Intelligent examination method and device

Abstract

The embodiment of the invention provides an intelligent examination method and device, and the method comprises the steps: receiving test paper parameters of an administrator, carrying out the correlation verification of the test paper parameters and a set teaching outline knowledge graph, obtaining a test paper rule, screening test questions from a preset teaching question bank according to the test paper rule, and determining a corresponding preliminary screening test question set; performing semantic matching on question stem keywords of the preliminary screening test question set and the teaching outline knowledge graph, constructing a secondary screening test question set, performing parameterization screening on the secondary screening test question set according to a project reaction theory to generate final test paper, sending the test paper to an examinee terminal, continuously monitoring the state of an examination network by the examinee terminal, and if the network is interrupted, stopping the examination. If yes, storing the examinee operation data to a local encryption container in a one-way hash chain mode; when the network is reconnected, the one-way hash chain sent by the examinee end is received, whether the one-way hash chain is broken or not is judged, if not, data synchronization is carried out according to the one-way hash chain, and the examination is completed, and the efficiency and accuracy of the examination system can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and specifically to an intelligent examination method and device. Background Art

[0002] In current school examination systems, computer-based online and offline testing systems have gradually become mainstream with the continuous development of information technology. These systems typically include backend question bank uploads and test paper generation, online testing, and related data processing and analysis capabilities. However, despite the progress made in existing technologies, some significant shortcomings and deficiencies remain.

[0003] Traditional test-generating systems often lack self-checking capabilities for syllabus compliance, resulting in deviations from teaching requirements and impacting exam effectiveness and accuracy. Furthermore, network outages are a common problem during exams, and existing methods for handling these outages are often inflexible and unreliable, easily leading to loss of candidate data or interruption of the exam process. Furthermore, traditional analysis systems often focus on simple statistical analysis of candidate responses, lacking in-depth data mining and analysis, making them unable to provide valuable feedback for teaching.

[0004] In response to the above problems, an intelligent examination method is urgently needed to improve the efficiency and accuracy of the examination system. Summary of the Invention

[0005] In response to the problems in the prior art, the present application provides an intelligent examination method and device that can improve the efficiency and accuracy of the examination system.

[0006] In order to solve at least one of the above problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides an intelligent examination method, comprising:

[0008] Receive the test parameters from the administrator, perform correlation verification on the test parameters and the set syllabus knowledge graph, determine the corresponding test rules based on the test parameters that have undergone correlation verification, screen test questions from the preset teaching question bank according to the test rules, and determine the corresponding preliminary screening test question set;

[0009] Extracting keywords from the question stems of the initial screening test set using entity extraction technology, performing semantic matching between the question stem keywords and the syllabus knowledge graph to determine corresponding semantic matching degrees, comparing the semantic matching degrees with a preset matching threshold, constructing a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, performing parameterized screening on the secondary screening test set based on item response theory, generating a final test paper based on the test questions that have undergone the parameterized screening, and sending the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain.

[0010] When the network is reconnected, the one-way hash chain sent by the candidate is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test.

[0011] Furthermore, before correlating and verifying the test test parameters with the set syllabus knowledge graph, the method further includes:

[0012] Perform semantic segmentation on the preset teaching syllabus document, determine the corresponding knowledge point entities, establish a knowledge point alias library, uniformly map the knowledge point entities according to the knowledge point alias library, and determine the corresponding unified entity nodes and the dependency relationships between the entity nodes;

[0013] Constructing a multi-layer ontology structure layer based on the unified entity node, assigning dynamic weights to the entity nodes of each ontology structure layer based on the examination frequency of the teaching question bank, and determining corresponding weighted entity nodes;

[0014] The empowered entity nodes are subject to rule constraints according to a preset OWL rule engine, and the corresponding set teaching syllabus knowledge graph is determined based on the indirect dependency relationships obtained after the rule constraints, the dependency relationships between the entity nodes, and the empowered entity nodes.

[0015] Furthermore, the test paper setting parameters are associated with the set syllabus knowledge graph for verification, and the corresponding test paper setting rules are determined according to the test paper setting parameters that have undergone the associated verification, including:

[0016] Parsing the knowledge point distribution weights in the test test parameters, matching the knowledge point distribution weights with the node weights in the syllabus knowledge graph, and determining the corresponding knowledge point distribution weight deviations;

[0017] Determine whether the knowledge point distribution weight deviation exceeds a preset deviation threshold; if not, determine the corresponding test paper composition rule according to the test paper composition parameters.

[0018] Furthermore, performing semantic matching between the question stem keywords and the syllabus knowledge graph to determine the corresponding semantic matching degree includes:

[0019] Vectorize the question keywords based on the preset BERT model to determine the corresponding semantic features;

[0020] Cosine similarity calculation is performed based on the semantic features and the teaching outline knowledge graph node vector to determine the corresponding semantic matching degree.

[0021] Furthermore, the parametric screening of the second screening test question set according to the item response theory and the generation of the final test paper based on the parametrically screened test questions include:

[0022] Constructing a three-parameter logic model, the three-parameter logic model including a discrimination parameter, a difficulty parameter, and a guessing parameter, wherein the discrimination parameter is used to characterize the ability of the test question to discriminate between examinees of different ability levels, the difficulty parameter is used to characterize the absolute difficulty level of the test question, and the guessing parameter is used to characterize the probability of the examinee randomly guessing the test question correctly;

[0023] The core features of the second-screened test question set are quantified according to the three-parameter logical model, the multi-parameter features of the corresponding second-screened test question set are determined, the multi-parameter features are screened based on preset constraints, and the final test paper is generated based on the screened test questions.

[0024] Furthermore, the candidate terminal continuously monitors the test network status. If the network is interrupted, the candidate's operation data is stored in a local encrypted container in the form of a one-way hash chain, including:

[0025] The examinee terminal continuously monitors the examination network status. If the network is interrupted, the examinee's real-time operation data is obtained, the real-time operation data is encrypted using a preset encryption algorithm, and the encrypted data is packaged into multiple independent data blocks based on preset time intervals, wherein the examinee's real-time operation data includes the examinee's operation log, answer data, and timestamp;

[0026] Generate a corresponding hash value for each independent data block, embed the hash value of the previous data block into the header of the current block, determine the corresponding one-way hash chain, and store the one-way hash chain in a local encrypted container.

[0027] Furthermore, the receiving of the one-way hash chain sent by the examinee terminal, determining whether the one-way hash chain is broken, and if not, performing data synchronization according to the one-way hash chain, includes:

[0028] Receive the one-way hash chain sent by the candidate, decrypt the one-way hash chain block by block, and verify the continuity;

[0029] If the header hash value of the current block matches the measured hash value of the previous block, it is determined that the hash chain is not broken, and data synchronization is performed according to the one-way hash chain.

[0030] In a second aspect, the present application provides an intelligent examination device, comprising:

[0031] The test paper and question set preliminary screening module is used to receive the test paper parameters from the administrator, associate and verify the test paper parameters with the set teaching syllabus knowledge map, determine the corresponding test paper rules based on the test paper parameters that have undergone the association verification, and screen test questions from the preset teaching question bank according to the test paper rules to determine the corresponding preliminary screening test question set, wherein the historical form usage data includes at least one of historical form performance data and historical form structure change data;

[0032] a test paper question set determination module, configured to extract question stem keywords from the primary screening test set based on entity extraction technology, perform semantic matching between the question stem keywords and the syllabus knowledge graph, determine the corresponding semantic matching degree, compare the semantic matching degree with a preset matching threshold, construct a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, perform parameterized screening on the secondary screening test set based on item response theory, generate a final test paper based on the test questions that have undergone the parameterized screening, and send the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain;

[0033] The offline re-examination module is used to receive the one-way hash chain sent by the candidate when reconnecting to the network, determine whether the one-way hash chain is broken, and if not, synchronize data according to the one-way hash chain to complete the examination.

[0034] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the intelligent examination method when executing the program.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the intelligent examination method when executed by a processor.

[0036] In a fifth aspect, the present application provides a computer program product, comprising a computer program / instruction, which implements the steps of the intelligent examination method when executed by a processor.

