A method for measuring and providing feedback on students' psychological resilience based on a large language model

Abstract

The present invention relates to a method for measuring and providing feedback on student psychological resilience based on a large language model, and belongs to the technical fields of natural language processing, large models, and mental health intelligent bodies. The method first constructs a psychological assessment prompt word template, a prompt word template for investigating leisure participation, and a prompt word template for investigating positive emotion arousal; then, multiple rounds of interactive dialogues are generated according to the above prompts, and based on the multiple rounds of dialogue records with the user, a prompt word template is constructed to predict the psychological resilience score; finally, based on the dialogue records and the predicted psychological resilience results, intervention suggestions for leisure activities, positive emotion training tasks, and thinking training activities are generated. The present invention can realize the automated execution of student psychological resilience measurement tasks, can effectively improve the interactivity, accuracy, and targeted feedback of psychological assessments, and enhance the suitability and practical value of the college mental health support system under the background of new technologies.

Description

Technical Field

[0001] The present invention relates to a student psychological resilience measurement and feedback method based on a large language model, belonging to the technical fields of natural language processing, large models, and mental health intelligent agents. Background Art

[0002] Resilience refers to an individual's ability to adapt and recover in the face of stress, challenges, and adversity. It is a core indicator of student mental health development. Traditional resilience assessments rely primarily on paper-and-pencil questionnaires (such as the CD-RISC scale) or structured interviews, which struggle to achieve dynamic interaction, personalized expression, and intelligent analysis. Furthermore, the assessment process relies on specialized personnel, resulting in high costs and low efficiency.

[0003] In recent years, the rapid advancement of large language models (such as ChatGPT) in natural language understanding and generation has made it possible to leverage them for personalized psychological conversations, sentiment analysis, and language feature inference. However, existing methods for emotional companionship or emotion recognition lack structured modeling and automated assessment mechanisms for core psychological constructs such as resilience. Furthermore, traditional psychological assessment methods generally overlook the correlation between external behavioral variables such as "leisure participation" and "emotional arousal" and psychological states, making it difficult to develop a comprehensive assessment perspective.

[0004] Therefore, there is an urgent need for an assessment and intervention method tailored to the structure of resilience, equipped with personalized language guidance and automated scoring and feedback capabilities. This method can not only achieve efficient and low-cost resilience assessment, but also serve as an intelligent supplement to mental health support systems in educational settings. This invention is proposed in this context. Summary of the Invention

[0005] The purpose of this invention is to provide a student psychological resilience measurement and feedback method based on a large language model, aiming to solve the technical problem that traditional methods are difficult to form a comprehensive evaluation perspective, and to achieve multidimensional measurement and intervention feedback of psychological resilience.

[0006] To achieve the above objectives, the present invention adopts a technical solution: a method for measuring and providing feedback on student resilience based on a large language model. This method constructs multiple prompt templates to prompt the large language model to conduct multi-dimensional assessment and intervention of resilience. From the three aspects of "leisure participation, positive emotions, and resilience", the method automatically guides, identifies, scores, and provides feedback on resilience-related traits through natural language interaction. Compared with directly using the large language model for emotional dialogue or simple question-and-answer analysis, this method can effectively improve the targeted nature of psychological assessments, the naturalness of language interaction, and the practicality of feedback generation. It includes the following steps:

[0007] Step 1: Build a psychological assessment prompt word template and set several semantic tags. Use the psychological assessment prompt word template to guide the large language model to generate questions related to the semantic tags to initiate dialogue interaction with the user.

[0008] Step 2: Construct a prompt word template for investigating leisure participation, guiding the large language model to generate questions to explore the frequency, type, and subjective experience of users' leisure activities;

[0009] Step 3: Construct a prompt word template to investigate positive emotional arousal, prompting the large language model to generate questions related to positive emotional arousal and positive experience recall;

[0010] Step 4: Based on the constructed psychological assessment prompt word template, the leisure participation survey prompt word template, and the positive emotion arousal survey prompt word template, the large language model is used to conduct multiple rounds of interactive question and answer sessions with the user;

[0011] Step 5: Set up a prompt word template for performing psychological resilience feature extraction and multi-dimensional score prediction, and combine multiple rounds of interactive question-and-answer records to obtain the psychological score of the user's semantic label;

[0012] Step 6: Set up a prompt word template for providing intervention suggestions. Based on the psychological scores and multiple rounds of interactive question-and-answer records, generate intervention suggestions on leisure activities, positive emotion training tasks, and thinking training activities.

