Psychological evaluation method and device based on large model, electronic equipment and storage medium

Abstract

This specification provides a psychological assessment method, device, electronic device, and storage medium based on a large model. The method includes: obtaining a user's identity; displaying candidate scenarios corresponding to the user's identity; responding to a target scenario selected by the user from the candidate scenarios; using a large model to call a psychological assessment question bank corresponding to the target scenario to generate dialogue questions; obtaining the user's answer data to the dialogue questions; determining a first information dimension not covered by the answer data in the key information dimensions required for scoring the dialogue questions; determining the information missing degree based on the first information dimension and the key information dimensions; if the information missing degree meets the follow-up question condition, generating follow-up questions using the large model and obtaining supplementary answer data from the user to the follow-up questions; generating scores corresponding to each dialogue question based on the answer data and supplementary answer data; and generating the user's psychological assessment result based on the scores. This improves assessment accuracy and enhances user experience and willingness to take assessments.

Description

Technical Field

[0001] This specification relates to the field of psychological assessment technology, and in particular to a psychological assessment method, device, electronic device and storage medium based on a large model. Background Technology

[0002] In the field of psychological assessment, traditional paper-based quality questionnaires (such as GHQ-12, DASS-21, and CD-RISC) are widely used standardized assessment tools. Although they have high reliability and validity, the process of filling them out is tedious, users are wary, and they are prone to missing or misanswering questions, leading to biased results.

[0003] Specifically, traditional questionnaires, in order to cover the broadest possible population and scenarios, are often highly generalized and abstract. For example, the question, "Have you lost interest in activities you used to enjoy?" requires respondents to accurately understand the specific content of "activities you used to enjoy" and recall their recent state. This process involves a certain comprehension cost, and respondents with different educational backgrounds and life experiences may have different understandings of the same question, leading to biased assessment results.

[0004] Furthermore, traditional scales often present closed-ended answers, such as "never," "occasionally," "more than half the time," and "almost always." Users not only need to recall complex life scenarios and emotional experiences over a period of time, but also need to force these continuous, vague feelings onto fixed options. In this "recall-mapping" process, insufficient thinking time, fuzzy memory, or misunderstanding can lead to information loss and decreased accuracy. Users' subjective feelings are multi-dimensional, while answer options are flat; this dimensionality reduction introduces certain assessment errors. Summary of the Invention

[0005] The main purpose of this specification is to provide a psychological assessment method, device, electronic device, and storage medium based on a large model, aiming to improve assessment accuracy and enhance user experience and willingness to take assessments. The technical solution is as follows: Firstly, the embodiments of this specification provide a psychological assessment method based on a large model, including: Obtain the user's identity and display the candidate scenarios corresponding to the user's identity; In response to the target scenario selected by the user from the candidate scenarios, a large model is used to call the psychological assessment question bank corresponding to the target scenario to generate dialogue questions; Obtain the user's response data to the dialogue question; Identify the first information dimension that is not covered in the key information dimensions required for scoring the dialogue question from the answer data; Based on the first information dimension and the key information dimension, the degree of information missing is determined; If the information missing degree meets the follow-up question condition, then the large model is used to generate follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained; Based on the answer data and the supplementary answer data, a score is generated for each of the dialogue questions; The psychological assessment results for the user are generated based on the rating.

[0006] Secondly, embodiments of this specification provide a psychological assessment device, comprising: The scene determination unit is used to obtain the user's identity and display the candidate scene corresponding to the user's identity. The question generation unit is used to generate dialogue questions by calling the psychological assessment question bank corresponding to the target scenario from the candidate scenarios in response to the user's selection of the target scenario using a large model; The answer acquisition unit is used to acquire the answer data input by the user in response to the dialogue question; An information dimension determination unit is used to determine a first information dimension that is not covered in the key information dimensions required for scoring the dialogue question in the answer data. An information missing assessment unit is used to determine the degree of information missing based on the first information dimension and the key information dimension; The follow-up questioning unit is used to generate follow-up questions using the large model if the information missing degree meets the follow-up questioning conditions, and to obtain supplementary answer data from the user for the follow-up question. A scoring unit is used to generate a score for each of the dialogue questions based on the answer data and the supplementary answer data; An assessment unit is used to generate a psychological assessment result for the user based on the score.

[0007] Thirdly, embodiments of this specification provide an electronic device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described above.

[0008] Fourthly, embodiments of this specification provide a storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described above.

[0009] Fifthly, embodiments of this specification provide a computer program product, including: a computer program that, when executed by a processor of an electronic device, enables the processor to at least implement the method described in the first aspect.

[0010] In the embodiments of this specification, by providing corresponding scenarios based on user identity and dynamically acquiring a psychological assessment question bank according to the target scenario selected by the user, a large model generates psychological assessment questions based on the question bank corresponding to the user scenario, better meeting the needs of different users. Through the natural language capabilities of the large model, the process is transformed into an empathetic and interactive "chat," reducing user resistance to psychological screening and improving completion rates. During the user's question-answering process, the large model performs deep semantic understanding of the user's natural language responses. The large model determines whether the user's answers meet the scoring requirements by assessing the degree of information gaps, and adaptively asks follow-up questions based on different levels of information gaps, guiding the user to answer questions clearly and improving assessment accuracy. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is an example diagram illustrating a psychological assessment method based on a large model provided in the embodiments of this specification; Figure 2 This is a flowchart illustrating a psychological assessment method based on a large model, as provided in the embodiments of this specification. Figure 3 This is a flowchart illustrating a psychological assessment method based on a large model, as provided in the embodiments of this specification. Figure 4 This is a flowchart illustrating a psychological assessment method based on a large model, as provided in the embodiments of this specification. Figure 5 This is a schematic diagram illustrating the results of a psychological assessment method based on a large model, as provided in the embodiments of this specification. Figure 6 This is a flowchart illustrating a psychological assessment method based on a large model, as provided in the embodiments of this specification. Figure 7 This is a schematic diagram of the structure of a psychological assessment device based on a large model, as provided in the embodiments of this specification. Figure 8 This is a schematic diagram of the structure of an electronic device provided in the embodiments of this specification. Detailed Implementation

[0013] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.

[0014] In related technologies, psychological assessments typically involve converting paper-based psychological scales into electronic questionnaires on web pages or within apps. Users complete the questions by clicking on options, and the system automatically calculates scores. Traditional paper-based psychological assessment methods have the following problems: the scales use closed-ended options (such as "never," "occasionally," "almost always," etc.), requiring the assessee to forcibly map a continuous, vague, multi-dimensional range of subjective emotions and life experiences over a period of time onto fixed, flat options. This "recall-matching" transformation process itself is a potential source of assessment error. Furthermore, answering requires the assessee to rely on memory recall and understanding of the options, which can easily lead to information loss due to insufficient thinking time, memory lapses, and biased understanding of the semantics of the options, thus reducing the accuracy of the assessment results.

[0015] To address the aforementioned issues, this specification provides a psychological assessment method based on a large model, which can be found in the embodiments described herein. Figure 1 This diagram illustrates an example of a large-model-based psychological assessment method provided in this specification. Candidate scenarios are provided based on the user's identity, and corresponding dialogue questions are generated according to the user's selected target scenario. Contextualized questioning and natural language responses replace abstract questions and fixed options, completely resolving the problems of comprehension bias and recall mapping bias. Users no longer need to perform "translation" and "matching" work; they only need to express themselves authentically, reducing cognitive load and improving user experience and willingness. Figure 1 As shown, after the user selects scenario 4, the large model retrieves the psychological assessment question bank corresponding to scenario 4 and rewrites it into dialogue questions. Figure 1 (A) The system initiates questions to users via chat. By analyzing the information dimensions contained in the user's answer data, the system obtains the information missing degree of the answer data. The large model adaptively generates follow-up questions based on the information missing degree of the user's answer to fill in the missing information and ensure that a score can be given based on the user's answer, thus guaranteeing the accuracy of the evaluation.

[0016] The following detailed description of the large-model-based psychological assessment method provided in this specification, with reference to specific embodiments, provides a detailed explanation.

[0017] Please see Figure 2 This document provides a flowchart illustrating a psychological assessment method based on a large model, as illustrated in the embodiments of this specification. Figure 2As shown, the method in the embodiments of this specification may include the following steps S101-S108.

[0018] S101, Obtain the user's identity and display the candidate scenarios corresponding to the user's identity; In one embodiment, when a user begins using the system, the system provides candidate scenarios based on the user's chosen identity (e.g., student, working professional, stay-at-home mom). Based on the assessment dimensions of a psychological scale, and combined with a pre-defined set of scenarios that trigger high-frequency psychological states for different user identities, each candidate scenario corresponds to an assessment direction for a specific psychological need. For example, for a student, candidate scenarios might include interpersonal relationships, life planning, learning scenarios, self-awareness, and role identification.

[0019] It should be noted that user identities can be categorized in various ways, including but not limited to age groups (such as teenagers, young adults, and seniors), occupation types (such as students, working professionals, and freelancers), and lifestyle tags (such as test-takers, new parents, and high-pressure workers). The specific method can be chosen based on actual needs.

