub-steps: S31, concatenating the preference data and non-preference data and inputting them into the role-playing model and the reference model, and respectively calculating the scores of the corresponding role-playing model and the reference model on the preference data and non-preference data; S32, calculating the reward of the role-playing model and the reward of the reference model according to the score; S33. Train the role-playing model based on the reward calculation loss function to obtain the Socrates education model.

[0040] In this embodiment, if Figure 3 As shown, based on the labeled preference data generated in the second step, the preference optimization algorithm is used to train the model so that the responses generated by the model can meet human preferences, that is, the model is more inclined to generate answers that choose preferences. In order to achieve the above goals, the DPO algorithm is used in the present invention to achieve the above goals. The DPO preference optimization algorithm directly uses the preference data to fine-tune the original education model, and finally makes the model generation mode meet user preferences.

[0041] In S32, the reward for playing the model and the reference model reward The specific expression is:

[0042]

[0043] In the formula, To score the role-playing model on the preference data, To score the role model on non-preferred data, Score the reference model on the preference data, Score the reference model on the non-biased data.

[0044] In this embodiment, during training, each question and the preferred answer and the non-preferred answer are spliced ​​together and input into the role-playing model and the reference model respectively, and then the scores of the corresponding role-playing model and the reference model on the preference data and the non-preference data are calculated respectively, and the preference and non-preference scores are calculated according to the maximum probability of relief predicted by the model. The reward of the role-playing model and the reward of the reference model are calculated according to the scores respectively, and the KL divergence of the two rewards is calculated as the loss to train the role-playing model. After the training is completed, the role-playing model is the Socratic education model obtained through training.

[0045] In S33, the loss function The specific expression is:

[0046] In the formula, To take the logarithm, is the sigmoid activation function, is the weight parameter.

[0047] like Figure 4 As shown, in this embodiment, an intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy is designed by the present invention, including three steps of generating Socratic teaching question-answering instructions, constructing Socratic preference data and non-preference data, and training the Socratic education model.

[0048] For the Socratic teaching question-answer instruction generation process, the process includes random sampling of instructions, prompt instructions based on a fixed template, and generating five different responses for each prompt.

[0049] For Socrates' method of constructing preference data and non-preference data, the evaluation dimensions of Socrates' instructions include four dimensions: problematic nature, relevance of the topic, helpfulness, and whether the answer is revealed. The scores are calculated using uniform weighting and normalization to obtain specific scores.

[0050] For the Socratic Educational Large Model Training Method, the model is trained using a direct preference optimization algorithm to enhance the model's Socratic question-answering capabilities.

[0051] The beneficial effects of the present invention are as follows: the present invention provides an intelligent question-answering method for educational knowledge that integrates the Socratic teaching philosophy, and introduces a training process and method for enhancing the Socratic teaching ability on the basis of the original educational large model, so that the trained model has heuristic teaching ability and can better adapt to educational scenarios, thereby enhancing the applicability of the large model in educational scenarios. The products involved have broad market prospects and a relatively broad application range in the field of education.

[0052] The method of the present invention uses a direct preference optimization algorithm to train the Socratic teaching ability of the model, and performs feedback enhancement training on the model, so that the generated Socratic education large model has the Socratic heuristic teaching question-answering ability, enhances its teaching characteristics, and thus better provides intelligent teaching services for students and teachers, achieving convenience and low cost.

[0053] In the description of the present invention, it is necessary to understand that the orientation or positional relationship indicated by the terms "center", "thickness", "upper", "lower", "horizontal", "top", "bottom", "inner", "outer", "radial", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", and "third" are used only for descriptive purposes, and cannot be understood as indicating or implying the relative importance or the number of implicitly specified technical features. Therefore, the features defined by "first", "second", and "third" may explicitly or implicitly include one or more of the features.

Claims

1. An intelligent question-answering method for educational knowledge that integrates Socratic teaching philosophy, characterized in that: The following steps are involved: S1. Establish Socratic teaching question-and-answer instructions to guide the model to generate answers to Socratic questions; S2, conduct Socratic teaching evaluation on the answers to obtain preference data and non-preference data; S3, training the original education model based on the preference data and non-preference data to obtain the Socratic education model; S4. Conduct Socratic-inspired teaching questions and answers through the Socratic education model.

2. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 1 is characterized in that: The S1 is specifically: Question-answering data are collected from public datasets as original seeds. Socratic instruction prompt templates are used according to the original seeds to input questions into the model and guide the model to generate answers to Socratic questions.

3. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 1 is characterized in that: The S2 comprises the following sub-steps: S21, input the answer into the evaluation model, evaluate it based on the evaluation indicators of Socratic teaching through the evaluation model, and calculate the indicator value of each evaluation indicator; S22. Calculate the evaluation score according to the index values ​​of all evaluation indicators, select the answer with the highest score as the preferred data, and select the answer with the lowest score as the non-preferred data.

4. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 3 is characterized in that: In S21, the evaluation indicators of Socratic teaching include question nature, relevance of the topic, helpfulness, and whether the answer is revealed; The method for calculating the index value of questionability is as follows: determine whether the answer contains at least one question directed to the student. If so, the index value is 1; if not, the index value is 0; The method for calculating the index value of the relevance of the topic is specifically as follows: setting the index value according to the relevance of the answer to the ongoing conversation, and the index value ranges from 1 to 5; The method for calculating the index value of the helpfulness of the topic is as follows: the index value is set according to the degree of usefulness of the answer in providing guidance and support to students, and the index value ranges from 1 to 5; The method for calculating the index value of whether the answer is revealed is specifically as follows: determine whether the answer directly reveals the answer, if so, the index value is 0, if not, the index value is 1.

5. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 4 is characterized in that: In S22, the evaluation score is calculated. The specific expression is: In the formula, is the problematic indicator value, is the index value of the relevance of the topic, The helpfulness indicator value of the theme, is the indicator value of whether to reveal the answer.

6. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 1 is characterized in that: In S3, the original educational model includes a role-playing model and a reference model, and S3 includes the following sub-steps: S31, concatenating the preference data and non-preference data and inputting them into the role-playing model and the reference model, and respectively calculating the scores of the corresponding role-playing model and the reference model on the preference data and non-preference data; S32, calculating the reward of the role-playing model and the reward of the reference model according to the score; S33. Train the role-playing model based on the reward calculation loss function to obtain the Socrates education model.

7. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 6 is characterized in that: In S32, the reward for playing the model and the reference model reward The specific expression is: In the formula, To score the role-playing model on the preference data, To score the role model on non-preferred data, Score the reference model on the preference data, Score the reference model on the non-biased data.

8. The intelligent question-answering method for educational knowledge integrating Socratic teaching philosophy according to claim 7 is characterized in that: In S33, the loss function The specific expression is: In the formula, To take the logarithm, is the sigmoid activation function, is the weight parameter.

Images