[0013] The Step 1 is specifically as follows:

[0014] Step 1.1: Set semantic labels, which are composed of multiple psychological dimensions;

[0015] Step 1.2: Set the prompt word template , used to prompt the large language model to generate open-ended psychometric questions for each semantic tag;

[0016] Step 1.3: Set the prompt word template Combined with semantic tags to form prompts and input into the large language model, a set of dialogue questions is generated and bound to the semantic tags, denoted as ,in, is a set of all generated questions, including specific questions and semantic labels of questions, For the A generated problem, is a semantic label;

[0017] Step 1.4: Randomly select each semantic tag from several semantic tags. Questions, dialogue with users, user feedback results are recorded as ,in, is the total number of semantic tags, This will be determined based on the duration of the proposed test.

[0018] The Step 2 is specifically as follows:

[0019] Step 2.1: Set the prompt word template , used for the first interaction between the large language model and the user;

[0020] Step 2.2: Set the prompt word template , used for prompts after the user answers;

[0021] Step 2.3: Use the prompt word template Generate questions and conduct the first conversation with the user, using prompt word templates based on the user's answers When the large language model completes all the active inquiries, the conversation ends and the user's answer in the first conversation is recorded as .

[0022] The Step 3 is specifically as follows:

[0023] Step 3.1: Set the prompt word template , used to guide users to recall positive experiences for the first time;

[0024] Step 3.2: Set the prompt word template , used to dynamically supplement questions during interaction;

[0025] Step 3.3: Use the prompt word template Generate the first round of questions to obtain the user's initial expression of positive emotional experience;

[0026] Step 3.4: Use prompt word templates based on the user's answers in each round Generate supplementary guiding questions and advance the conversation cyclically until the user has expressed the source, process and characteristics of his positive emotional experience. The user's answer in this conversation is recorded as .

[0027] The Step 4 is specifically as follows:

[0028] Step 4.1: The large language model starts a conversation and receives user responses in sequence based on the constructed psychological assessment prompt word template, the leisure participation survey prompt word template, and the positive emotion arousal survey prompt word template.

[0029] Step 4.2: Question and answer based on the three dimensions of psychological assessment, leisure participation and positive emotional arousal. , input it into the large language model to determine whether the current dimension has been expressed. If not, call the prompt template corresponding to the current dimension to generate the next round of questions. If it has been expressed, jump to the next dimension;

[0030] Step 4.3: Repeat Step 4.2 to form a multi-round dialogue sequence with dynamic control capabilities until all dimensions are completed or the dialogue termination conditions are met, and the interaction corpus is output. .

[0031] The Step 5 is specifically as follows:

[0032] Step 5.1: Set the prompt word template , used to perform the task of extracting psychological resilience features, and to convert the interactive corpus Fill in the prompt word template And input into the large language model to obtain the language feature set ;

[0033] Step 5.2: Set the prompt word template , used to guide the large language model to perform scoring reasoning based on language features;

[0034] Step 5.3: Interaction corpus and language feature sets Fill in the prompt word template And input into the large language model, output the psychological score results of all semantic tags ,in, For the The psychological score of each semantic tag is calculated, and the psychological score result and reasoning basis are used as the final output to provide the user's psychological resilience score report. , wherein the reasoning basis is the scoring basis generated by the large language model when writing the prompt word.

[0035] The Step 6 is specifically as follows:

[0036] Step 6.1: Set the prompt word template , used to generate intervention recommendations;

[0037] Step 6.2: Psychological scoring results With score report Fill in the prompt word template The data is input into a large language model to generate multiple intervention suggestions, which are then returned to the user in natural language.

