Learning achievement evaluation system for computer room based on real-name binding and intelligent collection

Abstract

The present application relates to the information technology field of knowledge imparting and skill training, and discloses a computer room learning achievement evaluation system based on real-name binding and intelligent collection, which comprises a real-name binding module, a data automatic collection module, a dynamic data matrix modeling module, a dynamic weight optimization module, a comprehensive score output module, a hybrid network adaptation module and a federal learning coordinator; the real-name binding module is used for uniquely binding the learner identity information and the terminal device hardware feature code; and the data automatic collection module is used for collecting the screen operation log, the code compilation result and the training activity space interactive behavior data of the learner end in real time. Through dynamic weight optimization, multi-dimensional data fusion modeling, intelligent network adaptation and visualization technology, the present application solves the defects of subjectivity, interaction delay and strategy vagueness of the traditional system scoring, and realizes the technical improvement of evaluation objectivity, transmission stability and decision accuracy.

Description

Technical Field

[0001] This invention relates to the field of information technology for knowledge transfer and skills training, specifically to a computer lab learning outcome evaluation system based on real-name binding and intelligent data collection. Background Technology

[0002] With the widespread application of information technology in knowledge and skills training activities, the evaluation system for training activities is gradually shifting from single-score assessments to multi-dimensional data fusion analysis. However, the differences in knowledge and skills training activities (such as programming classes emphasizing code quality while theory classes focus on test scores), the dynamic evolution of learners' performance (such as long-term learning curve fluctuations), and the complexity of network transmission environments (such as unstable signals in remote areas) place higher demands on the objectivity, adaptability, and stability of the evaluation system.

[0003] Traditional knowledge and skills evaluation systems typically employ a fixed-weighting method, where knowledge and skills assessment experts pre-determine the weighting ratios for each dimension (such as attendance, testing, and interaction) based on experience, and generate a comprehensive score through linear weighting. At the data collection level, attendance records, test scores, and the number of interactions in training activities are stored independently in different subsystems. Some systems attempt to introduce a time dimension to record performance in individual training activities. At the network transmission level, most systems rely on a single network protocol (such as Wi-Fi) to transmit interactive data. Some solutions employ a dual-link backup mechanism, but the switching logic is triggered based on a fixed threshold. Visualization modules often use static charts to display total scores or individual scores.

[0004] Existing technical solutions have some shortcomings. First, fixed weights cannot capture the differences in data distribution across different knowledge modules. For example, in programming courses, code difference is strongly negatively correlated with test scores, but traditional models do not dynamically adjust the weights of both, causing scores to deviate from true ability. This invention quantifies the statistical correlation between dimensions through covariance analysis and generates dynamic weights through constraint optimization, enabling the scoring model to adapt to the characteristics of different knowledge modules. Second, heterogeneous data stored in a dispersed manner is fragmented by differences in dimensions (e.g., attendance is binary, code difference is continuous) and time scale, making it difficult to construct a spatiotemporal evolution model of learner performance. This invention integrates multi-source data into a structured matrix across training activity spaces through normalization processing and time series matrix modeling, supporting long-term trend analysis. Furthermore, single-protocol transmission lacks adaptive switching capabilities when signal fluctuations occur, leading to delays or loss of real-time interactive data (e.g., online assessment results). This invention evaluates network status through fuzzy logic, dynamically switches protocols or enables dual-link redundant transmission, and improves packet loss resistance by combining erasure coding technology. Moreover, static charts cannot reveal the hidden performance patterns of learner groups (e.g., the "high interaction, low coding ability" group). This invention identifies cluster characteristics through cluster analysis and maps multidimensional data into spatiotemporal color codes using heatmaps, intuitively presenting individual weaknesses and group evolution patterns. Finally, traditional systems fail to effectively handle device tampering or impersonation (such as fake attendance), leading to data contamination. This invention uses real-name binding and a dynamic attenuation coefficient to suppress the influence of abnormal data, improving scoring credibility. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a computer room learning outcome evaluation system based on real-name binding and intelligent data collection. This system solves the problems in existing knowledge and skills evaluation systems, such as subjective bias in scoring, difficulty in data integration, poor real-time interaction stability, and weak targeting of knowledge and skills training strategies, caused by fixed weight allocation, fragmented heterogeneous data, rigid network transmission, and insufficient group pattern mining.

[0006] To achieve the above objectives, this invention provides the following technical solution: a computer lab learning outcome evaluation system based on real-name binding and intelligent data collection, the system comprising a real-name binding module, an automatic data collection module, a dynamic data matrix modeling module, a dynamic weight optimization module, a comprehensive score output module, a hybrid network adaptation module, and a federated learning coordinator.

[0007] The real-name binding module is used to uniquely bind learner identity information with terminal device hardware feature code, generate device identifier and store it in encryption;

[0008] The automatic data collection module is connected to the real-name binding module and is used to collect the learner's screen operation logs, code compilation results and interactive behavior data in the training activity space in real time after verifying the learner's identity based on the device identifier.

[0009] The dynamic data matrix modeling module receives data from the automatic data acquisition module, generates a reference code variant library using a generative adversarial network, compares learners’ code logic and structural defects with a discriminator, and outputs a dynamic difference score; and constructs a multi-dimensional data matrix by time series based on attendance status, test scores, code difference, and number of spatial interactions in training activities.

[0010] The dynamic weight optimization module calculates the covariance matrix based on the multidimensional data matrix and generates a scoring weight vector through principal component analysis and constraint optimization algorithms.