[0020] S102, in response to the target scenario selected by the user from the candidate scenarios, a large model is used to call the psychological assessment question bank corresponding to the target scenario to generate dialogue questions; In one embodiment, the target scenario is a specific scenario that the user chooses from a list of candidate scenarios and that matches their current psychological state and needs. The psychological assessment question bank refers to a set of questions formed by decomposing and reconstructing the assessment dimensions of traditional standardized psychological scales (such as GHQ-12, DASS-21, and CD-RISC) according to different candidate scenarios.

[0021] In one feasible implementation, once a user selects a target scenario, the system immediately triggers the scenario-question bank association mechanism, retrieving the psychological assessment question bank corresponding to the target scenario. Subsequently, the questions from the question bank are input into a pre-trained large model. During the input process, a scenario-based generation instruction needs to be passed to the large model. The instruction includes: "Based on the current [target scenario name], transform the input assessment questions into conversational, scenario-based chat-style questions, avoiding the use of technical jargon. The questions should guide the user to express themselves in natural language, without setting fixed options." After receiving the instruction and the question bank questions, the large model generates dialogue questions through semantic understanding and scenario adaptation reconstruction.

[0022] In another feasible implementation, different scenarios can be pre-set with questions based on a standard scale. When a user selects a target scenario, the model first asks questions based on a pre-set psychological assessment question bank corresponding to that scenario. As the user answers the questions, the model makes real-time judgments and retrieves more suitable questions for the user in subsequent questions to achieve a more accurate assessment of the user's current psychological problems.

[0023] S103, Obtain the user's answer data in response to the dialogue question; In one embodiment, the user inputs feedback in natural language to the dialogue questions generated by the large model. The feedback can be in the form of text descriptions, speech-to-text content, etc., which is different from the fixed answer options of traditional scales and has the characteristics of open content and rich semantics.

[0024] For example, input boxes matching the dialogue questions can be set in the front-end interactive interface for users to input their answer data.

[0025] S104, determine the first information dimension that is not covered in the key information dimensions required for scoring the dialogue question; In one embodiment, the key information dimension refers to the core psychological assessment indicator dimension corresponding to each dialogue question. It is derived from the assessment dimensions of the original psychological scale, and each dialogue question corresponds to at least one key information dimension. The key information dimensions can be manually decomposed by experts in advance. The more key information dimensions the user expresses, the higher the accuracy of the assessment. For example, the key information dimensions corresponding to the dialogue question "When you encounter difficulties at work recently, how do you usually handle them?" are "stress coping style dimension" and "emotion regulation ability dimension".

[0026] The user's answer data is input into the semantic analysis module of the big model. The answer data is semantically parsed to extract the information dimensions involved in the answer and compared with the preset key information dimensions (set). If a certain key information dimension is not mentioned at all in the answer data, the big model will determine that dimension as the first information dimension.

[0027] S105, determine the degree of information missing based on the first information dimension and the key information dimension; In one embodiment, the information missing degree of the dialogue question is evaluated based on the missing first information dimension and the required key information dimension. The magnitude of the information missing degree directly reflects whether the answer data meets the scoring requirements.

[0028] Optionally, the degree of information missing can be determined based on the number of first information dimensions and the total number of key information dimensions. Furthermore, different weights can be assigned to the importance of the key information dimensions.

[0029] S106, if the information missing degree meets the follow-up question condition, then the large model is used to generate follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained; In one embodiment, the follow-up question condition refers to a preset threshold standard for determining whether a follow-up question should be initiated. It is usually presented in the form of a numerical range of information missing degree. For example, when the information missing degree is ≥50%, the follow-up question mechanism is triggered. If the calculated information missing degree meets the follow-up question condition, the system generates a natural language question based on the large model for the first information dimension, which is used to guide the user to supplement key information. That is, a follow-up question. Its core objective is to fill in the missing key information dimensions in the answer data.

[0030] For example, the first information dimension and the original dialogue question are input into the large model, and a follow-up question generation instruction is issued. The instruction includes "For the missing information of [first information dimension name], combine the original question [original dialogue question content] to generate a natural language follow-up question. The question should be concise and clear to guide the user to supplement specific information."

[0031] S107, Based on the answer data and the supplementary answer data, generate a score corresponding to each of the dialogue questions; In one embodiment, the user's original and supplementary answer data are integrated to form a complete semantic text. Then, a large model is invoked to parse the complete semantic text and load a scoring standard library corresponding to the target scenario. This scoring standard library is constructed based on the scoring rules of traditional psychological scales. For example, a Likert scale can be used, which is essentially a "level quantification of attitude / state," such as a 4-point scale: 0 = completely disagree, 1 = somewhat disagree, 2 = neutral, 3 = agree; or the DASS-21 0-3 scoring rule can be used, converting it into a score range corresponding to the natural language description, such as "no relevant feelings at all" corresponding to 0 points, "occasionally occurring and short-lived" corresponding to 1 point, "frequently occurring and affecting daily life" corresponding to 2 points, and "frequently occurring and unable to be relieved on its own" corresponding to 3 points. The large model matches the integrated answer text with the score ranges in the scoring standard library through semantic parsing. For example, if a user answers, "This dull feeling occurs about two or three times a week, each time lasting one or two days, sometimes affecting my mood in class," the corresponding score obtained by matching with the scoring standard library is 2 points. Finally, the system stores the scoring results for each dialogue question in the evaluation results database, forming a detailed scoring table.

[0032] In one feasible implementation, the large model is used to analyze the severity, frequency, and impact of the content described by the user in the response data and the supplementary response data; the severity, frequency, and impact are semantically matched with the scoring criteria of the psychological assessment question bank to obtain the score corresponding to each dialogue question. Specifically, this refers to using a labeled psychological assessment corpus to fine-tune the general large model, enabling it to identify the three core scoring dimensions of "severity, frequency, and impact." Frequency dimension analysis extracts the frequency of occurrence of the user's psychological state by identifying time-related keywords and expressions in the text, such as "three or four days a week," "occasionally on weekends," and "feeling uncomfortable every morning," and matches them to the frequency feature range of the scoring criteria. Severity dimension analysis determines the severity of the psychological state by identifying words describing the intensity of emotions in the text, such as "slightly irritable," "unbearable depression," and "feeling depressed but able to self-regulate," and matches them to the severity features of the scoring criteria. Impact dimension analysis identifies the impact of psychological states described in the text on life, work, and social interactions, such as expressions like "does not affect normal life," "has difficulty concentrating at work," and "does not want to meet friends," and maps them to the impact feature range of the scoring criteria. For example, scores can be obtained separately for the three dimensions of severity, frequency, and impact, and then summed to obtain the score for the dialogue question.

[0033] For example, the system asks: Do you experience insomnia because you are worried about your exam results, not being able to finish your homework, or falling behind in your studies? User's answer: I do find math quite difficult, and I often can't keep up. (This answer cannot conclude whether I have insomnia.) The system followed up: University math is indeed quite difficult, and I completely understand how you feel. When you fall behind, do you experience insomnia? User's answer: It happens occasionally, maybe once or twice a week.

[0034] The system determines: Based on the user's answer being mapped to the scoring table as "occasionally appearing," this question is awarded 1 point. It should be noted that, due to the involvement of psychological assessment scenarios, the questions and follow-up questions in the large model need to meet the following constraints: avoid sensitive follow-up questions, such as asking "Have you thought about how to do it specifically?" to users who have self-harm or suicidal tendencies; instead, prioritize triggering the crisis intervention process; respect user boundaries; if a user clearly states "I don't want to talk about this," the algorithm should immediately stop asking follow-up questions in that dimension and switch to other questions or empathetic reassurance; adapt to different user groups, such as asking follow-up questions to student users in a gentler and more specific way, while asking follow-up questions to working professionals can be more direct.

[0035] S108, Generate the user's psychological assessment result based on the rating.

[0036] In one embodiment, the user's psychological assessment result is determined based on the scores of each dialogue question. The psychological assessment result may include a total score, sub-dimensional scores for each dimension, the user's current psychological state characteristics, strengths in psychological resources, and potential psychological distress. For example, the psychological assessment result is displayed on the front-end interface in the form of a visual report. It is understood that the psychological assessment result can be displayed to the user in any format, such as images, text descriptions, tables, audio, etc., depending on the specific needs.

[0037] In the embodiments of this specification, by obtaining the user's identity, candidate scenarios corresponding to the user's identity are displayed. In response to the target scenario selected by the user from the candidate scenarios, a large model is used to call the psychological assessment question bank corresponding to the target scenario to generate dialogue questions. The user's answer data for the dialogue questions is obtained, and the first information dimension not covered in the key information dimensions required for scoring the dialogue questions is determined. Based on the first information dimension and the key information dimension, the information missing degree is determined. If the information missing degree meets the follow-up question conditions, the large model generates follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained. Based on the answer data and supplementary answer data, a score corresponding to each dialogue question is generated, and the user's psychological assessment result is generated based on the score. Generating psychological assessment questions based on the psychological assessment question bank corresponding to the user's scenario using a large model better meets the needs of different users. Simultaneously, the natural language capabilities of the large model transform it into an empathetic and interactive "chat" process, reducing user resistance to psychological screening and improving the completion rate. Combining information missing degree assessment to determine whether the user's answer meets the scoring requirements, and adaptively asking follow-up questions according to different information missing degrees, guides users to answer questions clearly, improving the accuracy of the assessment.

[0038] Please see Figure 3 This document provides a flowchart illustrating a psychological assessment method based on a large model, as illustrated in the embodiments of this specification. Figure 3 As shown, the method in the embodiments of this specification may include the following steps S201-S209.