[0037] It can be seen from the above technical solution that the present application provides an intelligent examination method and device, which receives the test paper parameters of the administrator, associates and verifies the test paper parameters with the set teaching syllabus knowledge graph to obtain the test paper rules, and screens the test questions from the preset teaching question bank according to the test paper rules to determine the corresponding primary screening test question set; semantic matching is performed on the question stem keywords of the primary screening test question set and the teaching syllabus knowledge graph to construct a secondary screening test question set, and the secondary screening test question set is parameterized and screened according to the item response theory to generate the final test paper, and the test paper is sent to the candidate end, and the candidate end continuously monitors the test network status. If the network is interrupted, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test, thereby improving the efficiency and accuracy of the examination system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are 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.

[0039] Figure 1 This is one of the flow charts of the intelligent examination method in the embodiment of the present application;

[0040] Figure 2 This is the second flow chart of the intelligent examination method in the embodiment of the present application;

[0041] Figure 3 This is the third flow chart of the intelligent examination method in the embodiment of the present application;

[0042] Figure 4 This is a fourth flow chart of the intelligent examination method in an embodiment of the present application;

[0043] Figure 5 This is the fifth flow chart of the intelligent examination method in the embodiment of the present application;

[0044] Figure 6 This is the sixth flow chart of the intelligent examination method in the embodiment of the present application;

[0045] Figure 7 This is the seventh flow chart of the intelligent examination method in the embodiment of the present application;

[0046] Figure 8 This is a structural diagram of the intelligent examination device in an embodiment of the present application;

[0047] Figure 9 Schematic diagram of the structure of the electronic device in the embodiment of the present application.

[0048] Reference numerals:

[0049] Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver program storage unit 9144, antenna 9111, speaker 9131, microphone 9132. DETAILED DESCRIPTION

[0050] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0051] The acquisition, storage, use, and processing of data in this application's technical solution comply with relevant national laws and regulations.

[0052] Taking into account the problems that in the traditional examination process, the test paper composition method often lacks the self-checking function of compliance with the syllabus, and the network disconnection processing method during the examination is often not flexible and reliable. The present application provides an intelligent examination method and device, which receives the test paper composition parameters of the administrator, associates and verifies the test paper composition parameters with the set syllabus knowledge map to obtain the test paper composition rules, and screens the test questions from the preset teaching question bank according to the test paper composition rules to determine the corresponding primary screening test question set; semantically matches the question stem keywords of the primary screening test question set with the syllabus knowledge map, constructs the secondary screening test question set, and performs parameterized screening on the secondary screening test question set according to the item response theory to generate the final test paper, and sends the test paper to the candidate end. The candidate end continuously monitors the test network status. If the network is disconnected, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the examination, thereby improving the efficiency and accuracy of the examination system.

[0053] In order to improve the efficiency and accuracy of the examination system, this application provides an embodiment of an intelligent examination method. Figure 1 The intelligent examination method specifically includes the following contents:

[0054] Step S101: receiving the test parameters from the administrator, correlating and verifying the test parameters with the set syllabus knowledge graph, determining corresponding test rules based on the correlatively verified test parameters, screening test questions from a preset teaching question bank based on the test rules, and determining a corresponding preliminary screening test question set;

[0055] Optionally, in this embodiment, this step is to add a "syllabus compliance self-check" function to address the above-mentioned test paper composition defects.

[0056] Optionally, in this embodiment, the administrator's test-setting parameters are defined by the test-setting teacher. These parameters include the knowledge point coverage, question type ratio, difficulty distribution, and total score. Verifying these parameters against the syllabus ensures they meet course assessment requirements. This is the first step in implementing the "Compliance Self-Check" feature.

[0057] Optionally, in this embodiment, in order to achieve relevance verification with the syllabus, we first construct a syllabus knowledge graph.

[0058] First, extract the syllabus knowledge point entities and dependency relationships.

[0059] Specifically, the latest teaching syllabus document is parsed into a three-level structure of "chapter-section-knowledge point" using a Transformer-based hierarchical segmentation model to extract knowledge point entities.

[0060] Chapter division is divided into logical units through title style recognition (such as "Chapter 3 Functions") and semantic coherence analysis (BERT sentence embedding cosine similarity > 0.9).

[0061] Knowledge point extraction uses the pre-trained ALBERT model combined with domain-adaptive fine-tuning to identify knowledge point entities, such as extracting "quadratic function" and "image properties" from "understanding the image properties of quadratic functions."

[0062] Specifically, a dynamically updated knowledge point alias library is constructed to solve the problem of multiple representations of the same knowledge point entity.

[0063] The semantic similarity of candidate aliases is calculated based on contrastive learning (SimCSE) (e.g., "Pythagorean Theorem" ≈ "Pythagorean Theorem"), and a mapping is established when the threshold is greater than 0.85.

[0064] Specifically, based on dependency syntax analysis and rule template matching, explicit dependencies are identified and explicit dependency relationships between knowledge point entities are obtained.

[0065] Secondly, the extracted knowledge point entities and dependency relationships are processed, entity weight values ​​are assigned to the knowledge point entities, and hidden dependency relationships are mined for the dependency relationships.

[0066] Specifically, we use a multi-layer ontology structure and dynamic weighting to transform discrete knowledge points into a multi-layer network with teaching semantics. We design a four-layer ontology model as follows:

[0067] The subject layer ("mathematics") uses ISO 15926 standard coding to ensure cross-system compatibility;

[0068] Module layer ("algebra", "geometry");

[0069] Knowledge point layer ("quadratic function" and "equation");

[0070] Ability dimension layer (“quantitative level of logical reasoning”);

[0071] Hypergraph is used to represent cross-layer associations, allowing a single node to belong to multiple parents ("vector" belongs to both algebraic and geometric modules).

[0072] Specifically, knowledge point entities are weighted based on question bank data. The frequency of knowledge points in exam questions over the past five years is counted and normalized to a value between 0 and 1. A time decay factor (half-life of 1 year) is used to ensure the timeliness of the weights, and weight values ​​are dynamically assigned to knowledge point entities.

[0073] Specifically, we mine implicit dependencies for dependencies. We define a customized rule set for the teaching domain, build a rule engine, declare pre-requisites as transitive attributes based on the rule engine, automatically deduce indirect dependencies, and use SWRL rules to automatically generate implicit relationships to constrain the teaching rationality of the knowledge graph.

[0074] For example, if knowledge point A requires capability L2 and B is a prerequisite for A, then B requires at least capability L1.

[0075] After obtaining the above-mentioned empowered entities, explicit and implicit dependencies, we can construct a syllabus knowledge graph based on the four-layer ontology model.

[0076] Optionally, in this embodiment, the test parameters input by the test-setting teacher are correlated with the syllabus knowledge graph to obtain test-setting rules that comply with the syllabus, and a preliminary test question set is generated based on the rules to ensure that the scope of the test question set complies with the scope of the syllabus.

[0077] Specifically, the association analysis between the test-setting parameters and the syllabus knowledge graph involves weight matching and automatic correction of the test-setting parameters and the syllabus knowledge graph.

[0078] For example:

[0079] Receive the test parameters set by the administrator:

[0080] Scope of knowledge points (e.g., mathematics disciplines, chapters on derivatives, etc.);

[0081] Question type ratio (multiple choice questions 30%, fill-in-the-blank questions 20%, subjective questions 50%);

[0082] Difficulty distribution (easy 30%, medium 50%, hard 20%);

[0083] Total examination score and examination duration.

[0084] The system automatically parses the test parameters set by the administrator, extracts the target knowledge points, and stores them in a structured, computable data model. The system traverses the syllabus knowledge graph and calculates the Euclidean distance deviation between the target knowledge point weights in the test parameters (e.g., "derivative" accounts for 30%) and the preset weights in the knowledge graph (e.g., the standard weight of "derivative" is 25%). The preset weights in the knowledge graph are weighted by assigning weights to knowledge points based on the test frequency of the question bank data.

[0085] A dynamic tolerance threshold (e.g., ±10%) is preset. If the deviation exceeds this threshold, a weight adjustment suggestion module is triggered. This module combines historical high-frequency test data and teaching requirements to generate adjustment plans. For example, the weight of "derivative" can be reduced to 25% and the weight of "integral" can be increased.