[0011] The comprehensive scoring output module performs weighted fusion of the multidimensional data matrix according to the scoring weight vector, outputs the learner's comprehensive score, and generates knowledge and skills cultivation strategy suggestions based on the clustering analysis results.

[0012] The hybrid network adapter module communicates with all modules and is used to coordinate the transmission priority of real-time control commands, batch data and abnormal events in a wireless or wired hybrid network to ensure data synchronization between modules.

[0013] By using a blockchain smart contract executor, the scoring rules and strategy generation logic are written into the chaincode, triggering an immutable execution process.

[0014] The federated learning coordinator, embedded in the hybrid network adaptation module, generates a federated scoring model by aggregating local model parameters across institutions, supporting a unified evaluation standard mapping under data isolation across multiple institutions.

[0015] This invention provides a computer lab learning outcome evaluation system based on real-name binding and intelligent data collection. It has the following beneficial effects:

[0016] 1. This invention uses a dynamic weight optimization module to automatically generate weights based on covariance analysis and constraint optimization algorithms, completely eliminating the subjective bias of manually setting weights. At the same time, it adjusts the weight allocation in real time according to the dynamic distribution characteristics of training activity space data, so that the scoring model can adapt to the different needs of different knowledge and skills training activity scenarios and ensure that the scoring results are more scientific and reasonable.

[0017] 2. The dynamic data matrix modeling module normalizes and organizes heterogeneous data from multiple dimensions, such as attendance, testing, interaction, and code differences, into time series, solving the integration difficulties caused by inconsistent data dimensions and missing time dimensions in traditional scoring systems, and providing a structured and scalable data foundation for subsequent analysis.

[0018] 3. The hybrid network adaptation module adopts multi-protocol adaptive switching and erasure coding redundant transmission technology. In the event of network signal fluctuations or high packet loss rate, it automatically switches to the optimal transmission protocol or enables dual-link parallel transmission, which significantly reduces the risk of delay and interruption in real-time interaction in training activities. It is especially suitable for knowledge and skills training activities in remote areas with unstable network conditions.

[0019] 4. The comprehensive scoring output module divides learners into excellent / average / poor levels through cluster analysis, and generates targeted strategy suggestions by combining cluster characteristics with the deviation from the preset knowledge and skills training goals. This helps knowledge instructors quickly identify weak links in knowledge and skills training and achieve data-driven, precise improvement of knowledge and skills training. Attached Figure Description

[0020] Figure 1 This is a system structure diagram of the present invention;

[0021] Figure 2 This is an architecture diagram of the real-name binding module of the present invention;

[0022] Figure 3 This is an architecture diagram of the automatic data acquisition module of the present invention;

[0023] Figure 4 This is an architecture diagram of the dynamic data matrix modeling module of the present invention. Detailed Implementation

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see the appendix Figure 1 -Appendix Figure 4 This invention provides a computer lab learning outcome evaluation system based on real-name binding and intelligent data collection. The system includes a real-name binding module, an automatic data collection module, a dynamic data matrix modeling module, a dynamic weight optimization module, a comprehensive score output module, and a hybrid network adaptation module.

[0026] The real-name binding module is used to uniquely bind learner identity information with terminal device hardware feature code, generate device identifier and store it in encryption;

[0027] The real-name binding module establishes a unique association between the learner's identity and the terminal device through multi-level hardware feature extraction and encrypted binding technology, ensuring the non-repudiation of knowledge and skills training data collection and the legitimacy of the devices. The module includes a hardware feature code generation unit, an encrypted binding unit, and an anomaly detection unit. The collaborative operation of these units is as follows:

[0028] The hardware signature generation unit is used to extract tamper-proof hardware identification information from the terminal device and generate a unique hash value. The unit first collects the device's physical network address (MAC address) through the application programming interface provided by the operating system. Preferably, the physical network address is a globally unique identifier that is fixed at the factory of the device's network card, such as the MAC address of an Ethernet or wireless network card.

[0029] Simultaneously, the unit reads the CPU's serial number through the underlying instruction set. Preferably, the serial number is an unmodifiable identification code written into the chip by the CPU manufacturer. To avoid the risk of privacy leakage caused by directly storing raw hardware information, the unit performs hash processing on the aforementioned hardware identification information: first, it generates a hash value for the CPU serial number, calculated using the following formula:

[0030] H cpu =SHA-256(CPUSerial);

[0031] Wherein, SHA-256(·) is the SHA-256 hash function, which outputs a hexadecimal string of length 256; H cpu The hash value is the CPU serial number; CPUSerial is the original CPU serial number string.

[0032] Furthermore, the MAC address and CPU serial number hash value are concatenated and then hashed again to generate a unique device identifier:

[0033] H dev =SHA-256(M) mac ∥H cpu );

[0034] Among them, M mac The MAC address of the learner's terminal device; H cpu The hash value of the CPU serial number; ∥ is the string concatenation operator; H dev This is a unique identifier for the device, used for learner-device binding; SHA-256(·) is the SHA-256 hash function, which outputs a 256-bit hexadecimal string.

[0035] The encrypted binding unit is used to irreversibly encrypt and store the learner's student ID and device identifier. The unit receives the learner's student ID S provided by the academic affairs system. id Combine it with the device's unique identifier Hdev Encryption using the Advanced Encryption Standard (AES) algorithm generates ciphertext data:

[0036] E = AES - 256(S) id ∥H dev Key);

[0037] Where Key is a 256-bit encryption key; H dev E represents the bound device identifier stored in the database; E is the encrypted binding relationship data, stored in the database, and the original student ID or device identifier cannot be recovered through reverse engineering.

