Higher education data deep mining system based on cloud computing and artificial intelligence

Abstract

The invention provides a higher education data deep mining system based on cloud computing and artificial intelligence, and relates to the technical field of data mining, and the system comprises a multi-modal data fusion perception module which is used for collecting multi-modal data in an education scene, and carrying out the classification processing through an education scene exclusive data classification system, an education scene dynamic perception engine is constructed, and the attention state, the understanding depth and the emotion change of the learner are perceived in real time; the dynamic knowledge graph module is used for constructing an initial knowledge graph; the emotion calculation and analysis module is used for carrying out emotion calculation and cognitive calculation on the multi-modal data, and combining emotion calculation and cognitive calculation results to construct an emotion and cognitive two-dimensional evaluation model. The emotion and cognitive weight is automatically adjusted according to the learning stage, the evaluation deviation caused by a fixed proportion is avoided, and the evaluation accuracy is improved. And a time attenuation factor is introduced to reduce historical data interference, so that the fluctuation of a real-time evaluation result is reduced.

Description

Technical Field

[0001] This invention relates to the field of data mining technology, and in particular to a deep data mining system for higher education based on cloud computing and artificial intelligence. Background Technology

[0002] The level of higher education development is an important indicator of a country's comprehensive national strength and development potential. Currently, higher education faces new trends and challenges, making it essential for all countries to reconsider the important position and role of higher education at the provincial and municipal levels. A country with a mature higher education system can achieve lower enrollment costs, higher quality talent cultivation, stronger talent attraction, better research opportunities, a fairer social environment, and a greater output of high-caliber talent, among other benefits.

[0003] The field of higher education has accumulated a massive amount of multimodal data, such as online learning behavior logs, classroom interaction videos, and physiological signal monitoring data. However, traditional analysis methods have the following limitations: 1. Most systems rely solely on single indicators such as answer accuracy or attendance records, ignoring learners' emotional states, such as the impact of anxiety and fatigue on cognitive efficiency. Furthermore, the calculation of knowledge point mastery rate depends on a static question bank and does not incorporate dynamic weighting based on knowledge graph relevance or problem-solving process complexity. 2. The data collected by different systems are in heterogeneous formats and lack a unified standard, making cross-platform data fusion difficult.

[0004] Therefore, it is necessary to provide a new system for deep mining of higher education data based on cloud computing and artificial intelligence to solve the above-mentioned technical problems. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a system for deep mining of higher education data based on cloud computing and artificial intelligence.

[0006] The higher education data deep mining system based on cloud computing and artificial intelligence provided by this invention includes: The multimodal data fusion and perception module is used to collect multimodal data in educational scenarios and classify and process it through a data classification system specific to educational scenarios. It also builds a dynamic perception engine for educational scenarios to perceive learners' attention status, comprehension depth, and emotional changes in real time. The dynamic knowledge graph module is used to extract knowledge from multimodal data, construct an initial knowledge graph, and dynamically evolve the knowledge graph through time-aware graph reasoning. The dynamic evolution of the knowledge graph includes dynamically adjusting the relation weights based on the frequency and timeliness of knowledge use, and constructing a knowledge value assessment system to evaluate the value of new knowledge for teaching. The Affective Computing Analysis module is used to perform affective and cognitive computing on multimodal data, and to integrate the results of affective and cognitive computing to construct a dual-dimensional assessment model of affective and cognitive states, which integrates the analysis of learners' emotional and cognitive states.

[0007] Furthermore, the multimodal data fusion sensing module includes: The data acquisition unit is used to classify multimodal data into four categories according to educational scenarios: behavioral representation data, neurophysiological information data, human-computer interaction data, and learning context perception data. The multimodal feature extraction unit is used to extract facial expression features, speech intonation features, text semantic features, and physiological signal features; The data time alignment unit is used to synchronize the time of multimodal data.

[0008] Furthermore, the dynamic knowledge graph module includes: The knowledge extraction and verification unit is used to extract learning behavior data from multimodal data using an AI agent, and input the learning behavior data into the educational knowledge base to calculate semantic similarity for verification. The dynamic weight calculation unit is used to dynamically adjust the relationship weights based on the frequency and timeliness of knowledge use. The knowledge value assessment unit is used to evaluate the value of new knowledge to teaching, including five dimensions: teaching value, learning value, knowledge relevance, timeliness, and innovation. The knowledge fusion decision unit is used to determine whether new knowledge should be included in the knowledge graph.