[0038] The beneficial effects of this invention are: by constructing multiple prompt word templates with clear psychological dimension objectives and setting up a dynamic control process, this method uses large language models (such as ChatGPT and Wenxinyiyan) to conduct multiple rounds of interactive question-and-answer sessions, thereby conducting structured measurements of psychological resilience. Compared with traditional scale-based assessments or directly invoking general language models, this method enables a more adaptable, targeted, and expressive psychological assessment process, effectively improving the interactivity, personalization, and interpretability of psychological assessments. It also provides personalized intervention recommendations, promoting the sustainable development of psychological literacy. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of the steps of the present invention. DETAILED DESCRIPTION

[0040] Example 1: Figure 1 As shown in the figure, a method for measuring and providing feedback on students' psychological resilience based on a large language model is described. The specific steps are as follows:

[0041] Step 1: Construct a psychological assessment prompt word template, set several semantic tags, and use the psychological assessment prompt word template to guide the large language model to generate questions related to the semantic tags to initiate dialogue interaction with the user.

[0042] Step 1.1: Set semantic labels. The semantic labels are composed of multiple psychological dimensions. In this embodiment, three psychological dimensions, namely, tenacity, self-reliance, and optimism, are set as semantic labels, which are recorded as ;

[0043] Step 1.2: Set the prompt word template , used to prompt the large language model to generate open psychological assessment questions for each semantic tag. In this embodiment, the prompt word template Specifically:

[0044] “You will have a psychometric conversation with the student, with the goal of assessing Please generate an open-ended question to guide them to talk about their relevant experiences. Replaced with "tenacity", "self-reliance" and "optimism" in turn;

[0045] Step 1.3: Set the prompt word template Combined with semantic tags to form prompts and input into the large language model, a set of dialogue questions is generated and bound to the semantic tags, denoted as ,in, is a set of all generated questions, including specific questions and semantic labels of questions, For the A generated problem, is a semantic label;

[0046] Step 1.4: In the example, each semantic label is randomly selected Questions, dialogue with users, user feedback results are recorded as ,in, This will be determined based on the duration of the proposed test.

[0047] Specifically, in this embodiment, the generated questions include 9 questions:

[0048] (1) “Have you ever been through a particularly difficult time? What kept you going?”

[0049] (2) “How do you cope with and adjust yourself when you encounter failure or continuous setbacks?”

[0050] (3) “Can you share a time when you almost gave up but persevered?”

[0051] (4) “Is there a time when you completed a challenging task independently without any help from others?”

[0052] (5) “Have you ever set a goal and achieved it through consistent effort? Can you describe it in more detail?”

[0053] (6) “Please tell me about a time when you actively sought growth or breakthroughs. What did you do?”

[0054] (7) “When faced with difficulties or setbacks in life, how do you usually view these things?”

[0055] (8) “Can you share a time when you used a positive attitude to resolve negative emotions or influences?”

[0056] (9) “Have you ever been around someone who was very pessimistic, but you were able to influence them by being optimistic?”

[0057] Step 2: Construct a prompt word template for investigating leisure participation, and guide the large language model to generate questions for exploring the frequency, type, and subjective experience of users' leisure activities.

[0058] Step 2.1: Set the prompt word template , used for the first interaction between the large language model and the user, in this embodiment, the prompt word template 2 Specifically:

[0059] "You are investigating students' leisure participation. You have completed the psychological assessment before. The interaction record is Now we need to investigate students' participation in leisure activities, mainly including the types of leisure activities, frequency of leisure participation, and subjective emotional experiences brought about by leisure activities. Please design questions based on the interaction records to inquire about the students' possible involvement in leisure activities. "Among them, Provide feedback to students in Step 1 ;

[0060] Step 2.2: Set the prompt word template , used for prompts after the user answers, in this embodiment, the prompt word template Specifically:

[0061] "You are surveying students about their leisure time activities and students have already provided some responses. The types of activities you would like to survey include According to the interaction record The assessment should include whether all possible activities that students may participate in have been asked. If not, further information should be obtained on their participation in leisure activities, including specific activities, types of leisure activities, frequency of leisure participation, and subjective emotional experiences brought about by leisure activities. For the activity type, usually you can set "sports", "creative" and "social";

[0062] Step 2.3: Use the prompt word template Generate questions and conduct the first conversation with the user, using prompt word templates based on the user's answers When the large language model completes all the active inquiries, the conversation ends and the user's answer in the first conversation is recorded as .