[0038] Preferably, the key employs a segmented storage strategy, with part of the key stored in a secure hardware module on the server side, and the other part injected by the system administrator during the initialization phase. The encryption process ensures that even if the database is compromised, attackers cannot directly obtain the binding relationship between the learner's student ID and the device.

[0039] The anomaly detection unit is used to verify the legitimacy of terminal devices in real time. When a learner logs into the system, the unit re-collects the current device's MAC address and CPU serial number, and generates a real-time device identifier H′ according to the hashing process described above. dev By comparing the real-time identifier H′ dev H stored in the database dev Determine if the equipment has been changed:

[0040] H′ dev ≠H dev ;

[0041] Among them, H′ dev H is the real-time computing identifier for the current device; dev The bound device identifier stored in the database.

[0042] Preferably, the abnormal event record includes a timestamp, original device identifier, real-time identifier, and network IP address, and is reported to the knowledge transmitter management terminal in real time through an independent communication channel.

[0043] Furthermore, when the unit detects a device malfunction, it simultaneously performs the following operations: temporarily freezes the current terminal's data upload permissions and pushes a secondary authentication request to the learner's device. Preferably, the secondary authentication uses a dynamic password or biometric recognition (such as facial recognition via camera) to distinguish between device loss and malicious impersonation scenarios.

[0044] The hardware feature code generation unit solves the problem of single hardware identifiers being easily tampered with by using double hashing and concatenation operations, ensuring the uniqueness and collision resistance of device identifiers;

[0045] The encrypted binding unit uses the AES-256 algorithm and key segmentation management to prevent learner-device binding relationships from being reverse-engineered or leaked in bulk;

[0046] The anomaly detection unit uses real-time hash comparison and dynamic verification mechanisms to quickly identify and isolate risks associated with equipment changes, providing a fundamental guarantee for the reliability of subsequent data collection.

[0047] The automatic data collection module communicates with the real-name binding module and is used to collect screen operation logs, code compilation results and interactive behavior data of the training activity space in real time after verifying the learner's identity based on the device identifier.

[0048] The automatic data acquisition module uses an event-triggered mechanism and code semantic parsing technology to capture learners' terminal operation data in real time and quantify their programming behavior characteristics. The module includes an event-triggered unit, a code analysis unit, and a difference calculation unit. The technical implementation method is as follows:

[0049] The event triggering unit is used to drive screen capture and data acquisition according to preset rules. The unit listens for code compilation completion events on the learner's terminal through the operating system's underlying interface. Preferably, the compilation completion event is a successful compilation signal returned by the integrated development environment (IDE) or an executable file generation operation. When such an event is detected, the current screen image is immediately captured and saved as a lossless compressed image file.

[0050] Simultaneously, the unit periodically triggers a screenshot operation. Preferably, the period is set to a time interval to avoid conflict with learner's active operations. The resolution of the captured image is consistent with the screen's native resolution to ensure clear recognition of code and text. The screenshot operation is combined with event-driven data acquisition, balancing real-time performance and data integrity.

[0051] The code analysis unit extracts code text from screen images and parses it into a structured syntax tree. The unit first calls an Optical Character Recognition (OCR) engine to locate and convert the code regions in the screenshot. Preferably, the OCR engine uses an end-to-end recognition model based on a convolutional neural network, supporting character sets for multiple programming languages ​​(such as special symbols in C / C++ and Python). The recognition result is output as a UTF-8 encoded plain text code file, preserving the original indentation and grammatical structure.

[0052] Furthermore, the unit converts the text code into an Abstract Syntax Tree (AST) using a syntax parser. Preferably, the parser generates a tree structure based on the syntax rules of the target programming language, where each node represents a code element (such as a variable declaration, function call, or control statement). During the parsing process, the unit records the following key node information:

[0053] The variable declaration node specifies the name, data type, and scope.

[0054] The name, parameter list, and return type of the function call node;

[0055] Conditional expressions and branch paths for control flow nodes.

[0056] The difference calculation unit is used to quantify the structural and logical differences between the learner's code and the reference code. The unit receives the learner's code AST and the knowledge transmitter's pre-set reference code AST, and calculates the difference index using the following formula:

[0057] Structural Difference:

[0058]

[0059] Among them, AST stu Abstract Syntax Tree (AST) for learner code; ref An abstract syntax tree for reference code provided to knowledge transmitters; N(·) is the total number of nodes in the abstract syntax tree; D struct The structural difference degree reflects the differences in code form.

[0060] Logical difference:

[0061]

[0062] Where CF(·) represents the set of control flow paths in the code; ∩ is the intersection operator; N paths D represents the total number of control flow paths in the reference code. logic For logical difference degree.

[0063] The event triggering unit avoids the resource waste of traditional polling collection by using a collaborative mechanism between compiled events and periodic screenshots, ensuring the integrity of the code submission phase;

[0064] The code analysis unit uses a cascaded processing of OCR and a syntax parser to solve the problem that screenshots cannot directly obtain code text, while preserving the semantic information of the code through AST parsing.

[0065] The difference calculation unit overcomes the shortcomings of traditional line comparison methods in identifying code logic errors by using multi-dimensional quantification of structural difference and logical difference, thus providing reliable input for subsequent dynamic weight scoring.

[0066] The dynamic data matrix modeling module receives data from the automatic data acquisition module, generates a reference code variant library using a generative adversarial network, compares learners’ code logic and structural defects with a discriminator, and outputs a dynamic difference score; and constructs a multi-dimensional data matrix by time series based on attendance status, test scores, code difference, and number of spatial interactions in training activities.