[0086] If the deviation is within the limit, the test paper setting rules are generated directly based on the test paper setting parameters input by the teacher.

[0087] Specifically, the correlation analysis between the test-setting parameters and the syllabus knowledge graph also focuses on the coverage completeness detection of the core knowledge points by the test-setting parameters.

[0088] Based on the directed acyclic graph (DAG) structure of the knowledge graph, the system identifies core knowledge points and counts the number of nodes covered by the test-taking parameters. The coverage ratio is calculated as follows: Coverage ratio = Covered core points / Total core points.

[0089] If the coverage ratio is lower than the threshold, it is judged as insufficient coverage. A graph embedding algorithm (such as Node2Vec) is used to calculate the semantic relevance between knowledge points and generate a list of completion recommendations. The completion logic supports multi-strategy fusion, including teaching logic relevance, examination frequency relevance, etc. For example:

[0090] Teaching logic connection: If "trigonometric functions" are not covered, related "unit circle" or "sine theorem" questions are recommended.

[0091] Examination frequency correlation: Priority is given to high-frequency examination points, and knowledge points whose appearance rate in the real questions in the past five years is >60%.

[0092] Uncovered knowledge points are marked with red warning signs, and a topology gap map is generated on the administrator interface to visually display the missing parts.

[0093] Through the correlation analysis between the above-mentioned test parameters and the syllabus knowledge graph, the first step of "self-inspection of compliance with the syllabus" is achieved, ensuring that the test weights meet the teaching standards, while ensuring that no knowledge points are missed, and improving the test quality from the source of the test rules.

[0094] Step S102: extracting question stem keywords from the primary screening test set based on entity extraction technology, performing semantic matching between the question stem keywords and the syllabus knowledge graph, determining corresponding semantic matching degrees, comparing the semantic matching degrees with a preset matching threshold, constructing a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, performing parameterized screening on the secondary screening test set based on item response theory, generating a final test paper based on the test questions that have undergone the parameterized screening, and sending the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain.

[0095] Optionally, in this embodiment, this step is the second step in implementing the "syllabus compliance self-check" for intelligent test composition, namely, a secondary compliance check is performed on the test question set that has passed the initial screening. It is understandable that the test question set that was initially screened is formed based on the test question composition rules that passed the first step of the compliance self-check. However, the test question composition rules are still too broad and can only ensure that the overall distribution probability of the exercises conforms to the syllabus, but cannot guarantee that the exercises meet the syllabus requirements in terms of the quality of the knowledge points. Therefore, a second step of compliance self-check is required to conduct a secondary screening of the initially screened exercise set to ensure the scientific nature and effectiveness of the test paper.

[0096] Optionally, in this embodiment, a self-check for syllabus compliance is performed based on the semantic matching of the stems of the pre-screened test questions. Specifically, this technical solution uses a pre-trained BERT model to perform deep semantic vectorization of the test question stem keywords. Combined with the node vectors of the syllabus knowledge graph, high-precision semantic matching is achieved through cosine similarity calculation. This effectively addresses the semantic gap in traditional keyword matching and ensures a strong correlation between test questions and teaching objectives.

[0097] First, BERT is based on the Transformer architecture and uses a bidirectional attention mechanism to capture the contextual semantics of the key words in the question stem. For example, for the keyword "photosynthesis", BERT not only extracts its literal meaning, but also generates a dynamic vector representation based on the context of the question stem (such as "plant energy conversion" and "chloroplast participation"), avoiding the static representation defects of traditional TF-IDF or Word2Vec. BERT's CLS token output (768 / 1024-dimensional vector) serves as a semantic feature, covering grammatical, logical, and domain knowledge information. Fine-tuning is used to adapt it to the terminology in the education field and improve vectorization accuracy.

[0098] Secondly, the nodes of the syllabus knowledge graph are pre-vectored using a similar BERT model to form a standardized semantic space. According to the syllabus requirements, core knowledge nodes are assigned higher vector weights (such as frequently tested points), and their influence is amplified through weighting coefficients when similarity is calculated.

[0099] Next, the cosine similarity between the question stem keyword vector and the knowledge graph node vector is calculated (range [-1, 1]). Positive values ​​(semantically related) are filtered and normalized to the range [0, 1]. A preset matching threshold is dynamically adjusted based on the question type. This semantic matching ensures a strong relevance between the question and the teaching objectives.

[0100] Optionally, in this embodiment, after the above semantic matching, the remaining test paper and question set will be subjected to a final "syllabus" compliance check according to the Item Response Theory (IRT), and a test paper will be generated.

[0101] Specifically, using item response theory, a three-parameter logic model is constructed to quantify the core characteristics of test questions, including the discrimination parameter (a), the difficulty parameter (b), and the guessing parameter (c). This model accurately characterizes the measurement characteristics of test questions, ensuring the scientific nature and validity of the test paper.

[0102] Among them, the discrimination parameter (a) measures the ability of the test questions to distinguish candidates with different ability levels, and its value range is usually 0 to 2.

[0103] High discrimination (a>1.0): The test questions can effectively distinguish between high-ability and low-ability candidates and are suitable for selective examinations.

[0104] Low discrimination (a<0.5): The test questions have weak discrimination ability, which may be caused by vague expression or insufficient coverage of knowledge points, and need to be optimized or eliminated.

[0105] Difficulty parameter (b): Indicates the absolute difficulty of the test question, and the value range is usually -3 (very easy) to +3 (extremely difficult).

[0106] Medium difficulty (b≈0) is suitable as the main body of the test paper to ensure that most candidates can demonstrate their true level.

[0107] Extremely difficult questions (|b|>2) should be used with caution to avoid biasing the score distribution.

[0108] Guess parameter (c): reflects the probability of the examinee guessing the correct answer at random, with a value range of 0 to 0.5 (e.g. the theoretical guess probability for a four-choice multiple-choice question is 0.25).

[0109] An excessively high guessing parameter (c>0.3) will reduce the reliability of the test questions, and the option design needs to be adjusted or changed to a non-multiple-choice question type.

[0110] The test question optimization process based on the three-parameter logic model includes three stages: feature calculation, constraint screening, and test paper generation.

[0111] First, feature calculation involves the automated calibration of test parameters.

[0112] Use historical test-taker data (e.g., 10,000+ samples) to fit an IRT model, and calculate the a, b, and c parameters for each question using maximum likelihood estimation. For new questions, use a small sample size pretest (e.g., 200 test-takers) combined with the EM algorithm to initially estimate the parameters, and then gradually refine these parameters as the number of tests increases.

[0113] Second, constraint screening involves multi-objective optimization strategies.

[0114] Filter test questions by pre-set constraints to ensure that the test paper meets the test objectives:

[0115] Discrimination constraint: retain questions with a∈[0.5,2.0] and eliminate questions with insufficient discrimination ability.

[0116] Difficulty gradient control: Allocate questions of different difficulty levels in proportion (such as 20% easy, 60% medium, and 20% difficult) so that the difficulty curve of the test paper conforms to the normal distribution.

[0117] Guess probability suppression: Eliminate questions with c>0.3 and add interference item intensity analysis to high-risk question types (such as multiple-choice questions).

[0118] Finally, test paper generation involves IRT-driven test paper generation algorithms

[0119] Question combination optimization: Based on the genetic algorithm, a question combination that meets the following objectives is selected from the second-screened question set:

[0120] Maximize the overall information content (i.e. the test information function TIF has the highest peak value in the target ability range).

[0121] Minimize the repetition rate of knowledge points between questions (avoid overlapping test points through semantic similarity detection).

[0122] Anomaly detection: Monte Carlo simulation is performed on the final test paper to predict the score distribution. If skewness is found (such as excessive concentration in high score segments), the test parameters are readjusted.

[0123] It is understandable that the intelligent screening and optimization of test questions are achieved through the three-parameter logic model, combined with the IRT theory, to ensure that the generated test paper has high discrimination, reasonable difficulty distribution and low guessing interference.

[0124] Next, in order to conduct the exam completely and deal with the problem of network interruption during the exam, this application implements the terminal data offline continuation function based on the two-way linkage between the server and the exam end.