[0039] S201, The large model is used to perform semantic analysis on the response data to obtain the target issues contained in the response data; In one embodiment, different scenarios include multiple topics, each topic corresponding to one or more questions. The deep semantic understanding capabilities of the large model are used to determine the target topic related to the user's answer.

[0040] S202, Obtain the key information dimensions corresponding to the target issue; In one embodiment, for each target issue, a psychology expert, combining traditional psychological scale assessment indicators and clinical counseling experience, pre-determines the key information dimensions required for a complete assessment of the issue. For example, the key information dimensions corresponding to "conflicts in intimate relationships" include: the conflict-triggered event, the frequency of conflict, the emotional impact of the conflict, and the conflict-handling methods. A database can be stored to store the mapping relationship between all target issues and their corresponding key information dimensions, and the corresponding key information dimensions can be retrieved from the database after the target issue is determined.

[0041] S203, identify the second information dimension contained in the answer data, and compare it with the key information dimension to obtain the first information dimension not covered in the answer data; In one embodiment, a second information dimension is obtained by extracting information directions already covered by user statements from user response data through large-scale model semantic analysis. The portion of the key information dimensions corresponding to the target topic not covered by the second information dimension is the first information dimension. Specifically, the system compares the extracted second information dimension with the obtained list of key information dimensions for the target topic one by one; dimensions that do not match are the first information dimensions.

[0042] The information missing degree is calculated as follows: Information missing degree = Number of first information dimensions / Total number of key information dimensions × 100%.

[0043] S204, if the information missing degree meets the follow-up question condition, then select the initial follow-up questions under the target topic from the follow-up question pool; In one embodiment, a pre-built pool of follow-up questions categorized by target issue is stored. When the degree of information missing meets the follow-up question criteria, initial follow-up questions for that target issue are selected from the pool. For example, if the first information dimension is "frequency of conflict," then follow-up questions labeled "corresponding dimension: frequency of conflict" are selected. By using the target issue as the core, exclusive key information dimensions and follow-up question strategies are matched for different users, avoiding the homogenization defect of all users sharing the same set of questions.

[0044] S205, rewrite the target follow-up question using the large model to obtain an empathic expression of the initial follow-up question, output the empathic expression of the initial follow-up question, and obtain the user's supplementary answer data for the follow-up question; In one embodiment, the initial follow-up questions in the follow-up question pool are questions in a general format. A large model is used to transform the initial follow-up questions into natural language expressions that fit the user's emotional state, weaken the direct questioning attribute, and enhance emotional resonance. The core is to improve the user's willingness to answer while retaining the core intent of the follow-up question.

[0045] The instruction to rewrite the question with empathy was issued. The instruction was: "Please rewrite the following follow-up questions in an empathetic manner, taking into account the user's emotional state (extracted from historical answers). You are required to first express your understanding of the user's emotions before asking the follow-up questions, and avoid asking questions in a stiff manner. For example, rewrite 'What was the specific reason for the argument?' as 'You sound like you are feeling quite upset right now. Would you mind telling me what the main reason for this argument was?'"

[0046] Specifically, in one embodiment of this specification, determining whether the degree of information missing meets the follow-up questioning condition may include: S2041, if the information missing degree is greater than the first threshold, then the large model is used to obtain follow-up questions with a first gain value, and the user's supplementary answer data for the follow-up questions with the first gain value is obtained. In one embodiment, the system retrieves preset threshold parameters (e.g., a first threshold of 70% and a second threshold of 30%), compares the information missing degree with each threshold parameter, determines whether the current follow-up questioning conditions are met, and determines the follow-up questions based on the different follow-up questioning conditions met. The follow-up questioning mode can be flexibly switched according to the degree of information missing: for high missing degree, core information is prioritized; for medium missing degree, details and empathy are considered; and for low missing degree, scoring is initiated directly, making the evaluation process more flexible.

[0047] When the information gap exceeds the first threshold, a large model is used to obtain follow-up questions with the highest gain value, which are then provided to the user. The first gain value refers to the information gain range corresponding to high-gain follow-up questions, typically between 0.8 and 1.0. These questions directly hit the core content of the first information dimension, resulting in the highest information retrieval efficiency.

[0048] For example, when the information missing rate is greater than 70%, the system selects questions from the follow-up question pool that are under the target topic, correspond to the first information dimension, and have a gain value of 0.8-1.0. For example, if the first information dimension is "conflict triggering event", the system selects questions with a gain value of 1.0 such as "What specific thing caused this argument?", which are rewritten into empathetic expressions by the large model and output, prioritizing the supplementation of core information.

[0049] S2042, if the information missing degree is less than or equal to the first threshold and greater than the second threshold, then the large model is used to obtain follow-up questions with a second gain value, and the user's supplementary answer data for the follow-up questions with the second gain value is obtained; In one embodiment, if the information missing degree is determined to be less than or equal to a first threshold and greater than a second threshold, a large model is used to obtain follow-up questions with a second gain value. Here, the second threshold is less than the first threshold, and the second gain value is less than or equal to the first gain value. The second gain value refers to the information gain value range corresponding to a medium-gain follow-up question, typically between 0.4 and 0.7.

[0050] For example, when the information gap is 30% < 70%, the system filters questions with a gain value of 0.4-0.7. These questions can be generated in real time by a large model combined with the user's emotions, such as "After an argument, have you tried to communicate with your friends to resolve it?", which aims to supplement details and enhance emotional resonance.

[0051] The system can adjust the threshold parameters according to the accuracy requirements of different evaluation scenarios.

[0052] In addition, when initially building the pool of follow-up questions, gain values ​​can be manually labeled.

[0053] For example, follow-up questions that directly hit the key dimension of gain are marked with a high score (0.8-1.0), such as "What specific things caused this breakup?"; follow-up questions that are indirectly related are marked with a medium score (0.4-0.7), such as "After the breakup, how has your daily life changed compared to before?"; follow-up questions that are irrelevant or of low value are marked with a low score (0-0.3), such as "What do you like to eat to relieve your mood?".

[0054] These marked follow-up questions are stored in the database according to scenario categories, thus forming the initial follow-up question pool.

[0055] S2043, if the information missing degree is less than the second threshold, then the follow-up questioning condition is not met.

[0056] In one embodiment, if the information missing degree is less than the second threshold, it indicates that the information coverage is sufficient, and the process proceeds directly to the scoring stage.

[0057] S206, determine the third information dimension that is not covered in the key information dimensions required for scoring the dialogue question, based on the supplementary answer data; In one embodiment, after generating follow-up questions, the system can determine whether further follow-up questions are needed based on the information dimension coverage of the user's supplementary answer data. The user's supplementary answer data is preprocessed, input into the large model, and the information dimensions covered by the supplementary answers are extracted and denoted as the third coverage dimension. The system merges the second and third coverage dimensions to obtain the total coverage dimension, and then performs a full comparison with the key information dimension list of the target topic. Dimensions that do not match are the third information dimensions.

[0058] S207, Based on the third information dimension and the key information dimension, determine the degree of information missing after follow-up questioning; In one embodiment, after one round of follow-up questioning, the proportion of key information dimensions not covered by the user's answer to the total number of key information dimensions is determined, i.e., the information gap after follow-up questioning, to determine whether further follow-up questioning is necessary. For example, the information gap after follow-up questioning is calculated based on the number of third information dimensions and the total number of key information dimensions.

[0059] S208, if the information missing degree after the follow-up question meets the follow-up question condition, then proceed to the step of generating follow-up questions using the large model, obtaining the user's supplementary answer data for the follow-up questions, and recording the number of follow-up question rounds generated for the user. In one embodiment, the system compares the calculated information gap after follow-up questions with a preset threshold to verify whether the current information coverage meets the scoring requirements. If the follow-up questioning conditions are still met (e.g., >30%), the next round of follow-up questions is triggered. The system records the current number of follow-up questioning rounds in the background. If the number of rounds has not reached the preset round threshold (e.g., 3 rounds), the process proceeds to steps S204-S207, where the large model generates new follow-up questions. If the number of rounds has reached the threshold, adaptive follow-up questioning stops, and step S209 is executed.

[0060] Optionally, during multiple rounds of follow-up questions, the large model will optimize the empathy of the follow-up questions based on the user's historical answers and supplementary answers. For example, the second round of follow-up questions can be more tailored to the user's specific situation and avoid asking the same questions repeatedly.

[0061] S209, if the number of follow-up questions reaches the number of rounds threshold, then the scale questions in the psychological assessment question bank are called so that the user can select from the preset options of the scale questions to obtain supplementary answer data.

[0062] In one embodiment, when the number of follow-up questions recorded by the system reaches a preset threshold (e.g., 3 rounds), and the information gap after follow-up questions still does not meet the scoring requirements, a scale question invocation mechanism is triggered. The scale questions refer to closed-ended questions extracted from traditional standardized psychological scales that correspond to the third information dimension, and have fixed options (e.g., "never at all," "occasionally," "more than half the time," "almost every day"), used to supplement missing information when adaptive follow-up questions are ineffective.

[0063] Optionally, the method further includes: recording the effective response rate of the user to each of the follow-up questions; and adjusting the gain value of the follow-up questions based on the effective response rate.