[0067] In this embodiment, the module receives raw operation logs, voice interaction data, and code submission records from the automatic data acquisition module, and generates a standardized spatiotemporal data matrix by combining generative adversarial networks and statistical analysis. The technical implementation and collaborative logic are as follows:

[0068] Input preparation: Extract reference code (such as sorting algorithm implementation) of the target function from the course knowledge graph, parse its abstract syntax tree (AST) and control flow graph (CFG) as the input benchmark for the generative adversarial network.

[0069] Generator operations: The generator based on the Transformer architecture performs syntactic mutations on the reference code to generate candidate code that is logically equivalent but structurally different;

[0070] The generated code undergoes syntax parsing and unit testing verification, and only the variant code that compiles successfully and is functionally equivalent is stored in the variant library;

[0071] The learner's code and the reference code are input into the symbolic execution engine to generate a set of path constraints Φ. 学习 With Φ 参考 Calculate the deviation of the constraint coverage between the two:

[0072]

[0073] Where, Φ 参考 The set of path constraints for the reference code; Φ 学习 For the set of path constraints of the learner's code; |Φ 参考 | represents the cardinality of the reference code path constraint set, indicating the total number of logical paths that the reference code should ideally cover; |Φ 学习 ∩Φ 参考 | is the cardinality of the intersection set of the learner's code and the reference code's path constraint sets, representing the number of times the learner's code correctly covers the expected logical paths of the reference code; S 逻辑 Scoring for logical flaws.

[0074] AST difference calculation: Extract the learner's code AST and compare it with the AST of each reference variant in the variant library. Take the minimum similarity as the structural defect score.

[0075] The weighting of logical and structural flaws is dynamically adjusted based on the course objectives to generate a comprehensive difference score.

[0076] S 差异 =α·S 逻辑 +β·S 结构 ;

[0077] Among them, S 差异The learner's code is scored based on its overall difference from the reference code variant library; α is the weighting coefficient for logical flaw scoring; S 逻辑 β is the score for logical defects; β is the weighting coefficient for structural defects; S 结构 Score the structural defects.

[0078] If the learner's code encounters a serious error (such as an infinite loop or an unhandled exception) during symbolic execution or test case verification, then S is directly assigned. 差异 =1 (highest defect level), and trigger a real-time alarm to the teacher's end.

[0079] The data normalization unit is used to map heterogeneous data to a unified dimensional range, eliminating the impact of dimensional differences on scoring. The unit receives the raw data stream verified by the real-name binding module, including attendance status, test scores, number of interactions in training activity spaces, and code difference indicators, and performs standardization processing according to the following rules:

[0080] Attendance status: binary value conversion, absence is marked as 0, attendance is marked as 1. Preferably, the absence judgment is based on the number of consecutive unresponsive heartbeat packets in the device login log exceeding a preset threshold;

[0081] Test score: Linearly normalized to the 0,1 interval, the calculation formula is as follows:

[0082]

[0083] Among them, S raw The learner's raw score in a single training activity space test, typically on a percentage basis or a fixed score; S min The lowest score among all learners in the same test, dynamically calculated within the current training activity space data; S max The highest score achieved by all learners in the class on the same test is used to calibrate the normalized benchmark; S norm The normalized test score is mapped to the 0,1 interval.

[0084] Preferably, the score range is dynamically calculated based on the spatial test of a single training activity to avoid interference from historical data;

[0085] Interaction frequency in training activities: The hyperbolic tangent function was used to suppress abnormally high-activity behavior. The calculation formula is as follows:

[0086] a interact =tanh(N) raise +N ask );

[0087] Where, N raise N represents the number of times a learner raises their hand in a single training session. askThe number of questions asked by the learner in a single training session; tanh(·) is the hyperbolic tangent function; a interact This refers to the normalized spatial interaction indicators for training activities.

[0088] Preferably, this design prevents a small number of learners from artificially inflating their scores through frequent, ineffective interactions;

[0089] Code Difference: The structural difference D of directly referencing the output of the automatic data acquisition module. struct Logical difference degree D logic , retain its original value range of 0,1.

[0090] The matrix expansion unit is used to organize data by time series, constructing a dynamically growing multidimensional matrix. After each lesson, a new row of data is added to the unit, with fixed column dimensions: [Attendance Status, Test Score, Structural Difference, Logical Difference, Training Activity Space Interaction Value]. The data matrix representation for a single learner is as follows:

[0091]

[0092] Where k2 is the learner ID; n2 is the number of classes attended.

[0093] Preferably, the matrix adopts a sparse storage format, retaining only non-zero elements and their coordinates to save storage space.

[0094] The anomaly marking unit is used to identify and correct data deviations caused by device malfunctions. When the real-name binding module detects a device identifier mismatch (H′... dev ≠H dev When this occurs, the unit applies a decay coefficient to the data row corresponding to the anomaly knowledge module:

[0095]

[0096] Where λ is the attenuation factor; Let be the data vector of all columns in the i-th row of the data matrix of the k-th learner;

[0097] Preferably, the factor is set to a constant less than 1 to reduce the weight contribution of outlier data to the score. Preferably, the decay operation is performed before the data is written to the matrix to ensure that subsequent processing is not disturbed.

[0098] The data normalization unit addresses the heterogeneity of data such as attendance, testing, and interaction through multi-strategy mapping, ensuring data comparability within the matrix. The matrix expansion unit adopts a time-series organization method, preserving the dynamic evolution characteristics of learners' performance and providing a time-dimensional analysis basis for subsequent weight optimization based on covariance. The anomaly labeling unit isolates the interference of abnormal data from devices through a decay coefficient, improving the robustness of the scoring system. Its technical implementation relies on the real-time verification results of the real-name binding module.