[0009] Furthermore, the sentiment computing and analysis module includes: An educational emotion recognition unit is used to identify emotional features in facial expressions, voice tone, and physiological signals. The cognitive state assessment index unit is used to establish four dimensions: knowledge point mastery rate, problem-solving complexity, knowledge point review frequency, and knowledge point application breadth. It also assesses the degree of knowledge mastery, depth of thinking, and learning ability based on learning behavior data. The Emotional-Cognitive Fusion Analysis Unit is used to integrate emotional and cognitive states and calculate a comprehensive cognitive state score.

[0010] Furthermore, the system also includes a student cognitive path mapping module, used to identify the student's position in the knowledge graph; The student cognitive path mapping module includes: a cognitive path matching algorithm unit, which is used to calculate the matching degree between the student's learning path and the knowledge graph, transform the knowledge graph into a directed acyclic graph, where nodes are knowledge points and edges are prerequisite relationships, uses Dijkstra's algorithm to calculate the shortest edit distance between the student's learned path and the target path, and calculates the matching degree based on the shortest edit distance and the total number of nodes; The student knowledge mastery unit is used to display the student's position in the knowledge graph; The cognitive state visualization unit is used to generate a student knowledge graph profile, using color coding to indicate the degree of mastery.

[0011] Furthermore, facial expression feature extraction includes the following steps: Step 1: Face detection and alignment: Use the MTCNN algorithm to locate the face region, eliminate background interference, and align the face to the standard coordinate system through affine transformation to ensure the comparability of feature points; Step 2, Feature Point Localization: Key feature points are marked using Dlib's 68-point model, and the geometric distance and motion amplitude between feature points are calculated; Step 3, Facial Expression Classification: Based on FACS, facial expressions are decomposed into action unit combinations, deep features are extracted using a CNN model, and classification is performed using an SVM.

[0012] Furthermore, speech intonation feature extraction includes the following steps: Step 1, Preprocessing: The speech data is framed and windowed, and Hamming windowing is applied to reduce spectral leakage; Step 2, Time-Frequency Feature Extraction: Extract Mel-frequency cepstral coefficients and their first and second-order differences to capture dynamic changes in speech, and calculate acoustic parameters such as fundamental frequency, energy, and speech rate; Step 3: Analyze the speech sequence using the BiLSTM-Attention model and combine it with CRF to optimize label prediction.

[0013] Furthermore, text semantic feature extraction includes the following steps: Step 1, Preprocessing: Segment the text and remove stop words, and apply BERT to generate word vectors; Step 2, Context Modeling: Use the BERT architecture to capture long-distance dependencies and fine-tune the model according to the educational scenario; Step 3: Sentiment polarity analysis: Output sentiment labels through the Softmax layer and focus on key sentences using an attention mechanism.

[0014] Furthermore, physiological signal feature extraction includes the following steps: Step 1, Signal Denoising: Use wavelet transform to remove motion artifacts in ECG and EEG; Step 2, Feature Extraction: Calculate the standard deviation of RR interval and the low-frequency to high-frequency power ratio in the electrocardiogram, and extract the peak number and rise time characteristics of the skin conductance response; Step 3: State Classification: Use random forest to distinguish the state of physiological signals.

[0015] Furthermore, time synchronization of the multimodal data includes the following: generating a unified pulse signal through a GPS clock to trigger synchronous acquisition of multiple sensors; using cubic spline interpolation to generate virtual data under high-frequency timestamps for low-frequency data; applying sliding window averaging to high-frequency data to match the low-frequency rhythm; using the highest frequency sensor as the main axis; and aligning other data to the main axis timestamp through nearest neighbor interpolation.

[0016] Compared with related technologies, the cloud computing and artificial intelligence-based high-level data deep mining system provided by this invention has the following beneficial effects: 1. This invention solves the compatibility problem of heterogeneous data sources by cleaning and aligning facial images, voice, physiological signals and behavioral data in real time through edge computing nodes. It performs affective and cognitive computing on multimodal data and integrates the results of affective and cognitive computing to construct a dual-dimensional evaluation model of affective cognition. It integrates and analyzes the learner's emotional and cognitive states to improve the accuracy of affective recognition.