[0063] Specifically, in this embodiment, the initial dialogue and supplementary dialogue generate four types of questions:

[0064] (1) Examples of sports-related questions

[0065] 1) "Do you usually participate in any sports-related leisure activities, such as running, fitness, or playing ball? Can you tell me more about what kind of activities?"

[0066] 2) "How often do you do this type of exercise? Daily, weekly, or occasionally?"

[0067] 3) “How do you typically feel when you participate in this activity? For example, do you feel relaxed, energized, or relieve stress?”

[0068] (2) Examples of creative questions

[0069] 1) “Do you do any creative activities to relax yourself, such as drawing, writing, photography, or crafts?”

[0070] 2) How often do you do this kind of activity? Is it at a set time, or does it happen when inspiration strikes?

[0071] 3) “What kind of emotions do you typically experience when engaging in this type of creative activity? Do you feel healed, immersed, or satisfied?”

[0072] (3) Examples of social questions

[0073] 1) "Do you spend your free time hanging out with friends, chatting, shopping, or participating in social activities like clubs?"

[0074] 2) "Do you think your social activities are frequent or infrequent? Do you schedule them every week?"

[0075] 3) “Does this social activity help you emotionally? For example, by boosting your mood, alleviating loneliness, or increasing your sense of belonging?”

[0076] (4) Supplementary class for checking whether the three types of activities are covered

[0077] 1) "Besides the ones you just mentioned, are there any other leisure activities that you find important or that you participate in regularly?"

[0078] 2) "Do you tend to participate in sports, creative activities, or social activities? Are there any categories that you engage in more or less?"

[0079] 3) “Is there any activity you’d like to try but haven’t had the chance to participate in yet? What’s the reason?”

[0080] Step 3: Construct a prompt word template to investigate positive emotional arousal, prompting the large language model to generate questions related to positive emotional arousal and positive experience recall.

[0081] Step 3.1: Set the prompt word template , used to guide users to recall positive experiences for the first time. In this embodiment, the prompt word template Specifically:

[0082] "You are conducting a student psychological assessment. The student has already completed a psychological assessment and a leisure participation information survey. The interaction record is Now we need to guide students to recall and express a positive experience that reflects their positive emotions, such as pride, joy, hope, and peace of mind. Please generate appropriate open-ended questions based on the interaction records. This is the conversation record from Step 1 and Step 2;

[0083] Step 3.2: Set the prompt word template , used to dynamically supplement questions during interaction, in this embodiment, the prompt word template Specifically:

[0084] "The student's answer just now was , please judge whether the content has fully expressed the type, source and change process of its positive emotions; if the expression is insufficient, please generate an open-ended question to guide it to continue to develop. "Among them, Answer the text for the previous round of students;

[0085] Step 3.3: Use the prompt word template Generate the first round of questions to obtain the user's initial expression of positive emotional experience;

[0086] Step 3.4: Use prompt word templates based on the user's answers in each round Generate supplementary guiding questions and advance the conversation cyclically until the user has expressed the source, process and characteristics of his positive emotional experience. The user's answer in this conversation is recorded as .

[0087] Specifically, in this embodiment, multiple open-ended questions are generated based on different types of records:

[0088] (1) “You mentioned that you like playing basketball and have played as a key player in class competitions. Can you share with me a time when you felt particularly proud or happy while playing basketball?”

[0089] (2) “You said that painting makes you feel relaxed and focused. Is there a time when you felt particularly satisfied or accomplished after completing a piece of work? Can you describe that process?”

[0090] (3) “You mentioned earlier that you were able to persevere when faced with difficulties and received praise from the teacher during a class presentation. Could you please talk about the positive feelings that experience brought you?”

[0091] (4) “You said you and your friends organize a reading or movie club every week. Was there any gathering that made you feel particularly safe or warm? Could you tell us about that time?”

[0092] Step 4: Based on the constructed psychological assessment prompt word template, the prompt word template for investigating leisure participation, and the prompt word template for investigating positive emotional arousal, use the large language model to conduct multiple rounds of interactive question and answer sessions with users.