[0099] The dynamic weight optimization module calculates the covariance matrix based on a multidimensional data matrix and generates a scoring weight vector through principal component analysis and constraint optimization algorithms.

[0100] The dynamic weight optimization module dynamically generates scoring weights through covariance analysis and constraint optimization algorithms, eliminating subjective biases from manually setting weights and ensuring that the scoring model adapts to the data distribution characteristics of different training activity spaces. The module includes a covariance calculation unit, a principal component analysis unit, and a constraint optimization unit. Its technical implementation and data interaction logic are as follows:

[0101] The covariance calculation unit is used to quantify the statistical correlation between data in different dimensions, providing a mathematical basis for weight optimization. The unit receives the set of the whole class learner matrix {M} output by the dynamic data matrix modeling module. (1) M (2) ,…,M (K)}, where each learner matrix It includes five columns of data: attendance, test scores, structural differences, logical differences, and interactive values ​​of training activities.

[0102] The unit first calculates the global mean vector μ = [μ1, μ2, μ3, μ4, μ5] for each column of data.

[0103] Subsequently, the unit calculates the covariance matrix across learners and across training activity spaces. Its element C pq Defined as:

[0104]

[0105] Among them, C pq Let be the element in the p-th row and q-th column of the covariance matrix; μ is the normalized value in the i-th row and p-th column of the dynamic data matrix of the k-th learner; p and μ q represents the mean of the p-th and q-th dimensions, respectively; K is the total number of learners; n is the number of rows of training activity space data for a single learner; K·n-1 is the unbiased estimation correction term for covariance calculation.

[0106] Preferably, the covariance matrix reflects the correlation of data in each dimension, such as the negative correlation between test scores and code discrepancies (high-scoring learners usually have low code discrepancies).

[0107] Principal component analysis (PCA) is used to extract the main directions of change in the data and generate initial weight vectors. The unit performs eigenvalue decomposition on the covariance matrix C, solving for its eigenvalues ​​λ1≥λ2≥…≥λ5 and the corresponding eigenvectors v1,v2,…,v5. The eigenvector v1 corresponding to the largest eigenvalue is selected as the initial weight.

[0108] Preferably, the principal component direction represents the projection direction with the largest data variance, enabling the scoring model to capture the main differences in learner performance.

[0109] The constrained optimization unit is used to impose reasonable constraints on the initial weights to generate the final optimized weights. The objective function of the constrained optimization unit is:

[0110]

[0111] Among them, w j Let be the weight of the j-th dimension; Let be the normalized value of the j-th dimension for the i-th lesson of the k-th learner; K is the total number of learners; and n is the number of rows of training activity space data for a single learner.

[0112] The covariance calculation unit identifies key influencing dimensions (such as the strong correlation between code difference and test scores) by statistically analyzing the correlation of data across the entire class, providing a data-driven basis for weight allocation.

[0113] The comprehensive scoring output module performs weighted fusion of the multidimensional data matrix according to the scoring weight vector, outputs the learner's comprehensive score, and generates knowledge and skills cultivation strategy suggestions based on the cluster analysis results;

[0114] The comprehensive scoring output module uses multi-dimensional weighted fusion and cluster analysis techniques to generate a comprehensive score of learners' performance in training activities and suggestions for knowledge and skills development strategies, providing data-driven decision support for knowledge instructors. The module includes a weighted summation unit, a cluster analysis unit, a strategy recommendation unit, and a heatmap visualization unit. Its technical implementation and data interaction logic are as follows:

[0115] The weighted summation unit integrates the weight vector generated by the dynamic weight optimization module with the learner's dynamic data matrix to calculate the overall score. The unit receives the optimal weight vector W output by the dynamic weight optimization module. opt = [w1, w2, w3, w4, w5], where w1 to w5 correspond to the weights of attendance status, test score, structural difference, logical difference, and training activity space interaction value, respectively.

[0116] For the Kth learner, the formula for calculating their comprehensive score is:

[0117]

[0118] Where Score is the learner's overall score, reflecting the weighted performance of multi-dimensional data under dynamic weights; K is the total number of learners; n is the number of rows of training activity space data for a single learner; W opt,j The optimal weight for the j-th dimension; Let be the normalized value in the i-th row and j-th column of the data matrix for the k-th learner.

[0119] The clustering analysis unit is used to segment learner groups based on their performance and identify areas for improvement in knowledge and skills development. The unit contains a dynamic data matrix {M} of all learners in the class. (1) M (2) ,…,M (K) Input the clustering algorithm. Preferably, the K-means algorithm is used with a preset number of 3 categories. Calculate the category label for each learner and the coordinates of the center of each cluster.

[0120] During clustering, the Euclidean distance between the learner matrix and the cluster center is calculated for each unit:

[0121]

[0122] in, d represents the value in the i-th row and j-th column of the c-th cluster center matrix; (k,c) is the Euclidean distance between learner k and cluster c, used to determine the learner's performance category; i is the training activity space index; j is the dimension index.

[0123] The strategy recommendation unit generates suggestions for adjusting knowledge and skills development based on clustering results. The unit calculates the Euclidean distance between each cluster center and a preset reference center (e.g., the center of the high-achieving student cluster). like This triggers the generation of knowledge and skills development strategy suggestions. Preferably, the reference center is derived from historical high-achieving learner data, and the threshold θ is dynamically set according to the knowledge and skills development activity scenario.

