Multi-agent-based examinee evaluation method and system and electronic equipment

Abstract

The invention discloses a multi-agent-based examinee evaluation method and system and electronic equipment, and the method comprises the steps: designing different cue words, guiding a large language model to play a plurality of exclusive agent roles, including a cognitive evaluation agent, an emotion understanding agent and a mental health agent; the answer content of examinees is divided according to modules, and different parts are respectively input to different agents for processing: the cognitive evaluation agent receives logic and knowledge type questions, the emotion understanding agent receives open expression type questions, and the psychological health agent receives psychological test questions to respectively carry out analysis and evaluation; each agent outputs evaluation content in a structured mode, and the output content comprises an evaluation conclusion and a reason; a comprehensive scoring agent is introduced, and evaluation results of different agents are fused and analyzed. According to the invention, a multi-agent collaborative evaluation mechanism is utilized, and multi-view analysis of experts is simulated, so that the comprehensiveness, intelligence and personalized feedback capability of examinee evaluation are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the field of large-scale intelligent agent technology, and specifically relates to a candidate assessment method, system and electronic equipment based on multi-agent. Background Art

[0002] Candidate assessment is a core task in educational measurement. It aims to comprehensively analyze candidates' multi-dimensional characteristics, including their knowledge, thinking skills, emotional state, and mental health. This is crucial for teaching feedback, personalized educational planning, and psychological intervention. Scientific and accurate candidate assessment not only helps teachers understand student learning outcomes but also provides a basis for subsequent learning resource recommendations and the development of competency pathways.

[0003] However, existing candidate assessment methods mostly focus on matching objective test scores with the similarity of subjective test texts, relying primarily on manual marking or rule-driven systems. These assessment dimensions are relatively narrow and fail to capture complex expressions such as emotional expression, psychological characteristics, and reasoning ability. Furthermore, traditional methods often focus on the final answer, ignoring the linguistic clues and deeper cognitive features of the answering process. This results in insufficient interpretability and low intelligence.

[0004] Therefore, there is a need for a candidate assessment method that can integrate multi-dimensional intelligent analysis capabilities and simulate the collaborative evaluation process of multiple experts to improve the comprehensiveness, accuracy and automation level of the assessment, so as to adapt to the needs of modern education scenarios for in-depth candidate profiling and personalized teaching support. Summary of the Invention

[0005] The main purpose of the present invention is to overcome the shortcomings and deficiencies of the existing technology and provide a multi-agent-based candidate assessment method, system and electronic equipment. The present invention uses a large language model to simulate multi-role agents to collaboratively analyze the content of candidate answers, thereby improving the intelligence level of candidate assessment, the comprehensiveness of assessment dimensions, and the interpretability and practicality of the results.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a multi-agent based candidate assessment method, comprising the following steps:

[0008] Design different prompt words to guide the large language model to play multiple dedicated agent roles, including cognitive assessment agent, emotion understanding agent, and mental health agent;

[0009] The test takers' answers are divided into modules and input into different agents for processing. The cognitive assessment agent is used to receive and process logic and knowledge questions, the emotional understanding agent is used to receive and process open expression questions, and the mental health agent is used to receive and process psychological test questions.

[0010] Each agent outputs evaluation content in a structured manner, which includes evaluation conclusions and reasons;

[0011] A comprehensive scoring agent is introduced, which receives the evaluation results of the cognitive assessment agent, the emotional understanding agent and the mental health agent, integrates and analyzes the evaluation results of different agents, and outputs the comprehensive score and text comments of the candidates.

[0012] As an optimal technical solution, the cognitive assessment agent processes logic questions through the subject knowledge graph constructed by Neo4j and the BERT-wwm+BiLSTM fusion model. It combines the text encoding of the examinee's answer with the cognitive graph node vector, uses bidirectional LSTM to calculate the hidden state, and finally outputs the cognitive score.