[0125] Optionally, in this embodiment, the candidate's terminal continuously monitors the test network status. If the network is interrupted, the candidate's operation data is stored in a local encrypted container in the form of a one-way hash chain. This step involves the operation of the candidate's terminal when the network status is interrupted.

[0126] Specifically, the candidate's end uses a heartbeat detection mechanism (sending an ICMP request every 3 seconds) to continuously monitor the network connection status. If a network interruption is detected, the system immediately captures the current candidate's real-time operation data, including:

[0127] Operation logs (such as page turning, answer modification, and abnormal behavior events);

[0128] Answer data (structured answer content and binary handwriting data);

[0129] High-precision timestamp (based on local clock synchronized by NTP, error ≤ 50ms)

[0130] The AES-256-GCM algorithm is used to encrypt real-time operation data and generate a data integrity check tag (GCM Tag). The encrypted data is packaged into independent data blocks at preset time intervals (such as every 30 seconds). Each data block contains:

[0131] Header information: data block ID, timestamp range, encryption algorithm identifier;

[0132] Payload data: compressed encrypted operation log and answer data;

[0133] Trailing checksum: CRC-32 checksum, used to detect transmission corruption.

[0134] A SHA-3-256 hash value is generated for each data block, and the hash value of the previous data block is embedded in the current block header, forming a one-way hash chain. This ensures that tampering with any data block will invalidate the hash values ​​of all subsequent blocks. The dual binding of timestamps and hash values ​​prevents data replay attacks. The hash chain is stored in a local encrypted container, and the key is protected by a hardware-level security module.

[0135] Through the above steps, the candidate can ensure that the candidate's operation information will not be lost when the network is interrupted, and at the same time ensure the uniqueness and non-tamperability of the operation information.

[0136] Step S103: When the network is reconnected, the one-way hash chain sent by the examinee is received, and it is determined whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the examination.

[0137] Optionally, in this embodiment, in order to complete the exam and deal with the problem of network interruption during the exam, this application implements the terminal data offline test continuation function based on the two-way linkage between the server and the test end. This step involves the operation of the server after the network status is connected.

[0138] It is understandable that the two-way linkage between the server and the examination end is achieved based on two-way fingerprint verification technology.

[0139] Specifically, during a network outage, the candidate's terminal caches their answer data and generates a one-way hash chain to ensure data integrity and tamper-proofing. When the network is restored, the candidate's terminal uploads the hash chain to the server, which decrypts and verifies the received data to ensure it has not been tampered with or lost during transmission and storage.

[0140] More specifically, after receiving the data, the server first performs a data packet integrity check (such as verifying the size of the data block and whether the format complies with the protocol). If an abnormality is found (such as data block damage or incorrect format), the candidate is required to retransmit.

[0141] After receiving the data packet, the server decrypts each data block using a pre-shared key (e.g., an AES-256 key generated based on the candidate ID and device fingerprint). After decryption, the server extracts the header hash value (i.e., the hash value of the previous data block) and the current block content in preparation for continuity verification.

[0142] More specifically, the hash chain continuity check is performed. The header hash value of the first data block (without a predecessor block) should be the preset initial value (such as an all-zero hash), otherwise the chain is determined to have been tampered with. Subsequent block checks calculate the measured hash value for the content of the current block and compare this measured hash value with the header hash value of the next data block. If they match, it means that the data has not been tampered with, and the next block is checked. If they do not match, the hash chain is determined to be broken, synchronization is stopped, and an anomaly is marked. During this process, an additional check is performed to see if the timestamp of the data block is monotonically increasing to prevent replay attacks.

[0143] Next, if all data blocks pass verification, the server reorganizes the data in chronological order and writes it to the database. If a hash chain break is found, the server records the abnormal location and takes the following measures:

[0144] Partial synchronization: Only the valid data before the breakpoint is synchronized, and the subsequent data is required to be resubmitted by the candidate.

[0145] Alarm and manual review: Send an exception report to the administrator, who will manually decide whether to allow re-upload or initiate the exam exception process.

[0146] In this way, the system can ensure the integrity and traceability of offline examination data while taking into account verification efficiency and security.

[0147] This example demonstrates how this embodiment improves the efficiency and accuracy of the examination system based on intelligent test paper generation and offline re-examination.

[0148] From the above description, it can be seen that the intelligent examination method provided in the embodiment of the present application can receive the test paper parameters of the administrator, associate and verify the test paper parameters with the set teaching syllabus knowledge graph to obtain the test paper rules, and screen the test questions from the preset teaching question bank according to the test paper rules to determine the corresponding primary screening test question set; semantically match the question stem keywords of the primary screening test question set with the teaching syllabus knowledge graph to construct a secondary screening test question set, and perform parameterized screening of the secondary screening test question set according to the item response theory to generate the final test paper, and send the test paper to the candidate end. The candidate end continuously monitors the test network status. If the network is interrupted, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test, thereby improving the efficiency and accuracy of the examination system.

[0149] In one embodiment of the intelligent examination method of the present application, see Figure 2 , and can also include the following:

[0150] Step S201: semantically segmenting the preset teaching syllabus document, determining corresponding knowledge point entities, establishing a knowledge point alias library, uniformly mapping the knowledge point entities according to the knowledge point alias library, and determining corresponding unified entity nodes and dependency relationships between entity nodes;

[0151] Step S202: constructing a multi-layer ontology structure layer based on the unified entity node, assigning dynamic weights to the entity nodes of each ontology structure layer based on the examination frequency of the teaching question bank, and determining corresponding weighted entity nodes;

[0152] Step S203: performing rule constraints on the weighted entity nodes according to a preset OWL rule engine, and determining the corresponding set syllabus knowledge graph according to the indirect dependency relationships obtained after the rule constraints, the dependency relationships between the entity nodes, and the weighted entity nodes.

[0153] Optionally, in this embodiment, this step is the process of constructing the syllabus knowledge graph.

[0154] Specifically, first, the syllabus knowledge point entities and dependency relationships are extracted.

[0155] Specifically, the latest teaching syllabus document is parsed into a three-level structure of "chapter-section-knowledge point" using a Transformer-based hierarchical segmentation model to extract knowledge point entities.

[0156] Chapter division is divided into logical units through title style recognition (such as "Chapter 3 Functions") and semantic coherence analysis (BERT sentence embedding cosine similarity > 0.9).

[0157] Knowledge point extraction uses the pre-trained ALBERT model combined with domain-adaptive fine-tuning to identify knowledge point entities, such as extracting "quadratic function" and "image properties" from "understanding the image properties of quadratic functions."

[0158] Specifically, a dynamically updated knowledge point alias library is constructed to solve the problem of multiple representations of the same knowledge point entity.

[0159] The semantic similarity of candidate aliases is calculated based on contrastive learning (SimCSE) (e.g., "Pythagorean Theorem" ≈ "Pythagorean Theorem"), and a mapping is established when the threshold is greater than 0.85.

[0160] Specifically, based on dependency syntax analysis and rule template matching, explicit dependencies are identified and explicit dependency relationships between knowledge point entities are obtained.

[0161] Secondly, the extracted knowledge point entities and dependency relationships are processed, entity weight values ​​are assigned to the knowledge point entities, and hidden dependency relationships are mined for the dependency relationships.

[0162] Specifically, we use a multi-layer ontology structure and dynamic weighting to transform discrete knowledge points into a multi-layer network with teaching semantics. We design a four-layer ontology model as follows:

[0163] The subject layer ("mathematics") uses ISO 15926 standard coding to ensure cross-system compatibility;

[0164] Module layer ("algebra", "geometry");

[0165] Knowledge point layer ("quadratic function" and "equation");

[0166] Ability dimension layer (“quantitative level of logical reasoning”);

[0167] Hypergraph is used to represent cross-layer associations, allowing a single node to belong to multiple parents ("vector" belongs to both algebraic and geometric modules).

[0168] Specifically, knowledge point entities are weighted based on question bank data. The frequency of knowledge points in exam questions over the past five years is counted and normalized to a value between 0 and 1. A time decay factor (half-life of 1 year) is used to ensure the timeliness of the weights, and weight values ​​are dynamically assigned to knowledge point entities.