[0064] Specifically, the effective response rate refers to the proportion of users who provide effective answers (i.e., answers that cover the corresponding information dimensions) to a follow-up question. The effective response rate for each follow-up question is determined by recording the total number of times the question is submitted, the number of times users provide effective answers, and the number of times users avoid answering (e.g., "I don't want to say" or "I don't know"). The system then uses preset gain value adjustment rules to dynamically update the information gain value of the follow-up question based on the effective response rate. For example, if the effective response rate is ≥80%, the gain value is increased by 0.1 (maximum not exceeding 1.0), and its ranking priority in the follow-up question pool is improved; if 40% ≤ effective response rate < 80%, the gain value remains unchanged; if the effective response rate is <40%, the gain value is decreased by 0.1 (minimum not lower than 0.0); if the effective response rate is <30% for three consecutive times, the question is removed from the follow-up question pool.

[0065] Optionally, in addition to adjusting the gain value, the system will also collect effective follow-up questions added by consultants in actual conversations, add them to the corresponding topic categories after experts mark the gain value; at the same time, invalid questions with low response rates and low gain values ​​will be cleaned up regularly to achieve dynamic optimization of the follow-up question pool.

[0066] In this embodiment, a first information dimension is determined by semantic analysis of the response data. Follow-up questions are then obtained based on the topic to which the response data belongs to, supplementing the information. This improves the accuracy and completeness of the psychological assessment data and reduces assessment errors. Furthermore, by setting a threshold for the number of follow-up questions, both perfunctory answers due to excessive questioning and scoring bias caused by insufficient information are avoided. Unlike the mechanical questioning of traditional scales and the rigid follow-up questions of electronic questionnaires, this embodiment rewrites the initial follow-up questions into empathetic expressions using a large model. It first perceives and accepts the user's emotions before initiating targeted questions, weakening the sense of assessment and strengthening the sense of communication, effectively reducing user psychological resistance.

[0067] Please see Figure 4 This document provides a flowchart illustrating a psychological assessment method based on a large model, as illustrated in the embodiments of this specification. Figure 4 As shown, the method described in the embodiments of this specification may include the following steps S301-S304.

[0068] S301, using the large model based on the answer data and the supplementary answer data, generate a score corresponding to each of the dialogue questions; In one embodiment, the assessment types of dialogue questions include emotional distress assessment and psychological resilience assessment. The emotional distress assessment primarily uses a rating to evaluate the severity of a user's psychological distress. The psychological resilience assessment primarily uses a rating to evaluate a user's psychological resilience in coping with difficulties. The system first labels all dialogue questions in the target scenario to distinguish between the emotional distress assessment type and the psychological resilience assessment type. Questions can be posed to the user in a certain order; for example, first asking dialogue questions of the emotional distress assessment type, then asking dialogue questions of the psychological resilience assessment type.

[0069] In one feasible implementation, the integrated user response data and supplementary response data are categorized by question type and input into the large model. A scoring instruction is then issued: "Based on the contextual semantics, determine the degree of [emotional distress / psychological resilience] corresponding to the user's response and match it with a Likert score of 0-3." The large model, based on its pre-trained semantic understanding capabilities, extracts core information related to "severity, frequency, and impact" from the responses to achieve accurate mapping. For example, if a user answers "I feel a little anxious before exams, but it doesn't affect my studying and sleep," the large model outputs a score of 0 for emotional distress questions; for psychological resilience questions, if the user adds "I relieve anxiety by running," the model outputs a score of 3.

[0070] After each dialogue question is scored, the system outputs the score result in real time, providing a data basis for subsequent total score calculation and calibration.

[0071] Optionally, in one embodiment, after obtaining the score corresponding to the dialogue question, step S401 may be included: S401, the large model is used to calibrate the score based on the user's identity, the stage to which the user belongs under the user identity, and preset calibration rules, so as to obtain the calibrated score.

[0072] Specifically, the system retrieves a pre-defined scenario calibration rule library. This library stores calibration coefficients indexed by "user identity + stage," with corresponding rules for scenarios such as "college student - before final exams" and "middle school student - early school term." The large model matches the base score with the scenario calibration rules, correcting scores that do not conform to scenario cognition and generating a first score. For example, a college student facing emotional distress before final exams might have a base score of 2, but if the user's answer does not mention severe symptoms such as "insomnia / school aversion," the large model calibrates it to 1 according to the rules. Similarly, a middle school student at the beginning of the school term might have a base score of 1 for psychological resilience, but if the user mentions "trying to actively get to know new classmates," the large model calibrates it to 2 according to the rules.

[0073] Optionally, in one embodiment, the score is calibrated using the large model based on the user's identity, the stage the user belongs to under that identity, and preset calibration rules to obtain a calibrated score, including: S4011, The large model is used to calibrate the score based on the user's identity, the stage to which the user belongs under the user identity, and the preset calibration rules to obtain the calibrated first score; In one embodiment, considering that the scores directly output by the large model may be biased (e.g., the score for "pre-exam anxiety" may be too strict), it is necessary to calibrate them using scenario calibration rules to ensure that the scores conform to professional standards in psychological counseling. The system retrieves a preset scenario calibration rule library, which stores calibration coefficients indexed by "user identity + stage". For example, there are corresponding rules for "college students - before final exams" and "middle school students - early school term". The large model matches the basic score with the scenario calibration rules, corrects scores that do not conform to scenario cognition, and generates a first score.

[0074] S4012, compare the first score with the historical scores of historical users who have the same user identity and belong to the same stage to obtain the score deviation value; In one embodiment, the system retrieves a historical rating database for the same group of people at the same stage, and obtains the average rating of historical users that is consistent with the current user's identity (e.g., a college student) and stage (e.g., before an exam). Then, the difference between the first rating and the historical average rating is calculated. This difference is the rating deviation value, and the formula for calculating the rating deviation value is that the rating deviation value is equal to the absolute value of the difference between the first rating and the historical average rating.

[0075] S4013, if the scoring deviation value exceeds a preset threshold, the large model is used to refer to the historical scores of the same user identity and the same stage to re-score the dialogue question, and a calibrated second score is obtained.

[0076] Specifically, the system presets a deviation threshold, typically set to 1 point. If the deviation value does not exceed the threshold, the first score is the final calibrated score; if the deviation value exceeds the threshold, a second inference reassessment is triggered. For cases where the deviation exceeds the threshold, the system adds a prompt to the large model: "Please refer to the general state of people with the same identity and stage, and reassess based on the user's answer context." The large model completes the reassessment based on the supplementary instruction and outputs the calibrated second score as the final score for the question.

[0077] S302, determine the total emotional distress score corresponding to the dialogue question of the emotional distress assessment type and the total psychological resilience score corresponding to the dialogue question of the psychological resilience assessment type based on the scoring; In one embodiment, the final calibration scores of all dialogue questions are retrieved from the database, classified according to question type, and the total emotional distress score corresponding to the emotional distress assessment type dialogue questions and the total psychological resilience score corresponding to the psychological resilience assessment type dialogue questions are calculated.

[0078] S303, obtain the emotional distress score threshold corresponding to the total emotional distress score, and the psychological resilience score threshold corresponding to the total psychological resilience score; In one embodiment, the emotional distress score threshold (denoted as θ_X) refers to the critical value that distinguishes whether a user's distress level exceeds the normal range. It is determined by the statistical distribution of historical scores from the same group at the same stage, and the default value can be set to 12. The psychological resilience score threshold (θ_Y) refers to the critical value that distinguishes whether a user's resilience level is below the normal range. It is determined by the statistical distribution of historical scores from the same group at the same stage, and the default value can be set to 25. The emotional distress score threshold and the psychological resilience score threshold can be default values ​​or retrieved from historical score data, and the specific selection should be based on the actual situation.

[0079] Optionally, in one embodiment, before obtaining the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score, the method further includes steps S3031-S3032: S3031, Obtain the historical ratings of historical users who have the same user identity and belong to the same stage as the user; Specifically, based on the current user's identity (e.g., university student, high school student) and stage of life (e.g., before exams, early school term, year-end assessment period), the system retrieves the total emotional distress score and total psychological resilience score sets of users with the same identity and stage from the historical scoring database. "Historical users with the same identity and stage" refers to the group of historically assessed users who share the same user identity tags as the current user and are in the same stage of life / study / work.

[0080] S3032, Based on the historical scores, determine the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score.

[0081] Specifically, the quantile method is used to calculate two types of scoring thresholds. The emotional distress scoring threshold is the 75th quantile of the total emotional distress scores of the same group at the same stage. This means that 75% of users in the group have a total score below this value; scores above this value are considered to indicate a high level of distress. The psychological resilience scoring threshold is the 25th quantile of the total psychological resilience scores of the same group at the same stage. This means that 25% of users in the group have a total score below this value; scores below this value are considered to indicate a low level of resilience. If historical data is insufficient, the system automatically uses default thresholds, with a default value of 12 for the emotional distress scoring threshold and 25 for the psychological resilience scoring threshold.

[0082] S304. Based on the total emotional distress score, the emotional distress scoring threshold, the total psychological resilience score, and the psychological resilience scoring threshold, determine the psychological risk category to which the user belongs, and generate the user's psychological assessment result based on the psychological risk category.

[0083] In one embodiment, the system classifies users' psychological risk categories based on comparisons between their total emotional distress score and corresponding thresholds, and their total psychological resilience score and corresponding thresholds. The system then combines and matches the comparison results from these two dimensions based on a preset risk category determination logic, outputting the corresponding psychological risk category. The psychological risk category can be the severity level of the psychological problem, the type of psychological problem, etc. Subsequently, based on the determined psychological risk category, a standardized psychological assessment result is generated.