[0093] Step 4.1: The large language model starts a conversation and receives user responses in sequence based on the constructed psychological assessment prompt word template, the leisure participation survey prompt word template, and the positive emotion arousal survey prompt word template.

[0094] Step 4.2: Question and answer based on the three dimensions of psychological assessment, leisure participation and positive emotional arousal. , input it into the large language model to determine whether the current dimension has been expressed. If not, call the prompt template corresponding to the current dimension to generate the next round of questions. If it has been expressed, jump to the next dimension;

[0095] Step 4.3: Repeat Step 4.2 to form a multi-round dialogue sequence with dynamic control capabilities until all dimensions are completed or the dialogue termination conditions are met, and the interaction corpus is output. .

[0096] Step 5: Set up a prompt word template for performing psychological resilience feature extraction and multi-dimensional score prediction, and combine multiple rounds of interactive question-and-answer records to obtain the psychological score of the user's semantic label.

[0097] Step 5.1: Set the prompt word template , used to perform the task of extracting psychological resilience features. In this embodiment, the prompt word template Specifically:

[0098] “The following is a complete transcript of the student’s interaction with the system Please identify the explicit language features that reflect their psychological resilience, including: emotional words, resilient behavior words, self-belief expressions, positive emotional descriptions, etc., and output them in a structured manner. "Among them, For complete interactive corpus ;

[0099] Interactive Corpus Fill in the prompt word template And input into the large language model to obtain the language feature set ,in For texts involving emotional words, resilient behavior words, self-belief expressions, and positive emotion descriptions;

[0100] Specifically, in this embodiment, student answer example 1 is "Once, our class played a friendly match against another college. The score was very close. In the last two minutes, I hit a three-pointer to reverse the score and ultimately win the game for us. The whole class was cheering on the sidelines. At that moment, I was really proud and felt that my efforts were recognized." The structured output result is: [emotion words: proud, happy; tenacious behavior words: hard work, persistence in the game; self-belief expression: feeling that my efforts were recognized (implicit value affirmation); positive emotion descriptions: the whole class cheering, increased pride, and a sense of recognition];

[0101] Student answer example 2 is, "Some time ago, I drew a night scene of campus. I spent several days slowly sketching and adjusting the colors. After I finished it, I posted it on WeChat Moments and received many likes and comments. A junior even sent me a private message saying that he was very moved. At that moment, I felt very satisfied. It felt like I had expressed my emotions in my own way and resonated with others." The structured output result is: [emotion words: satisfied, moved; resilient behavior words: spent several days and slowly sketched; self-belief expression: expressed emotions, resonated (identified with the value of self-expression); positive emotion descriptions: satisfied, sense of accomplishment brought by creation, appreciated by others];

[0102] Step 5.2: Set the prompt word template , used to guide the large language model to perform scoring reasoning based on language features. In this embodiment, the prompt word template Specifically:

[0103] “The following is a record of multiple rounds of interaction between students and the system and language feature sets Based on the text, please rate their psychological resilience from the three dimensions of 'tenacity, self-reliance, and optimism' (out of 10 points) and explain your reasoning."

[0104] Step 5.3: Interaction corpus and language feature sets Fill in the prompt word template And input into the large language model, output the psychological score results of all semantic tags ,in, For the The psychological score of each semantic tag is calculated, and the psychological score result and reasoning basis are used as the final output to provide the user's psychological resilience score report. , wherein the reasoning basis is the scoring basis generated by the large language model when writing the prompt word.

[0105] Specifically, in this embodiment, the psychological score results =[9,8,7.5], scoring is based on:

[0106] (1) Resilience: The student clearly demonstrated the ability to persist in the face of difficulties (e.g., a comeback at a critical moment in a basketball game, overcoming nervousness during a speech); used resiliency words such as "hard work," "practice," and "preparation in advance"; demonstrated the student's ability to participate in social activities and seek support when faced with pressure (e.g., final exam anxiety), and possessed a certain degree of emotional regulation ability; demonstrated resiliency across multiple scenarios, demonstrated strong stress tolerance, and received a score close to full marks;