[0169] Specifically, we mine implicit dependencies for dependencies. We define a customized rule set for the teaching domain, build a rule engine, declare pre-requisites as transitive attributes based on the rule engine, automatically deduce indirect dependencies, and use SWRL rules to automatically generate implicit relationships to constrain the teaching rationality of the knowledge graph.

[0170] For example, if knowledge point A requires capability L2 and B is a prerequisite for A, then B requires at least capability L1.

[0171] After obtaining the above-mentioned empowered entities, explicit and implicit dependencies, we can construct a syllabus knowledge graph based on the four-layer ontology model.

[0172] Through step S203, this embodiment obtains the syllabus knowledge graph, laying a solid data foundation for subsequent compliance self-inspection based on the knowledge graph.

[0173] In one embodiment of the intelligent examination method of the present application, see Figure 3 , and can also include the following:

[0174] Step S301: parsing the knowledge point distribution weights in the test test parameters, matching the knowledge point distribution weights with the node weights in the syllabus knowledge graph, and determining the corresponding knowledge point distribution weight deviations;

[0175] Step S302: determining whether the knowledge point distribution weight deviation exceeds a preset deviation threshold; if not, determining a corresponding test composition rule according to the test composition parameters.

[0176] Optionally, in this embodiment, this step is a test paper composition rule determination process.

[0177] Specifically, the test parameters entered by the test-setting teacher are correlated with the syllabus knowledge graph to obtain test-setting rules that conform to the syllabus, and a preliminary test question set is generated based on the rules to ensure that the scope of the test question set conforms to the scope of the syllabus.

[0178] Specifically, the association analysis between the test-setting parameters and the syllabus knowledge graph involves weight matching and automatic correction of the test-setting parameters and the syllabus knowledge graph.

[0179] For example:

[0180] Receive the test parameters set by the administrator:

[0181] Scope of knowledge points (e.g., mathematics disciplines, chapters on derivatives, etc.);

[0182] Question type ratio (multiple choice questions 30%, fill-in-the-blank questions 20%, subjective questions 50%);

[0183] Difficulty distribution (easy 30%, medium 50%, hard 20%);

[0184] Total examination score and examination duration.

[0185] The system automatically parses the test parameters set by the administrator, extracts the target knowledge points, and stores them in a structured, computable data model. The system traverses the syllabus knowledge graph and calculates the Euclidean distance deviation between the target knowledge point weights in the test parameters (e.g., "derivative" accounts for 30%) and the preset weights in the knowledge graph (e.g., the standard weight of "derivative" is 25%). The preset weights in the knowledge graph are weighted by assigning weights to knowledge points based on the test frequency of the question bank data.

[0186] A dynamic tolerance threshold (e.g., ±10%) is preset. If the deviation exceeds this threshold, a weight adjustment suggestion module is triggered. This module combines historical high-frequency test data and teaching requirements to generate adjustment plans. For example, the weight of "derivative" can be reduced to 25% and the weight of "integral" can be increased.

[0187] If the deviation is within the limit, the test paper setting rules are generated directly based on the test paper setting parameters input by the teacher.

[0188] Specifically, the correlation analysis between the test-setting parameters and the syllabus knowledge graph also focuses on the coverage completeness detection of the core knowledge points by the test-setting parameters.

[0189] Based on the directed acyclic graph (DAG) structure of the knowledge graph, the system identifies core knowledge points and counts the number of nodes covered by the test-taking parameters. The coverage ratio is calculated as follows: Coverage ratio = Covered core points / Total core points.

[0190] If the coverage ratio is lower than the threshold, it is judged as insufficient coverage. A graph embedding algorithm (such as Node2Vec) is used to calculate the semantic relevance between knowledge points and generate a list of completion recommendations. The completion logic supports multi-strategy fusion, including teaching logic relevance, examination frequency relevance, etc. For example:

[0191] Teaching logic connection: If "trigonometric functions" are not covered, related "unit circle" or "sine theorem" questions are recommended.

[0192] Examination frequency correlation: Priority is given to high-frequency examination points, and knowledge points whose appearance rate in the real questions in the past five years is >60%.

[0193] Uncovered knowledge points are marked with red warning signs, and a topology gap map is generated on the administrator interface to visually display the missing parts.

[0194] Through the correlation analysis between the above-mentioned test parameters and the syllabus knowledge graph, the first step of "self-inspection of compliance with the syllabus" is achieved, ensuring that the test weights meet the teaching standards, while ensuring that no knowledge points are missed, and improving the test quality from the source of the test rules.

[0195] Through step S302, this embodiment realizes the correlation analysis between the test paper setting parameters and the teaching syllabus, obtains the test paper setting rules that conform to the content of the teaching syllabus, and lays the foundation for the subsequent generation of the preliminary screening test question set.

[0196] In one embodiment of the intelligent examination method of the present application, see Figure 4 , and can also include the following:

[0197] Step S401: vectorizing the question stem keywords based on a preset BERT model to determine corresponding semantic features;

[0198] Step S402: Calculate the cosine similarity based on the semantic features and the node vectors of the syllabus knowledge graph to determine the corresponding semantic matching degree.

[0199] Optionally, in this embodiment, this step involves the second step of "self-checking compliance with the syllabus".

[0200] Specifically, the system conducts a self-check of syllabus compliance based on the semantic matching of the stems of pre-screened test questions. This technical solution uses a pre-trained BERT model to perform deep semantic vectorization of test stem keywords. This, combined with the node vectors of the syllabus knowledge graph, achieves high-precision semantic matching through cosine similarity calculation. This effectively addresses the semantic gap in traditional keyword matching and ensures a strong relevance between test questions and teaching objectives.

[0201] First, BERT is based on the Transformer architecture and uses a bidirectional attention mechanism to capture the contextual semantics of the key words in the question stem. For example, for the keyword "photosynthesis", BERT not only extracts its literal meaning, but also generates a dynamic vector representation based on the context of the question stem (such as "plant energy conversion" and "chloroplast participation"), avoiding the static representation defects of traditional TF-IDF or Word2Vec. BERT's CLS token output (768 / 1024-dimensional vector) serves as a semantic feature, covering grammatical, logical, and domain knowledge information. Fine-tuning is used to adapt it to the terminology in the education field and improve vectorization accuracy.

[0202] Secondly, the nodes of the syllabus knowledge graph are pre-vectored using a similar BERT model to form a standardized semantic space. According to the syllabus requirements, core knowledge nodes are assigned higher vector weights (such as frequently tested points), and their influence is amplified through weighting coefficients when similarity is calculated.

[0203] Next, the cosine similarity between the question stem keyword vector and the knowledge graph node vector is calculated (range [-1, 1]). Positive values ​​(semantically related) are filtered and normalized to the range [0, 1]. A preset matching threshold is dynamically adjusted based on the question type. This semantic matching ensures a strong relevance between the question and the teaching objectives.

[0204] Through step S402, this embodiment successfully screens out a set of test questions that are highly relevant to the teaching objectives through semantic similarity, laying a foundation for subsequent generation of test papers that meet the teaching standards.

[0205] In one embodiment of the intelligent examination method of the present application, see Figure 5 , and can also include the following:

[0206] Step S501: Constructing a three-parameter logic model, wherein the three-parameter logic model includes a discrimination parameter, a difficulty parameter, and a guessing parameter, wherein the discrimination parameter is used to characterize the ability of the test question to distinguish between examinees of different ability levels, the difficulty parameter is used to characterize the absolute difficulty level of the test question, and the guessing parameter is used to characterize the probability of the examinee guessing the test question correctly at random;

[0207] Step S502: quantify the core features of the second-screened test question set according to the three-parameter logical model, determine the multi-parameter features of the corresponding second-screened test question set, screen the multi-parameter features based on preset constraints, and generate a final test paper based on the screened test questions.

[0208] Optionally, in this embodiment, this step involves the second step of "self-checking compliance with the syllabus".

[0209] Specifically, the test paper question set left after the semantic matching in step S602 is subjected to a final “syllabus” compliance check according to the Item Response Theory (IRT), and a test paper is generated.

[0210] Specifically, using item response theory, a three-parameter logic model is constructed to quantify the core characteristics of test questions, including the discrimination parameter (a), the difficulty parameter (b), and the guessing parameter (c). This model accurately characterizes the measurement characteristics of test questions, ensuring the scientific nature and validity of the test paper.