[0013] As a preferred technical solution, the processing process of the cognitive evaluation agent is as follows:

[0014] Based on Neo4j, a subject knowledge graph is constructed and the entity relationship is defined as G = (E, R), where E = {e1, e2, ..., e n} is an entity collection, for relationships between entities;

[0015] The cognitive score calculation process is as follows:

[0016]

[0017] h t =BiLSTM(h t-1 ,x t )

[0018] S cog =σ(W c MaxPooling(h 1:T ))

[0019] Among them, h0 represents the implicit state of time step 0, BERT-wwm(Q) is the encoding of the candidate text Q using the BERT-wwm model, and GraphEmb(e i ) is the entity e in the knowledge graph i vector, h t represents the hidden state at time step t; W crepresents the output layer weight matrix, σ represents the Sigmoid activation function, S cog Score cognition.

[0020] As a preferred technical solution, the emotion understanding agent uses a multi-level LSTM time series network to analyze emotion features and outputs emotion scores by calculating gating states such as forgetting gates.

[0021] As a preferred technical solution, the emotion understanding agent calculation process is as follows:

[0022] S emo =BiLSTM(I)

[0023] Among them, I={i1,i,...,i n} represents the set of emotional features at different time steps, S emo Rate the sentiment.

[0024] As a preferred technical solution, the mental health agent integrates the XLNet model and the PHQ-9 scale features, constructs a joint psychological feature vector, and outputs a mental health score.

[0025] As a preferred technical solution, the mental health agent is calculated as follows:

[0026]

[0027] S psy =MLP(H mental )

[0028] Among them, H mental represents the psychological feature vector; XLNet(T) is the text feature encoding of the test-taker's answer text to the psychological question using XLNet; PHQ9(S) is the structured feature converted using the PHQ-9 scale; S represents the test-taker's answer on the scale; represents feature concatenation, W p represents the feature projection matrix, MLP represents the fully connected layer neural network, S psy Score your mental health.

[0029] As a preferred technical solution, the evaluation results of different agents are analyzed and integrated, specifically:

[0030]

[0031] F final =α1S cog +α2S emo +α3S psy

[0032] Among them, α iis the learnable weight, F final is the final comprehensive score, S cog S is the cognitive score, emo Score the sentiment, psy Score your mental health.

[0033] In a second aspect, the present invention provides a multi-agent-based candidate assessment system, which is applied to the multi-agent-based candidate assessment method, and includes an agent module, a content evaluation module, an evaluation content output module, and a comprehensive scoring module;

[0034] The agent module is used to design different prompt words to guide the large language model to play multiple dedicated agent roles, including a cognitive assessment agent, an emotion understanding agent, and a mental health agent;

[0035] The content evaluation module is used to divide the test takers' answers into modules and input them into different agents for processing. The cognitive assessment agent is used to receive and process logic and knowledge questions, the emotion understanding agent is used to receive and process open expression questions, and the mental health agent is used to receive and process psychological test questions.

[0036] The evaluation content output module is used for each agent to output the evaluation content in a structured manner, and the output content includes the evaluation conclusion and reasons;

[0037] The comprehensive scoring module is used to introduce a comprehensive scoring agent, which receives the evaluation results of the cognitive assessment agent, the emotional understanding agent and the mental health agent, and integrates and analyzes the evaluation results of different agents to output the comprehensive score and text comments of the candidates.

[0038] In a third aspect, the present invention provides an electronic device, comprising:

[0039] at least one processor; and,

[0040] a memory communicatively connected to the at least one processor; wherein,

[0041] The memory stores computer program instructions that can be executed by the at least one processor, and the computer program instructions are executed by the at least one processor to enable the at least one processor to perform the multi-agent-based examinee assessment method.