[0084] For example, the psychological assessment results include the following three core components: The quantitative scoring section clearly displays the user's total emotional distress score X, total psychological resilience score Y, and corresponding scoring thresholds θ_X and θ_Y, and clearly marks the comparison relationship between each total score and the threshold; Qualitative risk assessment: Clearly record the psychological risk category to which the user belongs, as well as a description of the core psychological state characteristics corresponding to that category; Intervention recommendations: Based on the user's psychological risk category, a framework of directional follow-up intervention recommendations is provided to provide a basis for the generation of subsequent precise intervention plans.

[0085] Optionally, in one embodiment, an interpretation report is generated based on the total psychological resilience score, the total emotional distress score, the scores corresponding to the dialogue questions, the response data, the supplementary response data, and an interpretation framework developed by a psychologist. This interpretation report serves as the result of the psychological assessment. The interpretation report is used to explain the potential risk levels represented by the total psychological resilience score and the total emotional distress score, as well as intervention recommendations for the user.

[0086] In one embodiment, a large-scale model integrates multi-dimensional assessment data and combines it with an expert interpretation framework to generate personalized interpretation reports that are relevant to the user's specific scenario and text description. The interpretation framework refers to a set of report generation rules developed by psychology experts based on clinical assessment experience, standardized psychological scale interpretation guidelines, and the psychological development characteristics of different populations. The framework includes risk level determination criteria, empathy criteria for language, intervention suggestion matching logic, user-quoted language guidelines, and scenario-based association rules. Its core is to achieve a deep binding between assessment results and the user's specific scenario and text description, ensuring a balance between the professionalism, personalization, and scenario adaptability of the interpretation report.

[0087] The report interpretation should include at least an explanation of the potential risk levels represented by the total psychological resilience score and the total emotional distress score, as well as intervention recommendations for users.

[0088] Optionally, the report can also include original quotes from user responses for easier explanation and understanding. Specifically, the large model prioritizes 2-3 original statements from user text data that are strongly relevant to the target scenario and reflect the core psychological state for precise citation. These are then combined with scores for corresponding dialogue questions to conduct personalized analysis, enhancing the report's "tailor-made" nature. For example, citing a user's response, "When I can't concentrate on studying, I run a couple of laps on the track. After running, I feel much better and can study a little longer," combined with a 3-point score for psychological resilience questions and the pre-exam scenario, the analysis would be: "From your method of 'running to relieve stress when you can't concentrate on studying,' we can see that you are very good at finding suitable channels for emotional release during study breaks. This is the key reason for your high psychological resilience score. This method is suitable for the tense pre-exam study pace and can quickly relieve anxiety, making it a very practical self-regulation method."

[0089] Optionally, in one embodiment, the psychological risk category includes health, growth, crisis, and vulnerability. Before determining the user's psychological risk category based on the total emotional distress score, the emotional distress scoring threshold, the total psychological resilience score, and the psychological resilience scoring threshold, and generating the user's psychological assessment result based on the psychological risk category, the following steps S3041-S3043 are further included: S3041, construct a two-dimensional coordinate system with the total score of emotional distress as the first coordinate axis and the total score of psychological resilience as the second coordinate axis; In one embodiment, a Cartesian coordinate system is constructed with the total score of emotional distress as the first coordinate axis and the total score of psychological resilience as the second coordinate axis.

[0090] S3042, using the emotional distress score threshold as the boundary point of the first coordinate axis and the psychological resilience score threshold as the boundary point of the second coordinate axis, the straight lines formed by extending the boundary points on the first and second coordinate axes intersect, dividing the two-dimensional coordinate system into four quadrants, each quadrant corresponding to the psychological risk category. Specifically, the emotional distress score threshold is used as the dividing point of the first coordinate axis, and the psychological resilience score threshold is used as the dividing point of the second coordinate axis. Perpendicular lines are drawn to the corresponding coordinate axes through the two dividing points. The intersection of the two perpendicular lines divides the two-dimensional coordinate system into four quadrants, with each quadrant corresponding to a category of psychological risk.

[0091] The "healthy" category refers to a psychological state in which the user's emotional distress level is within the reference range and their psychological resilience level meets the reference standard. These users currently have no obvious psychological distress and have strong psychological resilience to cope with difficulties, which is an ideal and stable psychological state.

[0092] The "Growth" category refers to a psychological state in which the user's emotional distress exceeds the reference range, but their psychological resilience level meets the reference standard. These users currently have clear psychological distress, but they have strong psychological adjustment abilities and can resolve their distress and achieve optimization and improvement of their psychological state through targeted training, course learning, or external support.

[0093] The "crisis" category refers to a psychological state in which the user's emotional distress exceeds the reference range and their psychological resilience does not meet the reference standard. These users not only have significant psychological distress, but also lack effective self-regulation abilities, and their psychological state is at risk of imbalance and deterioration.

[0094] The Vulnerable Risk Category refers to a psychological state in which the user's emotional distress level is within the reference range, but their psychological resilience level does not meet the reference standard. These users do not currently have obvious psychological distress, but their psychological resilience is insufficient, and they are prone to psychological problems when faced with sudden stress or difficulties.

[0095] S3043, determine the target quadrant in which the total score of emotional distress and the total score of psychological resilience fall, and use the two-dimensional coordinate system as the psychological assessment result to display the psychological risk category to which the user belongs through the target quadrant.

[0096] Specifically, the target quadrant is defined by the user's total score for emotional distress and total score for psychological resilience, which form coordinate points. The quadrant in which these points fall within a two-dimensional coordinate system is the target quadrant. Optionally, the system uses a two-dimensional coordinate system, which includes coordinate axes, boundary points, quadrant divisions, and the user's coordinate point location, as the visualization part of the psychological assessment results.

[0097] Please see Figure 5 This diagram illustrates the results of a psychological assessment method provided in this embodiment of the specification. It shows a two-dimensional coordinate system, where the horizontal axis represents emotional distress assessment and the vertical axis represents psychological resilience score. The intersection of the axes is θ_X=12 and θ_Y=25. Based on this intersection, four quadrants are obtained, corresponding to four categories of psychological risk: health, growth, crisis, and vulnerability. Specifically, if X < θ_X and Y ≥ θ_Y, the intervention recommendation is mindfulness audio + exercise challenge; if X ≥ θ_X and Y ≥ θ_Y, the intervention recommendation is CBT and peer support; if X ≥ θ_X and Y < θ_Y, the intervention recommendation is 24-hour human assessment and referral; if X < θ_X and Y < θ_Y, the intervention recommendation is 8 weeks of resilience training and monthly follow-up.

[0098] In the embodiments of this specification, by categorizing the assessment types of dialogue questions into emotional distress assessment and psychological resilience assessment, a more detailed assessment of the user's current psychological state can be achieved. Furthermore, scene calibration and historical data calibration improve the accuracy of score generation, thereby generating more accurate psychological assessment results for the user.

[0099] Please see Figure 6 This document provides a flowchart illustrating a psychological assessment method as described in the embodiments of this specification. When a user clicks "Start," candidate scenarios are displayed based on their identity. The user selects the scenario of greatest interest and answers questions using natural language. Information gaps are assessed based on the user's responses. If the information gap is less than a second threshold, a Likert score is generated directly. If the second threshold is less than or equal to the information gap but less than or equal to the first threshold, medium-gain follow-up questions are initiated. If the information gap is greater than the first threshold, high-gain follow-up questions are initiated. This process repeats multiple times, generating an X-axis score (for emotional distress type questions) and continuing until all X-axis questions are completed. Then, a Y-axis score (for psychological resilience type questions) is generated, and this process is repeated until all Y-axis questions are completed. Finally, a total emotional distress score and a total psychological resilience score are obtained. Based on these scores, the user is assigned to the corresponding quadrant, resulting in a psychological assessment.

[0100] The following will be combined with the appendix Figure 7 This document provides a detailed description of the large-model-based psychological assessment device provided in the embodiments of this specification. It should be noted that the appendix... Figure 7 The large-model-based psychological assessment device described herein is used to perform the functions described in this manual. Figures 1-6 The methods shown in the embodiments are illustrated for ease of explanation, showing only the parts related to the embodiments of this specification. For specific technical details not disclosed, please refer to this specification. Figures 1-6 The example shown.

[0101] Please see Figure 7 This diagram illustrates a schematic representation of a large-model-based psychological assessment device provided in an exemplary embodiment of this specification. This large-model-based psychological assessment device can be implemented as all or part of a device through software, hardware, or a combination of both. The device 1 includes a scenario determination unit 11, a question generation unit 12, an answer acquisition unit 13, an information dimension determination unit 14, an information gap assessment unit 15, a follow-up questioning unit 16, a scoring unit 17, and an assessment unit 18.