[0211] Among them, the discrimination parameter (a) measures the ability of the test questions to distinguish candidates with different ability levels, and its value range is usually 0 to 2.

[0212] High discrimination (a>1.0): The test questions can effectively distinguish between high-ability and low-ability candidates and are suitable for selective examinations.

[0213] Low discrimination (a<0.5): The test questions have weak discrimination ability, which may be caused by vague expression or insufficient coverage of knowledge points, and need to be optimized or eliminated.

[0214] Difficulty parameter (b): Indicates the absolute difficulty of the test question, and the value range is usually -3 (very easy) to +3 (extremely difficult).

[0215] Medium difficulty (b≈0) is suitable as the main body of the test paper to ensure that most candidates can demonstrate their true level.

[0216] Extremely difficult questions (|b|>2) should be used with caution to avoid biasing the score distribution.

[0217] Guess parameter (c): reflects the probability of the examinee guessing the correct answer at random, with a value range of 0 to 0.5 (e.g. the theoretical guess probability for a four-choice multiple-choice question is 0.25).

[0218] An excessively high guessing parameter (c>0.3) will reduce the reliability of the test questions, and the option design needs to be adjusted or changed to a non-multiple-choice question type.

[0219] The test question optimization process based on the three-parameter logic model includes three stages: feature calculation, constraint screening, and test paper generation.

[0220] First, feature calculation involves the automated calibration of test parameters.

[0221] Use historical test-taker data (e.g., 10,000+ samples) to fit an IRT model, and calculate the a, b, and c parameters for each question using maximum likelihood estimation. For new questions, use a small sample size pretest (e.g., 200 test-takers) combined with the EM algorithm to initially estimate the parameters, and then gradually refine these parameters as the number of tests increases.

[0222] Second, constraint screening involves multi-objective optimization strategies.

[0223] Filter test questions by pre-set constraints to ensure that the test paper meets the test objectives:

[0224] Discrimination constraint: retain questions with a∈[0.5,2.0] and eliminate questions with insufficient discrimination ability.

[0225] Difficulty gradient control: Allocate questions of different difficulty levels in proportion (such as 20% easy, 60% medium, and 20% difficult) so that the difficulty curve of the test paper conforms to the normal distribution.

[0226] Guess probability suppression: Eliminate questions with c>0.3 and add interference item intensity analysis to high-risk question types (such as multiple-choice questions).

[0227] Finally, test paper generation involves IRT-driven test paper generation algorithms

[0228] Question combination optimization: Based on the genetic algorithm, a question combination that meets the following objectives is selected from the second-screened question set:

[0229] Maximize the overall information content (i.e. the test information function TIF has the highest peak value in the target ability range).

[0230] Minimize the repetition rate of knowledge points between questions (avoid overlapping test points through semantic similarity detection).

[0231] Anomaly detection: Monte Carlo simulation is performed on the final test paper to predict the score distribution. If skewness is found (such as excessive concentration in high score segments), the test parameters are readjusted.

[0232] Through step S502, this embodiment successfully realizes intelligent screening and optimization of test questions through the three-parameter logic model, and combines IRT theory to ensure that the generated test paper has high discrimination, reasonable difficulty distribution and low guessing interference.

[0233] In one embodiment of the intelligent examination method of the present application, see Figure 6 , and can also include the following:

[0234] Step S601: The examinee's terminal continuously monitors the examination network status. If the network is interrupted, the examinee's real-time operation data is obtained, the real-time operation data is encrypted using a preset encryption algorithm, and the encrypted data is packaged into multiple independent data blocks based on preset time intervals. The real-time operation data of the examinee includes the examinee's operation log, answer data, and timestamp.

[0235] Step S602: Generate a corresponding hash value for each independent data block, embed the hash value of the previous data block into the header of the current block, determine the corresponding one-way hash chain, and store the one-way hash chain in a local encryption container.

[0236] Optionally, in this embodiment, in order to complete the exam and deal with the problem of network interruption during the exam, this technical solution implements the terminal data offline test continuation function based on the two-way linkage between the server and the exam end. This step involves the operation of the candidate end.

[0237] Specifically, the candidate's end uses a heartbeat detection mechanism (such as sending an ICMP request every 3 seconds) to continuously monitor the network connection status. If a network interruption is detected, the system immediately captures the current candidate's real-time operation data, including:

[0238] Operation logs (such as page turning, answer modification, and abnormal behavior events);

[0239] Answer data (structured answer content and binary handwriting data);

[0240] High-precision timestamp (based on local clock synchronized by NTP, error ≤ 50ms)

[0241] The AES-256-GCM algorithm is used to encrypt real-time operation data and generate a data integrity check tag (GCM Tag). The encrypted data is packaged into independent data blocks at preset time intervals (such as every 30 seconds). Each data block contains:

[0242] Header information: data block ID, timestamp range, encryption algorithm identifier;

[0243] Payload data: compressed encrypted operation log and answer data;

[0244] Trailing checksum: CRC-32 checksum, used to detect transmission corruption.

[0245] A SHA-3-256 hash value is generated for each data block, and the hash value of the previous data block is embedded in the current block header, forming a one-way hash chain. This ensures that tampering with any data block will invalidate the hash values ​​of all subsequent blocks. The dual binding of timestamps and hash values ​​prevents data replay attacks. The hash chain is stored in a local encrypted container, and the key is protected by a hardware-level security module.

[0246] Through step S602, this embodiment successfully ensures that the examinee's operation information will not be lost when the network is interrupted, and at the same time ensures the uniqueness and non-tamperability of the operation information, laying the foundation for the server to recover data when the network is restored.

[0247] In one embodiment of the intelligent examination method of the present application, see Figure 7 , and can also include the following:

[0248] Step S701: receiving the one-way hash chain sent by the candidate, decrypting the one-way hash chain block by block, and verifying the continuity;

[0249] Step S702: If the header hash value of the current block matches the measured hash value of the previous block, it is determined that the hash chain is not broken, and data synchronization is performed according to the one-way hash chain.

[0250] Optionally, in this embodiment, in order to complete the exam and deal with the problem of network interruption during the exam, this technical solution implements the terminal data offline test continuation function based on the two-way linkage between the server and the exam end. This step involves the operation of the server end.

[0251] Specifically, during a network outage, the candidate's terminal caches their answer data and generates a one-way hash chain to ensure data integrity and tamper-proofing. When the network is restored, the candidate's terminal uploads the hash chain to the server, which decrypts and verifies the received data to ensure it has not been tampered with or lost during transmission and storage.

[0252] More specifically, after receiving the data, the server first performs a data packet integrity check (such as verifying the size of the data block and whether the format complies with the protocol). If an abnormality is found (such as data block damage or incorrect format), the candidate is required to retransmit.

[0253] After receiving the data packet, the server decrypts each data block using a pre-shared key (e.g., an AES-256 key generated based on the candidate ID and device fingerprint). After decryption, the server extracts the header hash value (i.e., the hash value of the previous data block) and the current block content in preparation for continuity verification.

[0254] More specifically, the hash chain continuity check is performed. The header hash value of the first data block (without a predecessor block) should be the preset initial value (such as an all-zero hash), otherwise the chain is determined to have been tampered with. Subsequent block checks calculate the measured hash value for the content of the current block and compare this measured hash value with the header hash value of the next data block. If they match, it means that the data has not been tampered with, and the next block is checked. If they do not match, the hash chain is determined to be broken, synchronization is stopped, and an anomaly is marked. During this process, an additional check is performed to see if the timestamp of the data block is monotonically increasing to prevent replay attacks.

[0255] Next, if all data blocks pass verification, the server reorganizes the data in chronological order and writes it to the database. If a hash chain break is found, the server records the abnormal location and takes the following measures:

[0256] Partial synchronization: Only the valid data before the breakpoint is synchronized, and the subsequent data is required to be resubmitted by the candidate.

[0257] Alarm and manual review: Send an exception report to the administrator, who will manually decide whether to allow re-upload or initiate the exam exception process.

[0258] Through step S702, this embodiment successfully ensures the integrity and traceability of offline examination data while taking into account verification efficiency and security.