[0042] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0043] The present invention comprehensively covers the multi-dimensional analysis of examinees' logical reasoning, emotional expression and psychological state by constructing a collaborative mechanism of three intelligent agents: cognitive assessment, emotional understanding and mental health, breaking through the single-dimensional limitation of traditional assessment. The cognitive assessment intelligent agent integrates subject knowledge graph and BERT-BiLSTM model to improve the accuracy of logical judgment, the emotional understanding intelligent agent adopts multi-level LSTM network to analyze temporal emotional characteristics, and the mental health intelligent agent combines XLNet and PHQ-9 scale to realize psychological risk grading, forming a deep technical integration advantage. Finally, the dynamic fusion mechanism of learnable weights of the comprehensive scoring intelligent agent is used to generate scores and structured comments that are both objective and personalized, thereby improving the comprehensiveness and accuracy of the assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0045] Figure 1 This is a flowchart of a multi-agent-based candidate assessment method according to an embodiment of the present invention;

[0046] Figure 2 This is a block diagram of a multi-agent-based candidate assessment system according to an embodiment of the present invention.

[0047] Figure 3 2 is a structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0049] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments.

[0050] See also Figure 1This embodiment provides a multi-agent-based candidate assessment method, which mainly includes the following steps:

[0051] Step S1: Design different prompt words to guide the large language model to play multiple dedicated agent roles, including: cognitive assessment agent (assessing IQ-related performance), emotional understanding agent (assessing EQ level) and mental health agent (assessing the psychological state of the examinee). The specific prompt words for different roles are:

[0052] For the cognitive assessment agent, the prompt is: "You are a cognitive ability assessment expert. Please evaluate the logic, knowledge application ability, and depth of reasoning of the candidates in the logic and knowledge questions, and explain the basis for your evaluation."

[0053] For the emotion understanding agent, the prompt is: "You are an emotion understanding expert. Please analyze the candidate's emotional expression, empathy, and emotion regulation ability during the open-ended expression questions, and provide your evaluation conclusions and reasons."

[0054] For the mental health agent, the prompt is: "You are a mental health expert. Please judge whether the candidate has psychological states such as anxiety, depression, avoidance, etc. based on the answers to the psychological test questions, and explain the basis for your judgment."

[0055] Step S2: Divide the candidate's answers into modules according to content type and input them into corresponding agents. The module set is: D = {d1, d1, d3}, representing logic knowledge questions, open expression questions, and psychological assessment questions, respectively; the agent set is: A = {A1, A2, A3}, representing cognitive assessment agents, emotional understanding agents, and psychological health agents, respectively. Each module is input into the corresponding agent for processing. The evaluation process is represented by the following formula:

[0056] O i =LLM(A i ,d i )

[0057] Among them O i For agent A i For module d i The above mechanism ensures that each type of question is handled by the agent that best fits its attributes, thereby improving the accuracy and pertinence of the evaluation.

[0058] Furthermore, the cognitive assessment agent processes logic questions using the subject knowledge graph built by Neo4j and the BERT-wwm+BiLSTM fusion model. After combining the text encoding of the examinee's answer with the cognitive graph node vector, it uses a bidirectional LSTM to calculate the hidden state and finally outputs the cognitive score. The processing process is as follows:

[0059] Based on Neo4j, a subject knowledge graph is constructed and the entity relationship is defined as G = (E, R), where E = {e1, e1, ..., e n} is an entity collection, for relationships between entities;

[0060] The cognitive score calculation process is as follows:

[0061]

[0062] h i =BiLSTM(h t-1 ,x t )

[0063] S cog =σ(W c MaxPooling(h 1:T ))

[0064] Among them, BERT-wwm(Q) is the encoding of the candidate text Q using the BERT-wwm model, GraphEmb(e i ) is the entity e in the knowledge graph i vector, h i represents the hidden state at time step i; W c represents the output layer weight matrix, σ represents the Sigmoid activation function, S cog Score cognition.

[0065] Furthermore, the emotion understanding agent uses a multi-level LSTM time series network to analyze emotion features and outputs an emotion score by calculating the gating states such as the forget gate. The calculation process is as follows:

[0066] S emo =BiLSTM(I)

[0067] Among them, I={i1,i,...,i n} represents the set of emotional features at different time steps, S emo Rate the sentiment.