[0124] The heatmap visualization unit maps learner matrix data to color intensities, and the rendering formula is as follows:

[0125] Color(a ij )=(255(1-a ij ),255a ij ,0);

[0126] Among them, Color(a ij ) represents the output of the entire color mapping function; a ijLet be the normalized value in the i-th row and j-th column of the learner data matrix; 255(1-a ij ) represents the intensity value of the red channel; 255a ij 0 represents the intensity value of the green channel; 0 represents the blue channel, which is fixed at 0 and does not participate in color rendering.

[0127] The hybrid network adapter module communicates with all modules and is used to coordinate the transmission priority of real-time control commands, batch data and abnormal events in a wireless or wired hybrid network to ensure data synchronization between modules.

[0128] The hybrid network adaptation module addresses latency jitter and packet loss issues in knowledge and skills training terminal data transmission within heterogeneous network environments through multi-protocol adaptive switching and transmission path optimization technologies, ensuring the stability of real-time training activity space interactions. The module comprises a network status monitoring unit, a protocol switching decision unit, and a transmission optimization unit, with its technical implementation and collaborative logic as follows:

[0129] The network status monitoring unit collects network environment parameters from terminal devices in real time, providing data support for protocol switching and path selection. The unit periodically checks the link quality indicators of the currently connected wireless network (such as Wi-Fi, 5G) and wired network, including:

[0130] Signal Strength Indicator (RSSI): The received signal strength indication value is obtained through the physical layer driver interface. Preferably, the value is smoothed by Kalman filtering to eliminate instantaneous fluctuations.

[0131] End-to-end delay: The round-trip time (RTT) is measured by sending ICMP probe packets, and the one-way delay τ = RTT² is calculated.

[0132] Packet loss rate: The percentage of TCP retransmissions per unit time out of the total number of transmissions. Ideally, the statistical window should be set to dynamically adjust to adapt to network fluctuations.

[0133] The unit encapsulates the above parameters into a network state vector Q = [RSSI, τ, p] loss ], where p loss The packet loss rate is calculated and pushed to the protocol switching decision unit.

[0134] The protocol switching decision unit triggers either a transmission protocol switch or dual-link parallel transmission based on network status assessment results and preset thresholds. The unit employs a fuzzy logic evaluation model, defining the input variable as Q and the output variable as the protocol switching action A ∈ {hold, switch to backup protocol, dual-link transmission}.

[0135] The core of the evaluation model consists of the membership function and the rule base.

[0136] Membership function: Defines three fuzzy sets, "Excellent", "Medium", and "Poor", for each network parameter. Fuzzy rule base: Contains combined rules such as "If RSSI is excellent and latency is excellent, then maintain the current protocol". Rule weights are adjusted based on historical handover success rate feedback.

[0137] The transmission optimization unit optimizes data transmission paths and redundancy strategies under a selected protocol. When dual-link transmission is enabled, the unit divides the data stream into multiple sub-blocks and transmits them in parallel via the primary link (e.g., Wi-Fi) and the backup link (e.g., 5G). The receiving end reassembles the data based on erasure coding. Preferably, the factor is dynamically adjusted according to the real-time packet loss rate.

[0138] Define transmission priority rules:

[0139]

[0140] Where Priority represents the transmission priority of data packets; 0 represents the multidimensional training activity space data generated by the dynamic data matrix modeling module; 1 represents device identifier mismatch or hardware tampering events; and 2 represents real-time control commands sent by the knowledge transmitter.

[0141] By quantifying the continuous changes in network state using fuzzy sets, the hard handover jitter problem of traditional threshold methods is avoided. Relying on the real-time evaluation of the network state vector Q, a closed-loop feedback is formed with the state monitoring unit to ensure the timeliness of the handover action.

[0142] By using a blockchain smart contract executor, the scoring rules and strategy generation logic are written into the chaincode, triggering an immutable execution process.

[0143] In this embodiment, the blockchain smart contract executor achieves immutable execution of scoring rules and strategies through chaincode logic encapsulation and distributed ledger technology. Specifically, the executor receives the scoring weight vector generated by the dynamic weight optimization module and the strategy generation conditions from the comprehensive scoring output module, converts them into verifiable smart contract code, and deploys them to the blockchain network, forming an automated execution system under a full-node consensus mechanism.

[0144] The chaincode logic injection unit uses modular programming to convert the scoring rules into chaincode executable logic. d The data is serialized and written to the chaincode state database, where the weight coefficients of each dimension strictly correspond to the column dimensions of the dynamic data matrix. The strategy generation logic is encoded as chaincode transaction trigger conditions. When a new data row is added to the dynamic data matrix or an abnormal event such as a device identifier is detected, the scoring calculation function is automatically invoked.

[0145]

[0146] Among them, w i This represents the normalized index value of the i-th dimension in the dynamic data matrix, including attendance status, test score, logic defect score, structural defect score, and interaction index; x i d represents the normalized index value of the i-th dimension; d represents the total number of column dimensions in the multidimensional data matrix.

[0147] Preferably, the chain code applies a normalization constraint to the weight coefficients to ensure... The mathematical conditions are automatically enforced with each weight update.