[0017] 2. This invention automatically adjusts the weights of emotion and cognition according to the learning stage to avoid evaluation bias caused by fixed ratios, dynamically adjusts the weights of relationships according to the frequency and timeliness of knowledge use, constructs a knowledge value evaluation system to evaluate the value of new knowledge to teaching, and introduces a time decay factor to reduce interference from historical data, thereby reducing the volatility of real-time evaluation results.

[0018] 3. This invention establishes four dimensions: knowledge point mastery rate, problem-solving complexity, knowledge point review frequency, and knowledge point application breadth. It assesses the degree of knowledge mastery, depth of thinking, and learning ability based on learning behavior data, integrates emotional state and cognitive state, calculates a comprehensive cognitive state score, and triggers personalized strategies based on an emotional-cognitive two-dimensional matrix for precise learning intervention. Attached Figure Description

[0019] Figure 1 A structural block diagram of the higher education data deep mining system based on cloud computing and artificial intelligence provided by the present invention; Figure 2 This is a structural block diagram of the multimodal data fusion sensing module provided by the present invention; Figure 3 This is a structural block diagram of the dynamic knowledge graph module provided by the present invention; Figure 4 This is a structural block diagram of the sentiment calculation and analysis module provided by the present invention; Figure 5 A flowchart illustrating the facial expression feature extraction process provided by this invention; Figure 6 A flowchart illustrating the speech intonation feature extraction process provided by this invention; Figure 7A flowchart illustrating the text semantic feature extraction process provided by this invention; Figure 8 The flowchart for physiological signal feature extraction provided by the present invention is shown. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Please refer to the following: Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 Figure 7 as well as Figure 8 ,in, Figure 1 A structural block diagram of the higher education data deep mining system based on cloud computing and artificial intelligence provided by the present invention; Figure 2 This is a structural block diagram of the multimodal data fusion sensing module provided by the present invention; Figure 3 This is a structural block diagram of the dynamic knowledge graph module provided by the present invention; Figure 4 This is a structural block diagram of the sentiment calculation and analysis module provided by the present invention; Figure 5 A flowchart illustrating the facial expression feature extraction process provided by this invention; Figure 6 A flowchart illustrating the speech intonation feature extraction process provided by this invention; Figure 7 A flowchart illustrating the text semantic feature extraction process provided by this invention; Figure 8 The flowchart for physiological signal feature extraction provided by the present invention is shown.

[0022] In the specific implementation process, such as Figures 1-4 As shown, the high-level data deep mining system based on cloud computing and artificial intelligence includes: The multimodal data fusion and perception module is used to collect multimodal data in educational scenarios and classify and process it through a data classification system specific to educational scenarios. It also builds a dynamic perception engine for educational scenarios to perceive learners' attention status, comprehension depth, and emotional changes in real time. In some embodiments, the multimodal data fusion sensing module includes: The data acquisition unit is used to classify multimodal data into four categories according to educational scenarios: behavioral representation data, neurophysiological information data, human-computer interaction data, and learning context perception data. The multimodal feature extraction unit is used to extract facial expression features, speech intonation features, text semantic features, and physiological signal features; The data time alignment unit is used to synchronize the time of multimodal data. The time synchronization is achieved by generating a unified pulse signal through the GPS clock to trigger the synchronous acquisition of multiple sensors. For low-frequency data, cubic spline interpolation is used to generate virtual data under high-frequency timestamps. For high-frequency data, sliding window averaging is applied to match the low-frequency rhythm. The highest frequency sensor is used as the main axis, and other data are aligned to the main axis timestamp through nearest neighbor interpolation.