[0068] Furthermore, the mental health agent integrates the XLNet model with the PHQ-9 scale features to construct a joint psychological feature vector and output a mental health score. The calculation process is as follows:

[0069]

[0070] S psy =MLP(H mental )

[0071] Among them, Hmental represents the psychological feature vector; XLNet(T) is the text feature encoding of the test-taker's answer text to the psychological question using XLNet; PHQ9(S) is the structured feature converted using the PHQ-9 scale; S represents the test-taker's answer on the scale; represents feature concatenation, W p represents the feature projection matrix, MLP represents the fully connected layer neural network, S psy Score your mental health.

[0072] Step S3: The output format of each agent is a structured binary:

[0073] O i =(c i ,r i )

[0074] Among them, c i The evaluation conclusions generated for the agent, such as "strong logical reasoning ability" and "positive emotional expression"; i The reasoning to support this conclusion comes from the semantic interpretation and inference process of the answer content.

[0075] Step S4: Finally, the comprehensive fusion module uses learnable weights to achieve dynamic weighted fusion of the three-modal evaluation results and generate a comprehensive evaluation report. The weighted fusion calculation is as follows:

[0076]

[0077] F final =α1S cog +α2S emo +α3S psy

[0078] Among them, α i is the learnable weight, F final is the final comprehensive score, S cog S is the cognitive score, emo Score the sentiment, psy Score your mental health.

[0079] The present invention constructs a collaborative mechanism of three intelligent agents, namely cognitive assessment, emotional understanding and mental health, through a large language model, which significantly improves the intelligence level and multi-dimensional coverage capability of candidate assessment: the cognitive assessment agent integrates subject knowledge graphs and deep neural network to analyze logical reasoning capabilities, the emotional understanding agent uses a temporal network to capture the dynamic characteristics of emotions, and the mental health agent combines psychological scales to achieve risk grading, forming a professional analysis system; through the division of labor in processing logic, open expression and psychological test questions, the three core qualities of reasoning ability, emotional literacy and psychological state are simultaneously covered; each intelligent agent outputs structured conclusions and natural language reasoning basis, and the comprehensive scoring agent generates traceable personalized comments based on dynamic weight fusion, which greatly enhances the practical value of assessment feedback while ensuring the interpretability of the results.

[0080] It should be noted that, for the sake of convenience, the aforementioned method embodiments are all expressed as a series of action combinations, but those skilled in the art should know that the present invention is not limited to the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously.

[0081] Based on the same concept as the multi-agent-based candidate assessment method in the above-mentioned embodiment, the present invention also provides a multi-agent-based candidate assessment system, which can be used to implement the above-mentioned multi-agent-based candidate assessment method. For ease of explanation, the structural diagram of the embodiment of the multi-agent-based candidate assessment system only shows the parts related to the embodiment of the present invention. Those skilled in the art will understand that the illustrated structure does not constitute a limitation of the device, and it can include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0082] See also Figure 2 In another embodiment of the present application, a multi-agent-based examinee assessment system 100 is provided, which includes an agent module 101, a content evaluation module 102, an evaluation content output module 103, and a comprehensive scoring module 104;

[0083] The agent module 101 is used to design different prompt words to guide the large language model to play multiple dedicated agent roles, including a cognitive assessment agent, an emotion understanding agent, and a mental health agent;

[0084] The content evaluation module 102 is used to divide the test-taker's answers into modules and input them into different agents for processing. The cognitive assessment agent is used to receive and process logic and knowledge questions, the emotion understanding agent is used to receive and process open expression questions, and the mental health agent is used to receive and process psychological test questions.

[0085] The evaluation content output module 103 is used for each agent to output the evaluation content in a structured manner, and the output content includes the evaluation conclusion and reasons;

[0086] The comprehensive scoring module 104 is used to introduce a comprehensive scoring agent, which receives the evaluation results of the cognitive assessment agent, the emotional understanding agent and the mental health agent, and integrates and analyzes the evaluation results of different agents to output the comprehensive score and text comments of the candidates.