[0102] Scene determination unit 11 is used to obtain the user's identity and display the candidate scene corresponding to the user's identity; Question generation unit 12 is used to generate dialogue questions by calling the psychological assessment question bank corresponding to the target scene from the candidate scenes in response to the user's selection of the target scene using a large model; Answer acquisition unit 13 is used to acquire the answer data input by the user in response to the dialogue question; Information dimension determination unit 14 is used to determine a first information dimension that is not covered in the key information dimensions required for scoring the dialogue question; Information missing assessment unit 15 is used to determine the degree of information missing based on the first information dimension and the key information dimension; The follow-up questioning unit 16 is used to generate follow-up questions using the large model if the information missing degree meets the follow-up questioning conditions, and to obtain the user's supplementary answer data for the follow-up question. Scoring unit 17 is used to generate a score for each of the dialogue questions based on the answer data and the supplementary answer data; Evaluation unit 18 is used to generate a psychological evaluation result for the user based on the score.

[0103] Optionally, the information dimension determination unit 14 is specifically used to perform semantic analysis on the answer data using the large model to obtain the target issues contained in the answer data; Obtain the key information dimensions corresponding to the target issue; Identify a second information dimension contained in the answer data and compare it with the key information dimension to obtain a first information dimension not covered in the answer data.

[0104] Optionally, the follow-up questioning unit 16 is specifically used to select initial follow-up questions under the target topic from the follow-up question pool if the information missing degree meets the follow-up questioning conditions; The target follow-up question is rewritten using the large model to obtain an empathic expression of the initial follow-up question. The empathic expression of the initial follow-up question is then output, and supplementary answer data from the user regarding the follow-up question is obtained.

[0105] Optionally, the follow-up questioning unit 16 is specifically used to obtain follow-up questions with a first gain value using the large model if the information missing degree is greater than a first threshold, and to obtain supplementary answer data from the user for the follow-up questions with the first gain value. If the information missing degree is less than or equal to the first threshold and greater than the second threshold, then the large model is used to obtain follow-up questions with a second gain value, and the user's supplementary answer data for the follow-up questions with the second gain value is obtained; wherein, the second threshold is less than the first threshold, and the second gain value is less than or equal to the first gain value; If the degree of missing information is less than the second threshold, then the follow-up questioning condition is not met.

[0106] Optionally, the follow-up questioning unit 16 is also used to record the effective response rate of the user to each of the follow-up questions; Adjust the gain value of the follow-up question based on the effective response rate.

[0107] Optionally, the follow-up questioning unit 16 is further configured to determine a third information dimension that is not covered in the key information dimensions required for scoring the dialogue question; Based on the third information dimension and the key information dimension, the degree of information missing after follow-up questioning is determined; If the information missing degree after the follow-up question meets the follow-up question condition, then proceed to the steps of generating follow-up questions using the large model, obtaining the user's supplementary answer data for the follow-up questions, and recording the number of follow-up question rounds generated for the user. If the number of follow-up questions reaches a threshold, then the scale questions in the psychological assessment question bank are invoked so that the user can select from the preset options of the scale questions to obtain supplementary answer data.

[0108] Optionally, the assessment types of the dialogue questions include emotional distress assessment types and psychological resilience assessment types; the scoring unit 17 is specifically used to generate scores corresponding to each of the dialogue questions based on the answer data and the supplementary answer data using the large model; Based on the scoring, determine the total emotional distress score corresponding to the dialogue question of the emotional distress assessment type and the total psychological resilience score corresponding to the dialogue question of the psychological resilience assessment type.

[0109] Optionally, the scoring unit 17 is specifically used to analyze the severity, frequency, and impact of the content described by the user in the answer data and the supplementary answer data using the large model; The severity, frequency, and impact are semantically matched with the scoring criteria of the psychological assessment question bank to obtain the score corresponding to each dialogue question.

[0110] Optionally, the scoring unit 17 is specifically used to calibrate the score using the large model based on the user's user identity, the stage to which the user belongs under the user identity, and preset calibration rules, so as to obtain a calibrated score.

[0111] Optionally, the scoring unit 17 is specifically used to calibrate the score using the large model based on the user's user identity, the stage to which the user belongs under the user identity, and preset calibration rules, to obtain a calibrated first score; The first rating is compared with the historical ratings of historical users who have the same user identity and belong to the same stage to obtain the rating deviation value. If the scoring deviation exceeds a preset threshold, the large model references historical scoring data from users with the same identity and at the same stage to re-score the dialogue question, resulting in a calibrated second score.

[0112] Optionally, the evaluation unit 18 is specifically used to obtain the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score. Based on the total score of emotional distress, the threshold for scoring emotional distress, the total score of psychological resilience, and the threshold for scoring psychological resilience, the psychological risk category to which the user belongs is determined, and a psychological assessment result for the user is generated based on the psychological risk category.

[0113] Optionally, the evaluation unit 18 is also used to obtain historical scores of historical users who have the same user identity and belong to the same stage as the user; Based on the historical scores, determine the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score.

[0114] Optionally, the psychological risk categories include health, growth, crisis, and vulnerability; the assessment unit 18 is specifically used to construct a two-dimensional coordinate system with the total score of emotional distress as the first coordinate axis and the total score of psychological resilience as the second coordinate axis. Using the emotional distress score threshold as the boundary point of the first coordinate axis and the psychological resilience score threshold as the boundary point of the second coordinate axis, the straight lines formed by extending the boundary points on the first and second coordinate axes intersect, dividing the two-dimensional coordinate system into four quadrants, each quadrant corresponding to the psychological risk category. Determine the target quadrant into which the total score of emotional distress and the total score of psychological resilience fall, and use the two-dimensional coordinate system as the psychological assessment result to display the psychological risk category to which the user belongs through the target quadrant.

[0115] Optionally, the assessment unit is further configured to generate an interpretation report based on the total psychological resilience score, the total emotional distress score, the scores corresponding to the dialogue questions, the response data, the supplementary response data, and the interpretation framework developed by the psychologist, and use the interpretation report as the psychological assessment result; the interpretation report is used to explain the possible risk levels represented by the total psychological resilience score and the total emotional distress score, as well as intervention suggestions for the user.

[0116] It should be noted that the above embodiments of the large-model-based psychological assessment device, when executing the large-model-based psychological assessment method, are only illustrative examples of the above functional module division. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the large-model-based psychological assessment device and the large-model-based psychological assessment method embodiments provided above belong to the same concept, and their implementation process is detailed in the method embodiments, which will not be repeated here.

[0117] It is understood that the large-model-based psychological assessment device provided in the embodiments of this specification can be a terminal device such as a mobile phone, computer, tablet computer, smartwatch or in-vehicle device, or it can be a module in the terminal device used to implement the large-model-based psychological assessment method.

[0118] The embodiment numbers in this specification are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0119] This specification also provides a storage medium storing a computer program, which, when executed by a processor, implements the above-described functionality. Figures 1-6 The detailed implementation process of the large-model-based psychological assessment method described in the illustrated embodiment can be found in [reference needed]. Figures 1-6 The specific details of the illustrated embodiments will not be elaborated here.

[0120] Please refer to Figure 8 This diagram illustrates the structure of an electronic device provided in an exemplary embodiment of this specification. The electronic device in this specification may include one or more components such as a processor 110, a memory 120, an input device 130, an output device 140, and a bus 150. The processor 110, memory 120, input device 130, and output device 140 may be connected via the bus 150.

[0121] Processor 110 may include one or more processing cores. Processor 110 connects to various parts of the electronic device using various interfaces and lines, and performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 120, and by calling data stored in memory 120. Optionally, processor 110 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). Processor 110 may integrate one or more of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user page, and applications; the GPU is responsible for rendering and drawing the displayed content; and the modem handles wireless communication. It is understood that the modem may also not be integrated into processor 110 and may be implemented separately using a communication chip.

[0122] The memory 120 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 120 may include non-transitory computer-readable storage medium. The memory 120 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the various method embodiments described above, etc. The operating system may be the Android system, including systems deeply developed based on the Android system, the iOS system developed by Apple Inc., including systems deeply developed based on the iOS system, or other systems.

[0123] The memory 120 can be divided into operating system space and user space. The operating system runs in the operating system space, while native and third-party applications run in user space. To ensure that different third-party applications can achieve good running performance, the operating system allocates corresponding system resources for each application. However, different application scenarios within the same third-party application have different requirements for system resources. For example, in local resource loading scenarios, third-party applications have high requirements for disk read speed; in animation rendering scenarios, third-party applications have high requirements for GPU performance. Since the operating system and third-party applications are independent of each other, the operating system often cannot promptly perceive the current application scenario of a third-party application, resulting in the operating system's inability to adapt system resources accordingly.

[0124] In order for the operating system to distinguish the specific application scenarios of third-party applications, it is necessary to establish data communication between the third-party applications and the operating system. This would allow the operating system to obtain the current scenario information of the third-party applications at any time, and then perform targeted system resource adaptation based on the current scenario.

[0125] The input device 130 is used to receive input instructions or data, and includes, but is not limited to, a keyboard, mouse, camera, microphone, or touch device. The output device 140 is used to output instructions or data, and includes, but is not limited to, a display device and a speaker. In one example, the input device 130 and the output device 140 can be combined, and the input device 130 and the output device 140 can be a touch display screen.

[0126] The touch display screen can be designed as a full-screen, curved screen, or irregularly shaped screen. It can also be designed as a combination of a full-screen and a curved screen, or a combination of an irregularly shaped screen and a curved screen; however, this specification does not limit the specific design of the embodiments.

[0127] In addition, those skilled in the art will understand that the structure of the electronic device shown in the above figures does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, the electronic device may also include radio frequency circuits, input units, sensors, audio circuits, WiFi modules, power supplies, Bluetooth modules, etc., which will not be described in detail here.

