CN-121997061-A - Intelligent guiding method, device and equipment for project learning

CN121997061ACN 121997061 ACN121997061 ACN 121997061ACN-121997061-A

Abstract

The invention provides an intelligent guiding method, device and equipment for project-oriented learning, which comprises the steps of identifying a learning stage according to acquired user input text and contextual information, determining a learning target matched with the learning stage, acquiring a corresponding stage guiding strategy according to the learning stage, calling a plurality of agents, driving each agent to generate feedback content corresponding to the user input text according to the stage guiding strategy, wherein each agent is configured with different role functions in advance, and carrying out fusion processing on the feedback content generated by each agent according to the learning target to generate multidimensional collaborative guiding information. The invention breaks through the limitation of single visual angle of a single language model, ensures that multidimensional collaborative guiding information not only maintains the professional depth of multiple intelligent agents, but also has good consistency and readability, so that students can obtain structured guidance under a unified context, the cognitive load is reduced, and the learning efficiency is improved.

Inventors

Zha Siyu
LIU YUJIA
GONG JIANGTAO
XU YINGQING

Assignees

清华大学

Dates

Publication Date: 20260508
Application Date: 20251229

Claims (10)

1. An intelligent guiding method for item-oriented learning is characterized by comprising the following steps: identifying a learning stage according to the acquired text input by the user and the acquired context information, and determining a learning target matched with the learning stage; Acquiring a corresponding stage guiding strategy according to the learning stage, and calling a plurality of agents to drive each agent to generate feedback content corresponding to the user input text according to the stage guiding strategy, wherein each agent is configured with different role functions in advance; And according to the learning target, carrying out fusion processing on feedback contents generated by each agent to generate multidimensional collaborative guiding information.
2. The intelligent guidance method for item-oriented learning according to claim 1, wherein obtaining a corresponding stage guidance policy according to the learning stage, and invoking a plurality of agents to drive each agent to generate feedback content corresponding to the user input text according to the stage guidance policy, comprises: Based on the learning stage, matching a corresponding stage guiding strategy from a preset strategy library, wherein the preset strategy library is constructed based on different learning stages and corresponding stage guiding strategies thereof in advance; Invoking a plurality of intelligent agents, inputting the stage guiding strategy and the user input text into each intelligent agent, driving each intelligent agent to perform visual angle conversion on the stage guiding strategy according to the role function of each intelligent agent so as to generate an emphasis point matched with the corresponding role function, and analyzing and processing the user input text according to the emphasis point so as to generate corresponding feedback content.
3. The intelligent guidance method for item-oriented learning according to claim 2, wherein when the learning phase is a question definition phase, the corresponding phase guidance strategy is used for helping students clearly express questions by clarity questions, task range limitation and/or guidance modes of specific target objects; When the learning phase is a creative generation phase, the corresponding phase guiding strategy is used for encouraging students to generate diversified alternative schemes through guiding modes of analogy, transformation, combination and/or reverse thinking; when the learning stage is a scheme integration and planning stage, the corresponding stage guiding strategy is used for guiding students to construct evaluation standards, comparing the advantages and disadvantages of different schemes and identifying risks and resource requirements so as to assist in perfecting the implementation scheme; When the learning stage is a thinking-back stage, the corresponding stage guiding strategy is used for guiding students to review the process, comb key decisions and/or refine movable experience so as to promote the students to conduct meta-cognition thinking-back; and when the role functions of the intelligent agents are task coordination, the corresponding visual angles are visual angles focusing on task rhythm and stage targets.
4. The intelligent guidance method for item-oriented learning according to claim 3, wherein when the role of the agent is knowledge support, the corresponding view angle is a view angle focusing on theoretical support and material expansion; When role functions of the intelligent agent are creative promotion, the corresponding visual angle is a visual angle for expanding attention ideas; when the role of the intelligent body is reviewed, the corresponding view angle is the view angle focusing on the duplication and value improvement.
5. The intelligent guidance method for item-oriented learning according to claim 1, wherein the fusion processing is performed on feedback content generated by each agent according to the learning objective to generate multidimensional collaborative guiding information, comprising: Determining fusion weights for the feedback contents of all the agents according to the learning targets corresponding to the current learning stage; adaptively adjusting feedback contents generated by each agent by utilizing a learning state determined based on a previous session so as to adjust the language and semantic depth of each feedback content; And generating multidimensional collaborative guiding information according to the adjusted fusion weights of the feedback contents and the feedback contents of the intelligent bodies.
6. The intelligent guidance method for item-oriented learning according to claim 1, wherein after performing fusion processing on feedback content generated by each agent according to the learning objective to generate multidimensional collaborative guiding information, the method comprises: Acquiring an interactive text returned by a user based on the multidimensional collaborative guiding information, performing language feature analysis on the interactive text, and identifying whether a predefined state keyword or a target expression mode exists in the interactive text; When the confusion state of stagnation, poor expression or repeated expression exists in the interactive text is determined, triggering an adaptive adjustment mechanism of a corresponding stage guiding strategy to adjust the corresponding stage guiding strategy into a refined guiding strategy comprising specific prompt and resolution of the step problem; When it is determined that there is a deep thought state of interest, excitement signal, or logic depth signal in the interactive text, an adaptive adjustment mechanism of the corresponding phase-guiding policy is triggered to adjust the corresponding phase-guiding policy to include an advanced challenge policy encouraging abstract induction or associated broad context.
7. The intelligent guidance method for item-oriented learning according to claim 1, wherein after performing fusion processing on feedback content generated by each agent according to the learning objective to generate multidimensional collaborative guiding information, the method further comprises: For the current session, extracting key content, combining the learning stage and student decision information, writing into an internal state record, and generating a learning path comprising stage switching nodes, stay time and key decision points; Before acquiring the corresponding stage guidance strategy according to the learning stage, the method comprises the following steps: When determining that the user input text contains the backtracking intention, analyzing the backtracking intention to acquire a target repositioning stage; Retrieving the target relocation phase in the internal state record based on a learning path generated by a preceding session and retrieving context information associated with the target relocation phase; And restoring the session state to the target relocation stage by using the context information, and continuing to execute interaction.
8. The intelligent guidance method for item-oriented learning of claim 1, wherein the identifying a learning phase based on the acquired user input text and context information comprises: inputting the acquired user input text into a stage identification model to perform joint coding on the user input text and context information comprising a pre-stage dialogue learning stage and a session context, so as to obtain joint coding characteristics; And determining the conditional probability of the joint coding feature relative to each preset learning stage in a preset typical learning stage set, and screening a target learning stage from the preset typical learning stage set according to a conditional probability maximization principle as a learning stage corresponding to the user input text.
9. An intelligent guiding device for item-oriented learning, comprising: The stage identifying module is used for identifying a learning stage according to the acquired text input by the user and the acquired context information, and determining a learning target matched with the learning stage; the multi-agent feedback module acquires a corresponding stage guiding strategy according to the learning stage, and invokes a plurality of agents to drive each agent to generate feedback content corresponding to the user input text according to the stage guiding strategy, wherein each agent is configured with different role functions in advance; And the fusion processing module is used for carrying out fusion processing on feedback contents generated by each intelligent agent according to the learning target to generate multidimensional collaborative guiding information.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and running on the processor, characterized in that the processor implements the intelligent guidance method for item-oriented learning according to any of claims 1 to 8 when executing the computer program.