[0023] The dynamic knowledge graph module is used to extract knowledge from multimodal data, construct an initial knowledge graph, and dynamically evolve the knowledge graph through time-aware graph reasoning. The dynamic evolution of the knowledge graph includes dynamically adjusting the relation weights based on the frequency and timeliness of knowledge use, and constructing a knowledge value assessment system to evaluate the value of new knowledge for teaching. In some embodiments, the dynamic knowledge graph module includes: The knowledge extraction and verification unit is used to extract learning behavior data from multimodal data using an AI agent, and input the learning behavior data into the educational knowledge base to calculate semantic similarity for verification. The dynamic weight calculation unit is used to dynamically adjust the relationship weights based on the frequency and timeliness of knowledge use. The knowledge value assessment unit is used to evaluate the value of new knowledge to teaching, including five dimensions: teaching value, learning value, knowledge relevance, timeliness, and innovation. The knowledge fusion decision unit is used to determine whether new knowledge should be included in the knowledge graph.

[0024] It should be noted that the AI ​​agent knowledge extraction uses LLM to parse teaching logs and student discussions, extract entities and relations, and output structured triples. The triples are then input into the educational knowledge base to calculate semantic similarity, and a manual review process is triggered for controversial knowledge.

[0025] The frequency calculation is as follows: The coefficient ranges from 1 to 1.5; The Affective Computing Analysis module is used to perform affective and cognitive computing on multimodal data, and to integrate the results of affective and cognitive computing to construct a dual-dimensional assessment model of affective and cognitive states, which integrates the analysis of learners' emotional and cognitive states.

[0026] In some embodiments, the sentiment computation and analysis module includes: An educational emotion recognition unit is used to identify emotional features in facial expressions, voice tone, and physiological signals. The cognitive state assessment index unit is used to establish four dimensions: knowledge point mastery rate, problem-solving complexity, knowledge point review frequency, and knowledge point application breadth. It also assesses the degree of knowledge mastery, depth of thinking, and learning ability based on learning behavior data. The Emotional-Cognitive Fusion Analysis Unit is used to fuse emotional and cognitive states and calculate a comprehensive cognitive state score. in, ; .

[0027] It should be noted that the reference Figure 5 As shown, facial expression feature extraction includes the following steps: Step 1: Face detection and alignment: Use the MTCNN algorithm to locate the face region, eliminate background interference, and align the face to the standard coordinate system through affine transformation to ensure the comparability of feature points; Step 2, Feature Point Localization: Key feature points are marked using Dlib's 68-point model, and the geometric distance and motion amplitude between feature points are calculated; Step 3, Facial Expression Classification: Based on FACS, facial expressions are decomposed into action unit combinations, deep features are extracted using a CNN model, and classification is performed using an SVM.

[0028] It should be noted that the reference Figure 6 As shown, speech intonation feature extraction includes the following steps: Step 1, Preprocessing: The speech data is framed and windowed, and Hamming windowing is applied to reduce spectral leakage; Step 2, Time-Frequency Feature Extraction: Extract Mel-frequency cepstral coefficients and their first and second-order differences to capture dynamic changes in speech, and calculate acoustic parameters such as fundamental frequency, energy, and speech rate; Step 3: Analyze the speech sequence using the BiLSTM-Attention model and combine it with CRF to optimize label prediction.

[0029] It should be noted that the reference Figure 7 As shown, text semantic feature extraction includes the following steps: Step 1, Preprocessing: Segment the text and remove stop words, and apply BERT to generate word vectors; Step 2, Context Modeling: Use the BERT architecture to capture long-distance dependencies and fine-tune the model according to the educational scenario; Step 3: Sentiment polarity analysis: Output sentiment labels through the Softmax layer and focus on key sentences using an attention mechanism.

[0030] It should be noted that the reference Figure 8 As shown, physiological signal feature extraction includes the following steps: Step 1, Signal Denoising: Use wavelet transform to remove motion artifacts in ECG and EEG; Step 2, Feature Extraction: Calculate the standard deviation of RR interval and the low-frequency to high-frequency power ratio in the electrocardiogram, and extract the peak number and rise time characteristics of the skin conductance response; Step 3: State Classification: Use random forest to distinguish the state of physiological signals.

[0031] Example 2 In a specific implementation process, the cloud computing and artificial intelligence-based higher education data deep mining system also includes a student cognitive path mapping module, which is used to identify the student's position in the knowledge graph; The student cognitive path mapping module includes: a cognitive path matching algorithm unit, which is used to calculate the matching degree between the student's learning path and the knowledge graph, transform the knowledge graph into a directed acyclic graph, where nodes are knowledge points and edges are prerequisite relationships, uses Dijkstra's algorithm to calculate the shortest edit distance between the student's learned path and the target path, and calculates the matching degree based on the shortest edit distance and the total number of nodes; in, .