[0128] exist Figure 8 In the illustrated electronic device, the processor 110 can be used to call computer applications stored in the memory 120 and specifically perform the following operations: Obtain the user's identity and display the candidate scenarios corresponding to the user's identity; In response to the target scenario selected by the user from the candidate scenarios, a large model is used to call the psychological assessment question bank corresponding to the target scenario to generate dialogue questions; Obtain the user's response data to the dialogue question; Identify the first information dimension that is not covered in the key information dimensions required for scoring the dialogue question from the answer data; Based on the first information dimension and the key information dimension, the degree of information missing is determined; If the information missing degree meets the follow-up question condition, then the large model is used to generate follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained; Based on the answer data and the supplementary answer data, a score is generated for each of the dialogue questions; The psychological assessment results for the user are generated based on the rating.

[0129] In one embodiment, when the processor 110 performs the operation of determining a first information dimension that is not covered in the key information dimensions required for scoring the dialogue question, it specifically performs the following operations: The large model is used to perform semantic analysis on the response data to obtain the target topics contained in the response data; Obtain the key information dimensions corresponding to the target issue; Identify a second information dimension contained in the answer data and compare it with the key information dimension to obtain a first information dimension not covered in the answer data.

[0130] In one embodiment, when the processor 110 executes the following steps: if the information missing degree meets the follow-up question condition, then the large model is used to generate a follow-up question, and the user's supplementary answer data for the follow-up question is obtained: If the degree of missing information meets the follow-up question criteria, then initial follow-up questions under the target topic are selected from the follow-up question pool; The target follow-up question is rewritten using the large model to obtain an empathic expression of the initial follow-up question. The empathic expression of the initial follow-up question is then output, and supplementary answer data from the user regarding the follow-up question is obtained.

[0131] In one embodiment, when the processor 110 executes the following steps: if the information missing degree meets the follow-up question condition, then the large model is used to generate a follow-up question, and the user's supplementary answer data for the follow-up question is obtained: If the information missing degree is greater than the first threshold, the large model is used to obtain follow-up questions with a first gain value, and the user's supplementary answer data for the follow-up questions with the first gain value is obtained. If the information missing degree is less than or equal to the first threshold and greater than the second threshold, then the large model is used to obtain follow-up questions with a second gain value, and the user's supplementary answer data for the follow-up questions with the second gain value is obtained; wherein, the second threshold is less than the first threshold, and the second gain value is less than or equal to the first gain value; If the degree of missing information is less than the second threshold, then the follow-up questioning condition is not met.

[0132] In one embodiment, the processor 110 may also perform the following operations: Record the effective response rate of the user to each of the follow-up questions; Adjust the gain value of the follow-up question based on the effective response rate.

[0133] In one embodiment, after the processor 110 executes the operation of generating follow-up questions using the large model if the information missing degree meets the follow-up question condition, and obtains the user's supplementary answer data for the follow-up question, it is further configured to perform the following operations: Identify a third information dimension that is not covered in the key information dimensions required for scoring the dialogue question, based on the supplementary answer data; Based on the third information dimension and the key information dimension, the degree of information missing after follow-up questioning is determined; If the information missing degree after the follow-up question meets the follow-up question condition, then proceed to the steps of generating follow-up questions using the large model, obtaining the user's supplementary answer data for the follow-up questions, and recording the number of follow-up question rounds generated for the user. If the number of follow-up questions reaches a threshold, then the scale questions in the psychological assessment question bank are invoked so that the user can select from the preset options of the scale questions to obtain supplementary answer data.

[0134] In one embodiment, the assessment types of the dialogue questions include emotional distress assessment and psychological resilience assessment; when the processor 110 generates scores corresponding to each of the dialogue questions based on the response data and the supplementary response data, it specifically performs the following operations: The large model is used to generate scores for each of the dialogue questions based on the answer data and the supplementary answer data. Based on the scoring, determine the total emotional distress score corresponding to the dialogue question of the emotional distress assessment type and the total psychological resilience score corresponding to the dialogue question of the psychological resilience assessment type.

[0135] In one embodiment, when the processor 110 generates scores for each dialogue question based on the answer data and the supplementary answer data using the large model, it specifically performs the following operations: The large model is used to analyze the severity, frequency, and impact of the content described by the users in the response data and the supplementary response data; The severity, frequency, and impact are semantically matched with the scoring criteria of the psychological assessment question bank to obtain the score corresponding to each dialogue question.

[0136] In one embodiment, after performing semantic matching of the severity, frequency, and impact with the scoring criteria of the psychological assessment question bank to obtain the scores corresponding to each of the dialogue questions, the processor 110 further performs the following operations: The large model is used to calibrate the score based on the user's identity, the stage the user belongs to under the user identity, and preset calibration rules to obtain a calibrated score.

[0137] In one embodiment, when the processor 110 calibrates the score using the large model based on the user's identity, the stage the user belongs to under that identity, and preset calibration rules to obtain a calibrated score, it specifically performs the following operations: The large model is used to calibrate the score based on the user's identity, the stage the user belongs to under the user identity, and preset calibration rules to obtain a calibrated first score; The first rating is compared with the historical ratings of historical users who have the same user identity and belong to the same stage to obtain the rating deviation value. If the scoring deviation exceeds a preset threshold, the large model references historical scoring data from users with the same identity and at the same stage to re-score the dialogue question, resulting in a calibrated second score.

[0138] In one embodiment, when the processor 110 generates a psychological assessment result for the user based on the rating, it specifically performs the following operations: Obtain the emotional distress score threshold corresponding to the total emotional distress score, and the psychological resilience score threshold corresponding to the total psychological resilience score; Based on the total score of emotional distress, the threshold for scoring emotional distress, the total score of psychological resilience, and the threshold for scoring psychological resilience, the psychological risk category to which the user belongs is determined, and a psychological assessment result for the user is generated based on the psychological risk category.

[0139] In one embodiment, before the processor 110 performs the operations of obtaining the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score, the processor 110 further performs the following operations: Obtain historical ratings from historical users who have the same user identity and belong to the same stage as the user mentioned above; Based on the historical scores, determine the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score.

[0140] In one embodiment, the psychological risk categories include health, growth, crisis, and vulnerability; when the processor 110 determines the psychological risk category to which the user belongs based on the total emotional distress score, the emotional distress rating threshold, the total psychological resilience score, and the psychological resilience rating threshold, and generates the user's psychological assessment result based on the psychological risk category, it specifically performs the following operations: A two-dimensional coordinate system is constructed with the total score of emotional distress as the first coordinate axis and the total score of psychological resilience as the second coordinate axis. Using the emotional distress score threshold as the boundary point of the first coordinate axis and the psychological resilience score threshold as the boundary point of the second coordinate axis, the straight lines formed by extending the boundary points on the first and second coordinate axes intersect, dividing the two-dimensional coordinate system into four quadrants, each quadrant corresponding to the psychological risk category. Determine the target quadrant into which the total score of emotional distress and the total score of psychological resilience fall, and use the two-dimensional coordinate system as the psychological assessment result to display the psychological risk category to which the user belongs through the target quadrant.

[0141] In one embodiment, when the processor 110 generates a psychological assessment result for the user based on the rating, it specifically performs the following operations: An interpretation report is generated based on the total psychological resilience score, the total emotional distress score, the scores corresponding to the dialogue questions, the response data, the supplementary response data, and the interpretation framework developed by psychologists. The interpretation report serves as the result of the psychological assessment. The interpretation report is used to explain the possible risk levels represented by the total psychological resilience score and the total emotional distress score, as well as intervention suggestions for the user.

[0142] In the embodiments of this specification, by obtaining the user's identity, candidate scenarios corresponding to the user's identity are displayed. In response to the target scenario selected by the user from the candidate scenarios, a large model is used to call the psychological assessment question bank corresponding to the target scenario to generate dialogue questions. The user's answer data for the dialogue questions is obtained, and the first information dimension not covered in the key information dimensions required for scoring the dialogue questions is determined. Based on the first information dimension and the key information dimension, the information missing degree is determined. If the information missing degree meets the follow-up question conditions, the large model generates follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained. Based on the answer data and supplementary answer data, a score corresponding to each dialogue question is generated, and the user's psychological assessment result is generated based on the score. Generating psychological assessment questions based on the psychological assessment question bank corresponding to the user's scenario using a large model better meets the needs of different users. Simultaneously, the natural language capabilities of the large model transform it into an empathetic and interactive "chat" process, reducing user resistance to psychological screening and improving the completion rate. Combining information missing degree assessment to determine whether the user's answer meets the scoring requirements, and adaptively asking follow-up questions according to different information missing degrees, guides users to answer questions clearly, improving the accuracy of the assessment.

[0143] Furthermore, semantic analysis of the response data identified the first information dimension. Follow-up questions were then derived based on the topic of the responses to supplement information. This improves the accuracy and completeness of the psychological assessment data and reduces assessment errors. In addition, setting a threshold for the number of follow-up questions avoids both perfunctory answers due to excessive questioning and scoring bias caused by insufficient information. Unlike the mechanical questioning of traditional scales and the rigid follow-up questions of electronic questionnaires, this embodiment rewrites the initial follow-up questions into empathetic statements using a large model. It first perceives and accepts the user's emotions before initiating targeted questions, weakening the sense of assessment and strengthening the sense of communication, effectively reducing user psychological resistance.