Description

Intelligent guiding method, device and equipment for project learning Technical Field The invention relates to the technical field of artificial intelligence, in particular to an intelligent guiding method, device and equipment for project-oriented learning. Background The generated artificial intelligence (GENERATIVE AI) is widely applied to the education field and becomes an important tool for assisting students in learning and knowledge acquisition. With the evolution of education ideas, teaching activities increasingly emphasize the cultivation of complex abilities such as project learning, interdisciplinary exploration and creative problem solving, and these activities require students to undergo continuous cognitive stages such as problem definition, information collection, divergent thinking, scheme integration, design iteration and dislike assessment in a longer period. The application target of the AI technology in the education scene is developing from single knowledge retrieval to support complete learning process and guiding deep thinking training, and aims to serve as an intelligent teacher to improve teaching effect and learning quality. At present, an AI system applied to the education field mainly adopts a question-answer based instant response mode, namely, by receiving text input by a user, the AI system utilizes the semantic understanding capability of a model with a single role, relies on a built-in universal knowledge base of the model to conduct question-answer or content creation so as to give instant and direct answers or complete text for questions of students. However, the question-answering mode of a single role is difficult to simulate the multi-role collaborative guidance environment of teachers, experts, directors and the like in project learning, so that feedback information is unidirectional, flat and lacks layering of professional viewing angles, the requirements of students on diversified guidance in different cognitive stages cannot be met, the AI system lacks stage identification and nonlinear flow management capabilities of learning processes, frequent reciprocation and jump of the students in task advancing cannot be effectively conducted, context continuity in long conversations is difficult to maintain, and advice is scattered, learning flow fragmentation and tracking is difficult. In addition, the AI system lacks a recording and arranging mechanism for learning process data, cannot identify and retain the thinking process, key decision and thinking rejection content of students, so that students are difficult to review a thinking path and a teacher is difficult to evaluate based on the process, and meanwhile, the system lacks self-adaptive adjustment capability for the students in the case of cognition stagnation or emotion fluctuation, cannot dynamically adjust interaction strategies like a real director, and limits effective support of the learning motivation and the adherence. Disclosure of Invention The invention provides an intelligent guiding method, device and equipment for item-oriented learning, which are used for solving the defects that feedback information is unidirectional, flat and lacked professional visual angle layering caused by a question-answer mode of a single role in the prior art, breaking the limitation of single visual angle of a single language model, ensuring that multidimensional collaborative guiding information not only maintains the professional depth of multiple intelligent agents, but also has good consistency and readability, so that students can obtain structured guiding under a unified context, the cognitive load is reduced, and the learning efficiency is improved. The invention provides an intelligent guiding method for project-oriented learning, which comprises the steps of identifying a learning stage according to acquired user input text and contextual information, determining a learning target matched with the learning stage, acquiring a corresponding stage guiding strategy according to the learning stage, calling a plurality of agents, and driving each agent to generate feedback content corresponding to the user input text according to the stage guiding strategy, wherein each agent is configured with different role functions in advance, and carrying out fusion processing on the feedback content generated by each agent according to the learning target to generate multidimensional collaborative guiding information. The intelligent guiding method for item-oriented learning comprises the steps of acquiring a corresponding stage guiding strategy according to a learning stage, calling a plurality of agents, driving the agents to generate feedback content corresponding to a user input text according to the stage guiding strategy, and matching the corresponding stage guiding strategy from a preset strategy library based on the learning stage, wherein the preset strategy library is constructed based on different learning stages and th