[0107] (2) Self-improvement: Demonstrates a willingness for independent growth: "I hope I can continue to challenge more difficult things"; exhibits behaviors of active involvement and self-improvement (such as actively practicing speeches and continuously creating paintings); expresses self-beliefs in a relatively rich manner, reflecting a certain sense of self-worth and intrinsic motivation; although the behaviors are clear, the self-statement lacks systematic planning or long-term goal expression, and is somewhat reserved;

[0108] (3) Optimism: Clearly express positive emotional experiences: pride, satisfaction, peace of mind, hope, etc.; can see positive meaning in failure and pressure (such as "being understood", "being recognized", "continue to challenge"); the expression is relatively mild, mainly based on feelings such as "being accepted", "relaxation", and "sense of accomplishment", without exaggerated or overly positive expressions; although the overall mood is positive, there is no obvious strong tendency to positively anticipate the future or reconstruct the meaning of adversity, so it is slightly lower than the self-reliance dimension.

[0109] Step 6: Set up a prompt word template for providing intervention suggestions. Based on the psychological scores and multiple rounds of interactive question-and-answer records, generate intervention suggestions on leisure activities, positive emotion training tasks, and thinking training activities.

[0110] Step 6.1: Set the prompt word template , used to generate intervention suggestions, in this embodiment, the prompt word template Specifically:

[0111] “Here are the students’ mental toughness scores: and analysis results Please generate a personalized suggestions, corresponding to: leisure activity recommendations, positive emotion training tasks, and thinking training paths"; among them, The number of items to be generated for each aspect;

[0112] Step 6.2: Psychological scoring results With score report Fill in the prompt word template The data is input into a large language model to generate multiple intervention suggestions, which are then returned to the user in natural language.

[0113] Specifically, in this embodiment, the generated intervention suggestions include:

[0114] (1) Try participating in regular challenging sports, such as rock climbing, endurance running, or team basketball leagues, and combine them with recording and feedback mechanisms to enhance your sense of persistence. Reason: You have excellent persistence and resilience in the face of difficulties and pressure. Effect: Such sports can continuously stimulate your "challenge-persistence-achievement" closed-loop experience, further strengthen your resilience, and provide a healthy outlet for emotional release;

[0115] (2) Complete a weekly "Positive Experience Reconstruction" writing exercise, recording a recent minor problem and trying to extract positive meaning or growth gains from it. Reason: You already have a certain ability to express positive emotions, but your ability to "reconstruct meaning" when facing setbacks can be further improved. Effect: This exercise can enhance your ability to find hope in adversity, improve your positive expectations for the future, and further enhance your optimism.

[0116] (3) Build a "Goal-Action-Feedback" micro-goal system: Set a challenging growth goal each month (such as public expression or content creation) and track your completion path and inner feelings in a structured manner. Reason: You already have self-motivation and a habit of hard work, making it suitable for the self-efficacy system building phase. Benefit: This path will help you develop a more stable self-motivation model, not only improving your motivation but also strengthening your belief that "you can continue to grow."

[0117] In summary, the present invention improves the immersion, personalization, and dynamic adaptability of psychological assessments by designing prompt word templates, dialogue generation control logic, and multi-round interaction processes for prompting large language models to generate evaluation dialogues, and is suitable for user mental health support and development service scenarios.