[0259] In order to improve the efficiency and accuracy of the examination system, the present application provides an embodiment of an intelligent examination device for implementing all or part of the contents of the intelligent examination method, see Figure 8 The intelligent examination device specifically includes the following contents:

[0260] The test paper and question set preliminary screening module 10 is used to receive the test paper setting parameters from the administrator, associate and verify the test paper setting parameters with the set teaching syllabus knowledge map, determine the corresponding test paper setting rules based on the test paper setting parameters that have undergone the association verification, and screen test questions from the preset teaching question bank according to the test paper setting rules to determine the corresponding preliminary screening test question set;

[0261] The test paper and question set determination module 20 is configured to extract question stem keywords from the primary screening test set based on entity extraction technology, perform semantic matching between the question stem keywords and the syllabus knowledge graph, determine the corresponding semantic matching degree, compare the semantic matching degree with a preset matching threshold, construct a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, perform parameterized screening on the secondary screening test set based on item response theory, generate a final test paper based on the test questions that have undergone the parameterized screening, and send the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain.

[0262] The offline resuming test module 30 is used to receive the one-way hash chain sent by the examinee when reconnecting to the network, determine whether the one-way hash chain is broken, and if not, synchronize data according to the one-way hash chain to complete the test.

[0263] From the above description, it can be seen that the intelligent examination device provided in the embodiment of the present application can receive the administrator's test paper parameters, associate the test paper parameters with the set teaching syllabus knowledge graph to obtain test paper rules, and screen test questions from the preset teaching question bank according to the test paper rules to determine the corresponding primary screening test question set; semantically match the question stem keywords of the primary screening test question set with the teaching syllabus knowledge graph to construct a secondary screening test question set, and perform parameterized screening of the secondary screening test question set according to the item response theory to generate the final test paper, and send the test paper to the candidate end. The candidate end continuously monitors the test network status. If the network is interrupted, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test, thereby improving the efficiency and accuracy of the examination system.

[0264] From a hardware perspective, in order to improve the efficiency and accuracy of the examination system, the present application provides an embodiment of an electronic device for implementing all or part of the contents of the intelligent examination method. The electronic device specifically includes the following contents:

[0265] A processor, a memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to transmit information between the intelligent examination method and related devices such as the core business system, user terminals, and related databases; the logic controller can be a desktop computer, a tablet computer, a mobile terminal, etc., but this embodiment is not limited thereto. In this embodiment, the logic controller can be implemented with reference to the embodiments of the intelligent examination method and the embodiments of the intelligent examination method in the embodiment, and their contents are incorporated herein, and repeated parts are not repeated.

[0266] It is understandable that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0267] In practical applications, portions of the intelligent examination method may be executed on the electronic device side as described above, or all operations may be performed on the client device. The specific selection may be based on the processing capabilities of the client device and the limitations of the user's usage scenario. This application does not impose any restrictions on this. If all operations are performed on the client device, the client device may also include a processor.

[0268] The client device may include a communication module (i.e., a communication unit) that can establish a communication connection with a remote server to implement data transmission with the server. The server may include a server on the task scheduling center side, and in other implementation scenarios, may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0269] Figure 9 Schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Figure 9 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that the Figure 9 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0270] In one embodiment, the intelligent examination method function may be integrated into the central processing unit 9100. The central processing unit 9100 may be configured to perform the following control:

[0271] Step S101: receiving the test parameters from the administrator, correlating and verifying the test parameters with the set syllabus knowledge graph, determining corresponding test rules based on the correlatively verified test parameters, screening test questions from a preset teaching question bank based on the test rules, and determining a corresponding preliminary screening test question set;

[0272] Step S102: extracting question stem keywords from the primary screening test set based on entity extraction technology, performing semantic matching between the question stem keywords and the syllabus knowledge graph, determining corresponding semantic matching degrees, comparing the semantic matching degrees with a preset matching threshold, constructing a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, performing parameterized screening on the secondary screening test set based on item response theory, generating a final test paper based on the test questions that have undergone the parameterized screening, and sending the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain.

[0273] Step S103: When the network is reconnected, the one-way hash chain sent by the examinee is received, and it is determined whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the examination.

[0274] From the above description, it can be seen that the electronic device provided in the embodiment of the present application receives the test paper setting parameters of the administrator, associates and verifies the test paper setting parameters with the set teaching syllabus knowledge graph to obtain the test paper setting rules, and screens the test questions from the preset teaching question bank according to the test paper setting rules to determine the corresponding primary screening test question set; semantic matching is performed on the question stem keywords of the primary screening test question set and the teaching syllabus knowledge graph to construct a secondary screening test question set, and the secondary screening test question set is parameterized and screened according to the item response theory to generate the final test paper, and the test paper is sent to the candidate end. The candidate end continuously monitors the test network status. If the network is interrupted, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test, thereby improving the efficiency and accuracy of the test system.

[0275] In another embodiment, the intelligent examination method can be configured separately from the central processing unit 9100. For example, the intelligent examination method can be configured as a chip connected to the central processing unit 9100, and the functions of the intelligent examination method can be realized through the control of the central processing unit.

[0276] like Figure 9 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Figure 9 In addition, the electronic device 9600 may also include all components shown in Figure 9 For components not shown, reference may be made to the prior art.

[0277] like Figure 9 As shown, the central processing unit 9100 is sometimes also referred to as a controller or operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0278] Memory 9140 can be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It can store the aforementioned failure-related information and also store programs that execute the relevant information. The CPU 9100 can execute the programs stored in memory 9140 to implement information storage or processing.

[0279] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 may be, for example, a keypad or touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display objects such as images and text. The display may be, for example, an LCD display, but is not limited thereto.

[0280] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), or a SIM card. Alternatively, it may be a memory that retains information even when power is off, can be selectively erased, and is provided with more data. Examples of such memory are sometimes referred to as EPROMs. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 by the central processing unit 9100.

[0281] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various driver programs for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0282] The communication module 9110 is a transmitter / receiver that transmits and receives signals via the antenna 9111. The communication module 9110 is coupled to the central processing unit 9100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0283] Based on different communication technologies, multiple communication modules 9110 can be provided in the same electronic device, such as a cellular network module, a Bluetooth module, and / or a wireless local area network module. The communication module 9110 is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby implementing common telecommunication functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Furthermore, the audio processor 9130 is also coupled to the central processing unit 9100, enabling local recording via the microphone 9132 and playback of stored audio via the speaker 9131.

[0284] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all steps of the intelligent examination method in the above embodiments, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the computer program implements all steps of the intelligent examination method in the above embodiments, where the execution subject is a server or a client. For example, when the processor executes the computer program, the following steps are implemented:

[0285] Step S101: receiving the test parameters from the administrator, correlating and verifying the test parameters with the set syllabus knowledge graph, determining corresponding test rules based on the correlatively verified test parameters, screening test questions from a preset teaching question bank based on the test rules, and determining a corresponding preliminary screening test question set;

[0286] Step S102: extracting question stem keywords from the primary screening test set based on entity extraction technology, performing semantic matching between the question stem keywords and the syllabus knowledge graph, determining corresponding semantic matching degrees, comparing the semantic matching degrees with a preset matching threshold, constructing a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, performing parameterized screening on the secondary screening test set based on item response theory, generating a final test paper based on the test questions that have undergone the parameterized screening, and sending the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain.

[0287] Step S103: When the network is reconnected, the one-way hash chain sent by the examinee is received, and it is determined whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the examination.

[0288] From the above description, it can be seen that the computer-readable storage medium provided in the embodiment of the present application receives the test paper setting parameters of the administrator, associates and verifies the test paper setting parameters with the set teaching syllabus knowledge graph to obtain the test paper setting rules, and screens the test questions from the preset teaching question bank according to the test paper setting rules to determine the corresponding primary screening test question set; semantically matches the question stem keywords of the primary screening test question set with the teaching syllabus knowledge graph to construct a secondary screening test question set, and performs parameterized screening of the secondary screening test question set according to the item response theory to generate the final test paper, and sends the test paper to the candidate end. The candidate end continuously monitors the test network status. If the network is interrupted, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test, thereby improving the efficiency and accuracy of the test system.