[0144] Furthermore, by categorizing the assessment types of dialogue questions into emotional distress assessment and psychological resilience assessment, a more detailed assessment of the user's current psychological state can be achieved. Scene calibration and historical data calibration further improve the accuracy of score generation, resulting in more precise psychological assessment results for users.

[0145] Additionally, embodiments of this specification provide a computer program product comprising a computer program that, when executed by a processor of an electronic device, enables the processor to at least perform the functions described above. Figures 1 to 6 The method provided in the illustrated embodiment.

[0146] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0147] The above-disclosed embodiments are merely preferred embodiments of this specification and should not be construed as limiting the scope of this specification. Therefore, any equivalent variations made in accordance with the claims of this specification shall still fall within the scope of this specification.

Claims

1. A psychological assessment method based on a large model, characterized in that, The method includes: Obtain the user's identity and display the candidate scenarios corresponding to the user's identity; In response to the target scenario selected by the user from the candidate scenarios, a large model is used to call the psychological assessment question bank corresponding to the target scenario to generate dialogue questions; Obtain the user's response data to the dialogue question; Identify the first information dimension that is not covered in the key information dimensions required for scoring the dialogue question from the answer data; Based on the first information dimension and the key information dimension, the degree of information missing is determined; If the information missing degree meets the follow-up question condition, then the large model is used to generate follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained; Based on the answer data and the supplementary answer data, a score is generated for each of the dialogue questions; The psychological assessment results for the user are generated based on the rating.

2. The method as described in claim 1, characterized in that, The determination of the first information dimension not covered by the answer data in the key information dimensions required for scoring the dialogue question includes: The large model is used to perform semantic analysis on the response data to obtain the target topics contained in the response data; Obtain the key information dimensions corresponding to the target issue; Identify a second information dimension contained in the answer data and compare it with the key information dimension to obtain a first information dimension not covered in the answer data.

3. The method as described in claim 2, characterized in that, If the information missing degree meets the follow-up question condition, then the large model is used to generate follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained, including: If the degree of missing information meets the follow-up question criteria, then initial follow-up questions under the target topic are selected from the follow-up question pool; The target follow-up question is rewritten using the large model to obtain an empathic expression of the initial follow-up question. The empathic expression of the initial follow-up question is then output, and supplementary answer data from the user regarding the follow-up question is obtained.

4. The method as described in claim 1, characterized in that, If the information missing degree meets the follow-up question condition, then the large model is used to generate follow-up questions, and the user's supplementary answer data for the follow-up questions is obtained, including: If the information missing degree is greater than the first threshold, the large model is used to obtain follow-up questions with a first gain value, and the user's supplementary answer data for the follow-up questions with the first gain value is obtained. If the information missing degree is less than or equal to the first threshold and greater than the second threshold, then the large model is used to obtain follow-up questions with a second gain value, and the user's supplementary answer data for the follow-up questions with the second gain value is obtained; wherein, the second threshold is less than the first threshold, and the second gain value is less than or equal to the first gain value; If the degree of missing information is less than the second threshold, then the follow-up questioning condition is not met.

5. The method as described in claim 4, characterized in that, The method further includes: Record the effective response rate of the user to each of the follow-up questions; Adjust the gain value of the follow-up question based on the effective response rate.

6. The method as described in claim 1, characterized in that, If the information missing degree meets the follow-up question condition, then after generating follow-up questions using the large model and obtaining the user's supplementary answer data for the follow-up questions, the method further includes: Identify a third information dimension that is not covered in the key information dimensions required for scoring the dialogue question, based on the supplementary answer data; Based on the third information dimension and the key information dimension, the degree of information missing after follow-up questioning is determined; If the information missing degree after the follow-up question meets the follow-up question condition, then proceed to the steps of generating follow-up questions using the large model, obtaining the user's supplementary answer data for the follow-up questions, and recording the number of follow-up question rounds generated for the user. If the number of follow-up questions reaches a threshold, then the scale questions in the psychological assessment question bank are invoked so that the user can select from the preset options of the scale questions to obtain supplementary answer data.

7. The method as described in claim 1, characterized in that, The assessment types for the dialogue questions include emotional distress assessment and psychological resilience assessment. The step of generating a score for each dialogue question based on the answer data and the supplementary answer data includes: The large model is used to generate scores for each of the dialogue questions based on the answer data and the supplementary answer data. Based on the scoring, determine the total emotional distress score corresponding to the dialogue question of the emotional distress assessment type and the total psychological resilience score corresponding to the dialogue question of the psychological resilience assessment type.

8. The method as described in claim 7, characterized in that, The process of generating scores for each dialogue question using the large model based on the answer data and the supplementary answer data includes: The large model is used to analyze the severity, frequency, and impact of the content described by the users in the response data and the supplementary response data; The severity, frequency, and impact are semantically matched with the scoring criteria of the psychological assessment question bank to obtain the score corresponding to each of the dialogue questions; After semantically matching the severity, frequency, and impact with the scoring criteria of the psychological assessment question bank to obtain the score corresponding to each dialogue question, the method further includes: The large model is used to calibrate the score based on the user's identity, the stage the user belongs to under the user identity, and preset calibration rules to obtain a calibrated score.

9. The method as described in claim 8, characterized in that, The process involves using the large model to calibrate the score based on the user's identity, the stage the user belongs to under that identity, and preset calibration rules, to obtain a calibrated score, including: The large model is used to calibrate the score based on the user's identity, the stage the user belongs to under the user identity, and preset calibration rules to obtain a calibrated first score; The first rating is compared with the historical ratings of historical users who have the same user identity and belong to the same stage to obtain the rating deviation value. If the scoring deviation exceeds a preset threshold, the large model references historical scoring data from users with the same identity and at the same stage to re-score the dialogue question, resulting in a calibrated second score.

10. The method as described in claim 7, characterized in that, The process of generating the user's psychological assessment result based on the rating includes: Obtain the emotional distress score threshold corresponding to the total emotional distress score, and the psychological resilience score threshold corresponding to the total psychological resilience score; Based on the total score of emotional distress, the threshold for scoring emotional distress, the total score of psychological resilience, and the threshold for scoring psychological resilience, the psychological risk category to which the user belongs is determined, and a psychological assessment result for the user is generated based on the psychological risk category.

11. The method as described in claim 10, characterized in that, Before obtaining the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score, the method further includes: Obtain historical ratings from historical users who have the same user identity and belong to the same stage as the user mentioned above; Based on the historical scores, determine the emotional distress score threshold corresponding to the total emotional distress score and the psychological resilience score threshold corresponding to the total psychological resilience score.

12. The method according to any one of claims 10-11, characterized in that, The psychological risk categories include health, growth, crisis, and vulnerability; The process of determining the user's psychological risk category based on the total emotional distress score, the emotional distress scoring threshold, the total psychological resilience score, and the psychological resilience scoring threshold, and generating the user's psychological assessment result based on the psychological risk category, includes: A two-dimensional coordinate system is constructed with the total score of emotional distress as the first coordinate axis and the total score of psychological resilience as the second coordinate axis. Using the emotional distress score threshold as the boundary point of the first coordinate axis and the psychological resilience score threshold as the boundary point of the second coordinate axis, the straight lines formed by extending the boundary points on the first and second coordinate axes intersect, dividing the two-dimensional coordinate system into four quadrants, each quadrant corresponding to the psychological risk category. Determine the target quadrant into which the total score of emotional distress and the total score of psychological resilience fall, and use the two-dimensional coordinate system as the psychological assessment result to display the psychological risk category to which the user belongs through the target quadrant.

13. The method as described in claim 7, characterized in that, The process of generating the user's psychological assessment result based on the rating includes: An interpretation report is generated based on the total psychological resilience score, the total emotional distress score, the scores corresponding to the dialogue questions, the response data, the supplementary response data, and the interpretation framework developed by psychologists. The interpretation report serves as the result of the psychological assessment. The interpretation report is used to explain the possible risk levels represented by the total psychological resilience score and the total emotional distress score, as well as intervention suggestions for the user.

14. A psychological assessment device based on a large model, characterized in that, The device includes: The scene determination unit is used to obtain the user's identity and display the candidate scene corresponding to the user's identity. The question generation unit is used to generate dialogue questions by calling the psychological assessment question bank corresponding to the target scenario from the candidate scenarios in response to the user's selection of the target scenario using a large model; The answer acquisition unit is used to acquire the answer data input by the user in response to the dialogue question; An information dimension determination unit is used to determine a first information dimension that is not covered in the key information dimensions required for scoring the dialogue question in the answer data. An information missing assessment unit is used to determine the degree of information missing based on the first information dimension and the key information dimension; The follow-up questioning unit is used to generate follow-up questions using the large model if the information missing degree meets the follow-up questioning conditions, and to obtain supplementary answer data from the user for the follow-up question. A scoring unit is used to generate a score for each of the dialogue questions based on the answer data and the supplementary answer data; An assessment unit is used to generate a psychological assessment result for the user based on the score.

15. An electronic device, characterized in that, include: Processor and memory; The memory stores a computer program adapted to be loaded by the processor and to execute the steps of the method as described in any one of claims 1 to 13.

16. A storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 13.

17. A computer program product, characterized in that, include: A computer program, when executed by a processor of an electronic device, causes the processor to perform the steps of the method as described in any one of claims 1 to 13.

Images