[0032] The student knowledge mastery unit is used to display the student's position in the knowledge graph; The cognitive state visualization unit is used to generate a student knowledge graph profile, using color coding to indicate the degree of mastery.

[0033] According to embodiments of the present invention, a computing device that can be used to implement the above method includes a processor and a memory; The processor can be a multi-core processor or include multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more special coprocessors, such as a graphics processing unit (GPU), a digital signal processor (DSP), etc. In some embodiments, the processor may be implemented using custom circuitry, such as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA).

[0034] Memory can include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. ROM can store static data or instructions required by the processor or other modules of the computer. Permanent storage devices can be read-write storage devices. Permanent storage devices can be non-volatile storage devices that retain stored instructions and data even when the computer is powered off. In some embodiments, permanent storage devices use mass storage devices (e.g., magnetic or optical disks, flash memory) as permanent storage devices. In other embodiments, permanent storage devices can be removable storage devices (e.g., floppy disks, optical drives). System memory can be a read-write storage device or a volatile read-write storage device, such as dynamic random access memory. System memory can store some or all of the instructions and data required by the processor during operation. Furthermore, memory can include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks can also be used. In some implementations, the memory may include removable storage devices that are readable and / or writable, such as laser discs (CDs), read-only digital versatile optical discs (e.g., DVD-ROMs, dual-layer DVD-ROMs), read-only Blu-ray discs, ultra-high density optical discs, flash memory cards (e.g., SD cards, mini SD cards, Micro-SD cards, etc.), magnetic floppy disks, etc. Computer-readable storage media do not include carrier waves and transient electronic signals transmitted wirelessly or via wired connections.

[0035] It should be understood that, unless otherwise expressly stated herein, there is no strict order restriction on the execution of the above steps, and these steps may be executed in other orders. Moreover, at least some steps in the processes involved in the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.

[0036] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or basic characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects. The scope of the invention is defined by the appended claims rather than the foregoing description. Therefore, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention, and no reference numerals in the claims should be construed as limiting the scope of the claims.