[0289] The embodiments of the present application also provide a computer program product capable of implementing all steps of the intelligent examination method in the above embodiments, where the execution subject is a server or a client. When the computer program / instructions are executed by a processor, the steps of the intelligent examination method are implemented. For example, the computer program / instructions implement the following steps:

[0290] Step S101: receiving the test parameters from the administrator, correlating and verifying the test parameters with the set syllabus knowledge graph, determining corresponding test rules based on the correlatively verified test parameters, screening test questions from a preset teaching question bank based on the test rules, and determining a corresponding preliminary screening test question set;

[0291] Step S102: extracting question stem keywords from the primary screening test set based on entity extraction technology, performing semantic matching between the question stem keywords and the syllabus knowledge graph, determining corresponding semantic matching degrees, comparing the semantic matching degrees with a preset matching threshold, constructing a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, performing parameterized screening on the secondary screening test set based on item response theory, generating a final test paper based on the test questions that have undergone the parameterized screening, and sending the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain.

[0292] Step S103: When the network is reconnected, the one-way hash chain sent by the examinee is received, and it is determined whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the examination.

[0293] From the above description, it can be seen that the computer program product provided in the embodiment of the present application receives the test paper setting parameters of the administrator, associates and verifies the test paper setting parameters with the set teaching syllabus knowledge graph to obtain the test paper setting rules, and screens the test questions from the preset teaching question bank according to the test paper setting rules to determine the corresponding primary screening test question set; semantic matching is performed on the question stem keywords of the primary screening test question set and the teaching syllabus knowledge graph to construct a secondary screening test question set, and the secondary screening test question set is parameterized and screened according to the item response theory to generate the final test paper, and the test paper is sent to the candidate end. The candidate end continuously monitors the test network status. If the network is interrupted, the candidate operation data is stored in the local encrypted container in the form of a one-way hash chain; when the network is reconnected, the one-way hash chain sent by the candidate end is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test, thereby improving the efficiency and accuracy of the test system.

[0294] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, apparatus, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0295] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0296] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0297] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0298] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. An intelligent examination method, characterized in that: Applied to a server side, the server side is in communication connection with a preset examinee side, and the method includes: Receive the test parameters from the administrator, perform correlation verification on the test parameters and the set syllabus knowledge graph, determine the corresponding test rules based on the test parameters that have undergone correlation verification, screen test questions from the preset teaching question bank according to the test rules, and determine the corresponding preliminary screening test question set; Extracting keywords from the question stems of the initial screening test set using entity extraction technology, performing semantic matching between the question stem keywords and the syllabus knowledge graph to determine corresponding semantic matching degrees, comparing the semantic matching degrees with a preset matching threshold, constructing a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, performing parameterized screening on the secondary screening test set based on item response theory, generating a final test paper based on the test questions that have undergone the parameterized screening, and sending the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain. When the network is reconnected, the one-way hash chain sent by the candidate is received to determine whether the one-way hash chain is broken. If not, data synchronization is performed according to the one-way hash chain to complete the test.

2. The intelligent examination method according to claim 1, characterized in that: Before correlating and verifying the test test parameters with the set syllabus knowledge graph, the method includes: Perform semantic segmentation on the preset teaching syllabus document, determine the corresponding knowledge point entities, establish a knowledge point alias library, uniformly map the knowledge point entities according to the knowledge point alias library, and determine the corresponding unified entity nodes and the dependency relationships between the entity nodes; Constructing a multi-layer ontology structure layer based on the unified entity node, assigning dynamic weights to the entity nodes of each ontology structure layer based on the examination frequency of the teaching question bank, and determining corresponding weighted entity nodes; The empowered entity nodes are subject to rule constraints according to a preset OWL rule engine, and the corresponding set teaching syllabus knowledge graph is determined based on the indirect dependency relationships obtained after the rule constraints, the dependency relationships between the entity nodes, and the empowered entity nodes.

3. The intelligent examination method according to claim 2, characterized in that: The correlating and verifying the test paper setting parameters with the set syllabus knowledge graph, and determining corresponding test paper setting rules according to the test paper setting parameters that have undergone the correlating and verification, includes: Parsing the knowledge point distribution weights in the test test parameters, matching the knowledge point distribution weights with the node weights in the syllabus knowledge graph, and determining the corresponding knowledge point distribution weight deviations; Determine whether the knowledge point distribution weight deviation exceeds a preset deviation threshold; if not, determine the corresponding test paper composition rule according to the test paper composition parameters.

4. The intelligent examination method according to claim 1, characterized in that: The semantic matching of the question stem keywords with the syllabus knowledge graph to determine the corresponding semantic matching degree includes: Vectorize the question stem keywords based on the preset BERT model to determine the corresponding semantic features; Cosine similarity calculation is performed based on the semantic features and the teaching outline knowledge graph node vector to determine the corresponding semantic matching degree.

5. The intelligent examination method according to claim 1, characterized in that: The parametric screening of the second screening test question set according to the item response theory and generating a final test paper based on the parametric screened test questions include: Constructing a three-parameter logic model, the three-parameter logic model including a discrimination parameter, a difficulty parameter, and a guessing parameter, wherein the discrimination parameter is used to characterize the ability of the test question to discriminate between examinees of different ability levels, the difficulty parameter is used to characterize the absolute difficulty level of the test question, and the guessing parameter is used to characterize the probability of the examinee randomly guessing the test question correctly; The core features of the second-screened test question set are quantified according to the three-parameter logical model, the multi-parameter features of the corresponding second-screened test question set are determined, the multi-parameter features are screened based on preset constraints, and the final test paper is generated based on the screened test questions.

6. The intelligent examination method according to claim 1, characterized in that: The candidate terminal continuously monitors the test network status. If the network is interrupted, the candidate's operation data is stored in a local encrypted container in the form of a one-way hash chain, including: The examinee terminal continuously monitors the examination network status. If the network is interrupted, the examinee's real-time operation data is obtained, the real-time operation data is encrypted using a preset encryption algorithm, and the encrypted data is packaged into multiple independent data blocks based on preset time intervals, wherein the examinee's real-time operation data includes the examinee's operation log, answer data, and timestamp; Generate a corresponding hash value for each independent data block, embed the hash value of the previous data block into the header of the current block, determine the corresponding one-way hash chain, and store the one-way hash chain in a local encrypted container.

7. The intelligent examination method according to claim 6, characterized in that: The receiving the one-way hash chain sent by the examinee terminal, determining whether the one-way hash chain is broken, and if not, performing data synchronization according to the one-way hash chain, including: Receive the one-way hash chain sent by the candidate, decrypt the one-way hash chain block by block, and verify the continuity; If the header hash value of the current block matches the measured hash value of the previous block, it is determined that the hash chain is not broken, and data synchronization is performed according to the one-way hash chain.

8. An intelligent examination device, characterized in that: The device comprises: The test paper and question set preliminary screening module is used to receive the test paper parameters from the administrator, associate and verify the test paper parameters with the set teaching syllabus knowledge map, determine the corresponding test paper rules based on the test paper parameters that have undergone the association verification, and screen the test questions from the preset teaching question bank according to the test paper rules to determine the corresponding preliminary screening test question set; a test paper question set determination module, configured to extract question stem keywords from the primary screening test set based on entity extraction technology, perform semantic matching between the question stem keywords and the syllabus knowledge graph, determine the corresponding semantic matching degree, compare the semantic matching degree with a preset matching threshold, construct a secondary screening test set based on test questions corresponding to semantic matching degrees exceeding the matching threshold, perform parameterized screening on the secondary screening test set based on item response theory, generate a final test paper based on the test questions that have undergone the parameterized screening, and send the test paper to the examinee terminal so that the examinee can take the test based on the test paper. The examinee terminal continuously monitors the test network status. If the network is interrupted, the examinee's operation data is stored in a local encrypted container in the form of a one-way hash chain; The offline re-examination module is used to receive the one-way hash chain sent by the candidate when reconnecting to the network, determine whether the one-way hash chain is broken, and if not, synchronize data according to the one-way hash chain to complete the examination.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the intelligent examination method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent examination method according to any one of claims 1 to 7 are implemented.

Images