[0118] The above describes the specific embodiments of the present invention in detail with reference to the accompanying drawings and specific examples. However, the present invention is not limited to the above embodiments. Various changes can be made within the knowledge of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A method for measuring and providing feedback on student resilience based on a large language model, characterized by: The steps include: Step 1: Build a psychological assessment prompt word template and set several semantic tags. Use the psychological assessment prompt word template to guide the large language model to generate questions related to the semantic tags to initiate dialogue interaction with the user. Step 2: Construct a prompt word template for investigating leisure participation, guiding the large language model to generate questions to explore the frequency, type, and subjective experience of users' leisure activities; Step 3: Construct a prompt word template to investigate positive emotional arousal, prompting the large language model to generate questions related to positive emotional arousal and positive experience recall; Step 4: Based on the constructed psychological assessment prompt word template, the leisure participation survey prompt word template, and the positive emotion arousal survey prompt word template, the large language model is used to conduct multiple rounds of interactive question and answer sessions with the user; Step 5: Set up a prompt word template for performing psychological resilience feature extraction and multi-dimensional score prediction, and combine multiple rounds of interactive question-and-answer records to obtain the psychological score of the user's semantic label; Step 6: Set up a prompt word template for providing intervention suggestions. Based on the psychological scores and multiple rounds of interactive question-and-answer records, generate intervention suggestions for leisure activities, positive emotion training tasks, and thinking training activities. The Step 4 is specifically as follows: Step 4.1: The large language model starts a conversation and receives user responses in sequence based on the constructed psychological assessment prompt word template, the leisure participation survey prompt word template, and the positive emotion arousal survey prompt word template. Step 4.2: Question and answer based on the three dimensions of psychological assessment, leisure participation and positive emotional arousal. , input it into the large language model to determine whether the current dimension has been expressed. If not, call the prompt template corresponding to the current dimension to generate the next round of questions. If it has been expressed, jump to the next dimension; Step 4.3: Repeat Step 4.2 to form a multi-round dialogue sequence with dynamic control capabilities until all dimensions are completed or the dialogue termination conditions are met, and the interaction corpus is output. ; The Step 5 is specifically as follows: Step 5.1: Set the prompt word template , used to perform the task of extracting psychological resilience features, and to convert the interactive corpus Fill in the prompt word template And input into the large language model to obtain the language feature set ; Step 5.2: Set the prompt word template , used to guide the large language model to perform scoring reasoning based on language features; Step 5.3: Interaction corpus and language feature sets Fill in the prompt word template And input into the large language model, output the psychological score results of all semantic tags ,in, For the The psychological score of each semantic tag is calculated, and the psychological score result and reasoning basis are used as the final output to provide the user's psychological resilience score report. , wherein the reasoning basis is the scoring basis generated by the large language model when writing the prompt word.

2. The method for measuring and providing feedback on student resilience based on a large language model according to claim 1, characterized in that: The Step 1 is specifically as follows: Step 1.1: Set semantic labels, which are composed of multiple psychological dimensions; Step 1.2: Set the prompt word template , used to prompt the large language model to generate open-ended psychometric questions for each semantic tag; Step 1.3: Set the prompt word template Combined with semantic tags to form prompts and input into the large language model, a set of dialogue questions is generated and bound to the semantic tags, denoted as ,in, is a set of all generated questions, including specific questions and semantic labels of questions, For the A generated problem, is a semantic label; Step 1.4: Randomly select each semantic label from several semantic labels. Questions, dialogue with users, user feedback results are recorded as ,in, is the total number of semantic tags, This will be determined based on the duration of the proposed test.

3. The method for measuring and providing feedback on student resilience based on a large language model according to claim 1, characterized in that: The Step 2 is specifically as follows: Step 2.1: Set the prompt word template , used for the first interaction between the large language model and the user; Step 2.2: Set the prompt word template , used for prompts after the user answers; Step 2.3: Use the prompt word template Generate questions and conduct the first conversation with the user, using prompt word templates based on the user's answers When the large language model completes all the active inquiries, the conversation ends and the user's answer in the first conversation is recorded as .

4. The method for measuring and providing feedback on student resilience based on a large language model according to claim 1, characterized in that: The Step 3 is specifically as follows: Step 3.1: Set the prompt word template , used to guide users to recall positive experiences for the first time; Step 3.2: Set the prompt word template , used to dynamically supplement questions during interaction; Step 3.3: Use the prompt word template Generate the first round of questions to obtain the user's initial expression of positive emotional experience; Step 3.4: Use prompt word templates based on the user's answers in each round Generate supplementary guiding questions and advance the conversation cyclically until the user has expressed the source, process and characteristics of his positive emotional experience. The user's answer in this conversation is recorded as .

5. The method for measuring and providing feedback on student resilience based on a large language model according to claim 1, characterized in that: The Step 6 is specifically as follows: Step 6.1: Set the prompt word template , used to generate intervention recommendations; Step 6.2: Psychological scoring results With score report Fill in the prompt word template The data is input into a large language model to generate multiple intervention suggestions, which are then returned to the user in natural language.

Images