[0148] The on-chain and off-chain storage units employ a hierarchical storage architecture to ensure the verifiability and scalability of the scoring rules. The chaincode only stores the hash value H(W) = SHA256(W) of the scoring weight vector and the bytecode fingerprint of the policy triggering logic. Complete weight data and policy rules are processed using cryptographic algorithms and stored in a decentralized file system. Preferably, a content addressing identifier is used to establish an index association between the on-chain hash value and the off-chain data. The specific generation method is as follows:

[0149] CID = Multihash(Encrypt(W,K) pub ));

[0150] Among them, K pub The public key assigned to the federated learning coordinator ensures controllable access permissions for cross-agency data access; W is the scoring weight vector generated by the dynamic weight optimization module; Encrypt(W,K) pub ) is the public-key encryption function; Multihash(·) is the multi-hash function.

[0151] When it is necessary to verify the integrity of the scoring rules, the off-chain data is retrieved through the CID and decrypted, and the hash value is recalculated to match the on-chain record.

[0152] The cross-chain verification unit implements multi-chain data consistency verification through a lightweight node protocol. When the federated learning coordinator aggregates the scoring models of each participant, the smart contract executor submits the rule hash list of its local chaincode to the cross-chain relay, and verifies the consistency of the chaincode logic of each node with the global constraints through Merkle tree proof. Preferably, a relay chain mode is used to perform secondary consensus on the verification results of heterogeneous blockchain networks to prevent model bias caused by data tampering on a single chain.

[0153] In this embodiment, the smart contract executor and the hybrid network adaptation module form a closed loop for anomaly handling. When a device identifier change event is detected, the chaincode automatically triggers the following transaction process: first, the scoring output channel of the associated learner is frozen; second, the characteristics of the anomaly event (including device hardware fingerprint change records and threshold exceeding limits for logical and structural defect scoring differences) are written into the immutable ledger of the blockchain; and finally, a manual review request is pushed to the knowledge transmitter. The frozen state continues until the federated learning coordinator verifies the authenticity of the node's historical data, at which point it is lifted by a majority vote of the nodes.

[0154] The interaction between the executor and the dynamic data matrix modeling module is achieved through an event-driven mechanism. An event listener for adding a new data row to the matrix triggers a chaincode transaction call in real time. After the score calculation is completed within the blockchain network, the hash value of the result is written back to the matrix metadata field. Preferably, an asynchronous callback mechanism is used to avoid chaincode execution blocking caused by high-concurrency data transmission, while timestamp signatures ensure the temporal consistency between the data row and the score result.

[0155] The federated learning coordinator, embedded in the hybrid network adaptation module, generates a federated scoring model by aggregating local model parameters across institutions, supporting a unified evaluation standard mapping under data isolation across multiple institutions.

[0156] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A computer lab learning outcome evaluation system based on real-name binding and intelligent data collection, characterized in that: The system includes a real-name binding module, an automatic data collection module, a dynamic data matrix modeling module, a dynamic weight optimization module, a comprehensive scoring output module, a hybrid network adaptation module, and a federated learning coordinator. The real-name binding module is used to uniquely bind learner identity information with terminal device hardware feature code, generate device identifier and store it in encryption; The automatic data collection module is connected to the real-name binding module and is used to collect the learner's screen operation logs, code compilation results and interactive behavior data in the training activity space in real time after verifying the learner's identity based on the device identifier. The dynamic data matrix modeling module receives data from the automatic data acquisition module, generates a reference code variant library using a generative adversarial network (GAN), compares learners' code logic and structural defects using a discriminator, and outputs a dynamic difference score. It also constructs a multi-dimensional data matrix based on attendance status, test scores, code difference scores, and the number of interactions in training activities over time. The generation of the reference code variant library using a GAN includes: generating logically equivalent but syntactically different reference code variant libraries based on reference code in the course knowledge graph, covering three programming features: loop nesting optimization, exception handling completeness, and code reusability. The dynamic weight optimization module calculates the covariance matrix based on the multidimensional data matrix and generates a scoring weight vector through principal component analysis and constraint optimization algorithms. The comprehensive scoring output module performs weighted fusion of the multidimensional data matrix according to the scoring weight vector, outputs the learner's comprehensive score, and generates knowledge and skills cultivation strategy suggestions based on the clustering analysis results. The hybrid network adapter module communicates with all modules and is used to coordinate the transmission priority of real-time control commands, batch data and abnormal events in a wireless or wired hybrid network to ensure data synchronization between modules. By using a blockchain smart contract executor, the scoring rules and strategy generation logic are written into the chaincode, triggering an immutable execution process. The federated learning coordinator, embedded in the hybrid network adaptation module, generates a federated scoring model by aggregating local model parameters across institutions, supporting a unified evaluation standard mapping under data isolation across multiple institutions.

2. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection as described in claim 1, characterized in that, The real-name binding module includes: The hardware signature generation unit is used to collect the device's MAC address and CPU serial number, and generate a hash value H. dev =SHA-256(M) mac ||H cpu ); Among them, M mac The MAC address of the learner's terminal device; H cpu The hash value of the CPU serial number; || is the string concatenation operator; H dev This is a unique identifier for the device, used for learner-device binding; SHA-256(·) is the SHA-256 hash function, which outputs a 256-bit hexadecimal string. Encrypted binding unit, learner's student ID S id With H dev Storage is encrypted using the AES algorithm; The anomaly detection unit detects the current device hash value H′. dev ≠H dev When an exception event is triggered, an exception event log is recorded, where H′ dev H is the real-time computing identifier for the current device. dev The bound device identifier stored in the database.

3. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 1, characterized in that, The automatic data acquisition module includes: The event triggering unit triggers a screenshot at a preset time interval ΔT or upon completion of code compilation. The code analysis unit uses OCR to recognize code text in screenshots and parses it into an abstract syntax tree; The difference calculation unit calculates the structural difference D between the learner's code and the reference code. struct and logical difference D logic .

4. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 3, characterized in that, In the difference calculation unit: Structural Difference D struct Normalized value for the difference in the number of AST nodes: Among them, AST stu Abstract Syntax Tree (AST) for learner code; ref An abstract syntax tree for reference code provided to knowledge transmitters; N(·) is the total number of nodes in the abstract syntax tree; D struct The degree of structural difference reflects the differences in code form; Logical difference D logic The complement of the control flow path matching rate: Where CF(·) represents the set of control flow paths in the code; ∩ is the intersection operator; N paths D represents the total number of control flow paths in the reference code. logic For logical difference degree.

5. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 1, characterized in that, The dynamic data matrix modeling module includes: The following comparative analysis was performed on the code submitted by learners: Logical defect detection: Calculate the deviation between the learner's code and the reference code in the control flow graph path coverage; Structural defect detection: Compare the abstract syntax tree similarity between learner code and generator variant library to identify redundant or inefficient structures; Output a dynamic difference score, where logical defects have a weight of 0.6 and structural defects have a weight of 0.4; The data normalization unit maps attendance status, test scores, code differences, and training activity space interaction counts to the [0,1] interval, where the training activity space interaction counts are suppressed for overactivity using a hyperbolic tangent function. a interact = tanh(N raise +N ask ); Where, N raise N represents the number of times a learner raises their hand in a single training session. ask The number of questions asked by the learner in a single training session; tanh(·) is the hyperbolic tangent function; a interact The normalized spatial interaction indicators for training activities; The matrix expansion unit adds a row of data to the learner matrix for each lesson, with column dimensions including attendance, tests, structural differences, logical differences, and interactions; The anomaly marking unit adds an attenuation coefficient λ = 0.5 to the training activity space data of equipment changes.

6. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 1, characterized in that, The dynamic weight optimization module includes: The covariance calculation unit calculates the covariance matrix based on the set of learners in the whole class. Its elements are: Among them, C pq Let be the element in the p-th row and q-th column of the covariance matrix; Let be the normalized value of the i-th row and p-th dimension in the dynamic data matrix of the k-th learner; μ is the normalized value of the i-th row and q-th dimension of the dynamic data matrix of the k-th learner; p and μ q represents the mean of the p-th and q-th dimensions respectively; K is the total number of learners; n is the number of rows of training activity space data for a single learner; K·n-1 is the unbiased estimation correction term for covariance calculation; Principal component analysis unit extracts the eigenvector V1 corresponding to the largest eigenvalue of the covariance matrix as the initial weight W. pca ; Constrained optimization unit, constructing Lagrangian function and solving for the optimal weight W that maximizes the objective function. opt .

7. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 6, characterized in that, The objective function of the constrained optimization unit is: Among them, w j Let be the weight of the j-th dimension; is the normalized value of the j-th dimension in the i-th row of the dynamic data matrix of the k-th learner; K is the total number of learners; n is the number of rows of training activity space data for a single learner.

8. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 7, characterized in that, The comprehensive scoring output module includes: The weighted summation unit performs a weighted summation of the matrix row data lesson by lesson according to the optimal weights: Where Score is the learner's overall score, reflecting the weighted performance of multi-dimensional data under dynamic weights; n is the number of rows of training activity space data for a single learner; W opt,j The optimal weight for the j-th dimension; Let be the normalized value of the i-th row and j-th dimension in the dynamic data matrix of the k-th learner; The clustering analysis unit identifies anomalous learner groups using the K-means algorithm, with a cluster size of 3. The strategy recommendation unit outputs knowledge and skill development adjustment suggestions when the Euclidean distance between the class center and the reference center exceeds the threshold θ, where θ is the Euclidean distance threshold between the class center and the reference center.

9. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 1, characterized in that, The hybrid network adaptation module includes: The protocol selection unit transmits real-time control commands via UDP broadcast and transmits matrix incremental data via TCP fragmented compression. The priority management unit defines the transmission priority rules: Where Priority represents the transmission priority of data packets; 0 represents the multidimensional training activity space data generated by the dynamic data matrix modeling module; 1 represents device identifier mismatch or hardware tampering events; and 2 represents real-time control commands sent by the knowledge transmitter.

10. The computer lab learning outcome evaluation system based on real-name binding and intelligent data collection according to claim 1, characterized in that, The triggering of the immutable execution process includes: The chaincode logic injection unit is used to write the scoring weight vector W generated by the dynamic weight optimization module and the strategy generation conditions of the comprehensive scoring output module into the blockchain chaincode, defining the following immutable triggering rules: When the dynamic data matrix modeling module adds a new data row, it triggers the chaincode to perform score calculation. And store the resulting hash value on the blockchain, where w j The weight coefficients for the j-th dimension generated by the dynamic weight optimization module; x j The normalized index value of the j-th dimension in the dynamic data matrix modeling module; d is the total number of column dimensions of the multidimensional data matrix; When the hybrid network adaptation module detects an abnormal device identifier event, the chaincode automatically freezes the associated learner's scoring output and generates a manual review request event. The on-chain-off-chain storage unit is used to encrypt and store the original data of the scoring rules in IPFS. Only the IPFS content identifier and the rule hash value are retained in the chaincode. Verifiable query of off-chain data is achieved by indexing the rule hash value. The cross-chain verification unit verifies the consistency of chaincode rules among various institutions through a cross-chain protocol when the federated learning coordinator aggregates cross-institutional scoring models, ensuring that the generation of federated model weights meets preset constraints.

Images