[0087] It should be noted that the multi-agent-based candidate assessment system of the present invention corresponds one-to-one to the multi-agent-based candidate assessment method of the present invention. The technical features and beneficial effects described in the above-mentioned embodiment of the multi-agent-based candidate assessment method are applicable to the embodiment of the multi-agent-based candidate assessment. For specific contents, please refer to the description in the embodiment of the method of the present invention. No further details will be given here. This is hereby declared.

[0088] In addition, in the implementation of the multi-agent candidate assessment system in the above embodiment, the logical division of each program module is only an example. In actual application, the above functions can be assigned to different program modules as needed, for example, for the configuration requirements of the corresponding hardware or the convenience of software implementation. That is, the internal structure of the multi-agent candidate assessment system can be divided into different program modules to complete all or part of the functions described above.

[0089] See also Figure 3 In one embodiment, an electronic device for implementing a multi-agent-based candidate assessment method is provided. The electronic device 200 may include a first processor 201, a first memory 202 and a bus, and may also include a computer program stored in the first memory 202 and executable on the first processor 201, such as a multi-agent-based candidate assessment program 203.

[0090] The first memory 202 includes at least one type of readable storage medium, including a flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the first memory 202 may be an internal storage unit of the electronic device 200, such as a mobile hard disk of the electronic device 200. In other embodiments, the first memory 202 may also be an external storage device of the electronic device 200, such as a plug-in mobile hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash memory card, etc. equipped on the electronic device 200. Furthermore, the first memory 202 may include both an internal storage unit of the electronic device 200 and an external storage device. The first memory 202 can be used not only to store application software and various types of data installed in the electronic device 200, such as the code of the multi-agent candidate assessment program 203, but also to temporarily store data that has been output or is about to be output.

[0091] In some embodiments, the first processor 201 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including a combination of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The first processor 201 is the control core (Control Unit) of the electronic device, connecting the various components of the entire electronic device using various interfaces and lines, and executing or executing programs or modules stored in the first memory 202, as well as calling data stored in the first memory 202, to perform various functions of the electronic device 200 and process data.

[0092] Figure 3 Only the electronic device with components is shown, and it can be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the electronic device 200 , and the electronic device 200 may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0093] The multi-agent candidate assessment program 203 stored in the first memory 202 of the electronic device 200 is a combination of multiple instructions. When executed in the first processor 201, the program can achieve the following:

[0094] Design different prompt words to guide the large language model to play multiple dedicated agent roles, including cognitive assessment agent, emotion understanding agent, and mental health agent;

[0095] The test takers' answers are divided into modules and input into different agents for processing. The cognitive assessment agent is used to receive and process logic and knowledge questions, the emotional understanding agent is used to receive and process open expression questions, and the mental health agent is used to receive and process psychological test questions.

[0096] Each agent outputs evaluation content in a structured manner, which includes evaluation conclusions and reasons;

[0097] A comprehensive scoring agent is introduced, which receives the evaluation results of the cognitive assessment agent, the emotional understanding agent and the mental health agent, integrates and analyzes the evaluation results of different agents, and outputs the comprehensive score and text comments of the candidates.

[0098] Furthermore, if the modules / units integrated in the electronic device 200 are implemented as software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium. The computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0099] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0100] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0101] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. A multi-agent-based candidate assessment method, characterized in that: The steps include: Design different prompt words to guide the large language model to play multiple dedicated agent roles, including cognitive assessment agent, emotion understanding agent, and mental health agent; The test takers' answers are divided into modules and input into different agents for processing. The cognitive assessment agent is used to receive and process logic and knowledge questions, the emotional understanding agent is used to receive and process open expression questions, and the mental health agent is used to receive and process psychological test questions. Each agent outputs evaluation content in a structured manner, which includes evaluation conclusions and reasons; A comprehensive scoring agent is introduced, which receives the evaluation results of the cognitive assessment agent, the emotional understanding agent and the mental health agent, integrates and analyzes the evaluation results of different agents, and outputs the comprehensive score and text comments of the candidates.