[0037] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment includes only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A cloud computing and artificial intelligence-based higher education data deep mining system, characterized in that, The system comprises: a multi-modal data fusion perception module for collecting multi-modal data in an educational scene, classifying the multi-modal data through an educational scene-specific data classification system, constructing an educational scene dynamic perception engine, and performing real-time perception on learner attention state, understanding depth, and emotional changes; a dynamic knowledge graph module for extracting knowledge from the multi-modal data, constructing an initial knowledge graph, and performing dynamic evolution of the knowledge graph through time-aware graph reasoning, wherein the dynamic evolution of the knowledge graph comprises dynamically adjusting relationship weights according to knowledge usage frequency and timeliness, and constructing a knowledge value evaluation system to evaluate the value of new knowledge to teaching; an emotional computing analysis module for performing emotional computing and cognitive computing on the multi-modal data, and fusing the results of emotional computing and cognitive computing to construct an emotional and cognitive dual-dimension evaluation model for fused analysis of learner emotional state and cognitive state. 2.The cloud computing and artificial intelligence-based higher education data deep mining system according to claim 1, wherein, The multi-modal data fusion perception module comprises: a data collection unit for classifying multi-modal data into four categories of behavior representation data, neurophysiological information data, human-computer interaction data, and learning context perception data according to educational scenes; a multi-modal feature extraction unit for extracting facial expression features, speech intonation features, text semantic features, and physiological signal features; a data time alignment unit for time synchronization of multi-modal data. 3.The cloud computing and artificial intelligence-based higher education data deep mining system according to claim 1, characterized in that, The dynamic knowledge graph module comprises: a knowledge extraction and verification unit for extracting learning behavior data from multi-modal data using an AI agent, and inputting the learning behavior data into an educational knowledge base to calculate semantic similarity for verification; a dynamic weight calculation unit for dynamically adjusting relationship weights according to knowledge usage frequency and timeliness; a knowledge value evaluation unit for evaluating the value of new knowledge to teaching, including five dimensions of teaching value, learning value, knowledge correlation, timeliness, and innovation; a knowledge fusion decision unit for determining whether new knowledge should be included in the knowledge graph. 4.The cloud-computing and artificial-intelligence-based higher-education data deep-mining system according to claim 1, wherein, The emotional computing analysis module comprises: an educational emotion recognition unit for recognizing emotional features in facial expressions, speech intonation, and physiological signals; a cognitive state evaluation index unit for establishing four dimensions of knowledge point mastery rate, problem solving complexity, knowledge point review frequency, and knowledge point application breadth, and evaluating knowledge mastery level, thinking depth, and learning ability based on learning behavior data; an emotional and cognitive fusion analysis unit for fusing emotional state and cognitive state, and calculating a comprehensive cognitive state score. 5.The higher education data deep mining system based on cloud computing and artificial intelligence according to claim 1, wherein, The system further comprises a student cognitive path mapping module for identifying the position of a student in the knowledge graph; wherein the student cognitive path mapping module comprises: a cognitive path matching algorithm unit for calculating the matching degree of a student's learning path and the knowledge graph, converting the knowledge graph into a directed acyclic graph with knowledge points as nodes and prerequisite relationships as edges, using Dijkstra's algorithm to calculate the shortest edit distance between the student's learned path and the target path, and calculating the matching degree according to the shortest edit distance and the total number of nodes; a student knowledge mastery unit for displaying the position of a student in the knowledge graph; a cognitive state visualization unit for generating a student knowledge graph portrait and using color coding to represent mastery level. 6.The higher education data deep mining system based on cloud computing and artificial intelligence according to claim 2, wherein, The facial expression feature extraction includes the following steps: Step one, face detection alignment: use MTCNN algorithm to locate the face area, eliminate background interference, align the face to the standard coordinate system through affine transformation, and ensure the comparability of feature points; Step two, feature point positioning: use Dlib's 68-point model to mark key feature points, and calculate the geometric distance and motion amplitude between feature points; Step three, expression classification: based on FACS, the expression is decomposed into action unit combination, the deep features are extracted using CNN model, and the classification is combined with SVM. 7.The higher education data deep mining system based on cloud computing and artificial intelligence according to claim 2, wherein, The speech intonation feature extraction includes the following steps: Step one, preprocessing: frame and window the speech data, and apply Hamming window to reduce spectral leakage; Step two, time-frequency feature extraction: extract mel-frequency cepstral coefficients and their first and second order differences, capture speech dynamic changes, and calculate fundamental frequency, energy and speech rate acoustic parameters; Step three, use BiLSTM-Attention model to analyze speech sequence, and combine CRF to optimize label prediction. 8.The higher education data deep mining system based on cloud computing and artificial intelligence of claim 2, wherein, The text semantic feature extraction includes the following steps: Step one, preprocessing: segment the text and remove stop words, and apply BERT to generate word vectors; Step two, context modeling: use BERT architecture to capture long-distance dependency, and fine-tune the model according to the education scene; Step three, sentiment polarity analysis: output sentiment labels through Softmax layer, and combine attention mechanism to focus on key sentences. 9.The higher education data deep mining system based on cloud computing and artificial intelligence of claim 2, wherein, The physiological signal feature extraction includes the following steps: Step one, signal denoising: use wavelet transform to remove motion artifacts in electrocardiogram and electroencephalogram; Step two, feature extraction: calculate the standard deviation of RR interval and the ratio of low frequency to high frequency power in electrocardiogram, and extract the peak value and rise time features of galvanic skin response; Step three, state classification: use random forest to distinguish physiological signal states. 10.The cloud-computing and artificial-intelligence-based higher education data deep mining system according to claim 2, characterized in that, The time synchronization of the multi-modal data includes the following: generate a unified pulse signal through GPS clock, trigger multi-sensor synchronous acquisition, use cubic spline interpolation to generate virtual data under high-frequency time stamp for low-frequency data, apply sliding window average to high-frequency data, match low-frequency rhythm, take the highest frequency sensor as the main axis, and align other data to the main axis timestamp through nearest neighbor interpolation.

Images