2. The multi-agent based examinee assessment method according to claim 1, characterized in that: The cognitive assessment agent processes logic questions through the subject knowledge graph built by Neo4j and the BERT-wwm+BiLSTM fusion model. It combines the text encoding of the examinee's answer with the cognitive graph node vector, uses bidirectional LSTM to calculate the hidden state, and finally outputs the cognitive score.

3. The multi-agent based examinee assessment method according to claim 2, characterized in that: The processing process of the cognitive evaluation agent is as follows: Based on Neo4j, a subject knowledge graph is constructed and the entity relationship is defined as G = (E, R), where E = {e1, e2, ..., e n } is an entity collection, for the relationships between entities; The cognitive score calculation process is as follows: h t =BiLSTM(h t-1 ,x t ) S cog =σ(W c ·MaxPooling(h 1:T )) Among them, h0 represents the implicit state of time step 0, BERT-wwm(Q) is the encoding of the candidate text Q using the BERT-wwm model, and GraphEmb(e i ) is the entity e in the knowledge graph i vector, h t represents the hidden state at time step t; W c represents the output layer weight matrix, σ represents the Sigmoid activation function, S cog Score cognition.

4. The multi-agent based examinee assessment method according to claim 1, characterized in that: The emotion understanding agent uses a multi-level LSTM time series network to analyze emotion features and outputs emotion scores by calculating gating states such as forgetting gates.

5. The multi-agent based examinee assessment method according to claim 4, characterized in that: The calculation process of the emotion understanding agent is as follows: S emo =BiLSTM(I) Among them, I={i1,i,...,i n } represents the set of emotional features at different time steps, S emo Rate the sentiment.

6. The multi-agent based examinee assessment method according to claim 1, characterized in that: The mental health agent integrates the XLNet model and the PHQ-9 scale features, constructs a joint psychological feature vector, and outputs a mental health score.

7. The multi-agent based examinee assessment method according to claim 6, characterized in that: The mental health agent calculates as follows: S psy =MLP(H mental ) Among them H mental represents the psychological feature vector; XLNet(T) uses XLNet to encode the text features of the test-taker's answer text to the psychological question; PHQ9(S) is a structural feature converted using the PHQ-9 scale; S represents the examinee’s answer on the scale; represents feature concatenation, W p represents the feature projection matrix, MLP represents the fully connected layer neural network, S pst Score your mental health.

8. The multi-agent based examinee assessment method according to claim 1, characterized in that: The evaluation results of different agents are analyzed and integrated, specifically: F final =α1S cog +α2S emo +α3S psy Among them, α i is the learnable weight, F final is the final comprehensive score, S cog S is the cognitive score, emo Score the sentiment, psy Score your mental health.

9. The multi-agent-based candidate assessment system is characterized by: The multi-agent-based examinee assessment method applied to any one of claims 1-8, comprising an agent module, a content evaluation module, an evaluation content output module, and a comprehensive scoring module; The agent module is used to design different prompt words to guide the large language model to play multiple dedicated agent roles, including a cognitive assessment agent, an emotion understanding agent, and a mental health agent; The content evaluation module is used to divide the test takers' answers into modules and input them into different agents for processing. The cognitive assessment agent is used to receive and process logic and knowledge questions, the emotion understanding agent is used to receive and process open expression questions, and the mental health agent is used to receive and process psychological test questions. The evaluation content output module is used for each agent to output the evaluation content in a structured manner, and the output content includes the evaluation conclusion and reasons; The comprehensive scoring module is used to introduce a comprehensive scoring agent, which receives the evaluation results of the cognitive assessment agent, the emotional understanding agent and the mental health agent, and integrates and analyzes the evaluation results of different agents to output the comprehensive score and text comments of the candidates.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores computer program instructions that can be executed by the at least one processor, and the computer program instructions are executed by the at least one processor to enable the at least one processor to execute the multi-agent-based examinee assessment method as described in any one of claims 1-8.

Images