English auxiliary teaching software multi-modal optimization method based on embedded system

Abstract

The invention discloses an English auxiliary teaching software multi-modal optimization method based on an embedded system, and aims to solve the problem that the existing embedded English auxiliary teaching software lacks systematic extraction and intelligent scheduling capabilities for key parameters such as text density, voice rhythm and image resolution. Content logic confusion, cognitive load imbalance and poor equipment adaptability are caused. The method comprises the following steps: extracting and grouping the text density, the voice rhythm and the image resolution of the multi-modal content; determining an optimization set in combination with teaching target matching and supplementary resources; constructing an initial presentation structure based on the learner cognitive level; logic continuity is guaranteed through difficulty gradient rearrangement; dynamically replacing resources according to personalized requirements; and the loading sequence is adjusted according to the low power consumption characteristic of the embedded system. By adopting the technical scheme, hierarchical, coherent and personalized presentation of teaching contents can be realized, cognitive load is reduced, a synergistic effect is improved, and efficient and stable operation on resource-constrained equipment is ensured.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of computer software combined with educational technology, and particularly relates to a multi-modal optimization method for English auxiliary teaching software based on an embedded system. BACKGROUND

[0002] With the deepening of education informatization, English auxiliary teaching software, as the core carrier of the digital transformation of language education, is gradually integrated into various intelligent terminals and embedded devices, providing learners with flexible and convenient language training environments. Embedded systems, with their low power consumption, high integration and portability, have become an important platform for mobile English teaching and are widely used in students' daily learning scenarios. However, due to the limitations of hardware resources and algorithm efficiency, existing embedded English auxiliary teaching software has obvious shortcomings in content organization logic and multi-modal presentation strategy, making it difficult to balance the dual needs of teaching scientificity and user experience.

[0003] Among them, the dynamic optimization of multi-modal teaching content is a key direction to improve the effectiveness of auxiliary teaching. English teaching materials usually integrate multiple modalities such as text, speech and images, and their effective coordination directly affects the depth of learners' understanding and cognitive load. An ideal teaching system should be able to automatically build a content sequence that is from simple to deep and logically coherent according to the learners' cognitive stage and teaching goals, and achieve reasonable matching between modalities. However, current systems generally use static preset content arrangement methods, lacking quantitative analysis and intelligent scheduling capabilities for the inherent characteristics of the materials.

[0004] Existing technologies are difficult to systematically extract and group key parameters such as text density, speech rhythm and image resolution of multi-modal resources, resulting in a lack of hierarchy and adaptability in content organization. At the same time, the teaching content and knowledge point system are disconnected, and the presentation order cannot be dynamically adjusted according to the difficulty gradient and time distribution, which easily causes cognitive overload for beginners or loss of interest for advanced learners. In addition, in the embedded environment, the system also needs to consider memory occupation and processing efficiency, while existing solutions often ignore the coordinated optimization of multi-modal attributes such as visual complexity and speech speed, resulting in significant deficiencies in the logical coherence, cognitive adaptability and device adaptability of content presentation structure. Therefore, there is an urgent need for a method that can realize intelligent organization and dynamic optimization of multi-modal English teaching content in resource-constrained embedded systems. SUMMARY

[0005] The purpose of the present application is to provide an English auxiliary teaching software multi-modal optimization method based on an embedded system, which can effectively solve the problems in the background art. The existing embedded English auxiliary teaching software lacks systematic extraction and intelligent scheduling capability for key parameters such as text density, speech rhythm and image resolution in multi-modal content organization, resulting in chaotic content presentation logic, unbalanced cognitive load, poor device adaptability, and difficulty in dynamically generating scientific, coherent and efficient learning sequences according to the learner's cognitive stage and teaching goals.

[0006] To achieve the above purpose, the technical solution adopted by the present application is: A multi-modal optimization method for English auxiliary teaching software based on an embedded system, comprising the following specific steps: Step 1: Extract the feature parameters of multi-modal content resources, the feature parameters including text density, speech rhythm and image resolution, group the feature parameters using a data grouping method to obtain a classified content resource set; Step 2: Match the classified content resource set with the preset teaching goal, the teaching goal including knowledge point coverage and interaction requirement, if the matching degree is higher than the preset threshold, the set is retained, otherwise, supplementary resources are obtained from the content resource library, the supplementary resources meet the theme relevance and diversity requirements, and the optimized content resource set is determined; Step 3: Obtain the difficulty level and time length distribution attributes of each resource in the optimized content resource set, the attributes are based on learner's cognitive level and content duration evaluation, use sequence generation method combined with progressive learning principle to construct initial content presentation structure, and obtain preliminary presentation structure; Step 4: For the difficulty level distribution in the preliminary presentation structure, the distribution reflects the gradient of resources from simple to complex, if the distribution is uneven, reorder by adjusting the position of adjacent resources, judge the logical coherence of the presentation structure to ensure smooth connection of knowledge points; Step 5: Analyze the visual complexity and speech speed attributes from the reordered presentation structure, the attributes involve image element density and pronunciation unit rate, adjust the collocation mode of the attributes using optimization model, and obtain enhanced content presentation structure; Step 6: According to the relevance of the enhanced content presentation structure and learning needs, the learning needs include individualized progress and interest preference, if the needs change, dynamically replace the resources in the presentation structure, the replacement maintains overall balance, and determine the final content presentation structure; Step 7: Obtain the running parameters of the final content presentation structure in the embedded system, the parameters including memory occupation and processing speed, adjust the loading order of the presentation structure for low power consumption characteristics, and obtain the content organization structure adapted to mobile devices.

[0007] Preferably, in step 1, the text density is calculated by the number of effective words per unit area and the syntactic complexity, the speech rhythm is quantified by the syllable interval time and the stress distribution entropy, the image resolution is comprehensively evaluated by the pixel density and the edge gradient information, and the three are normalized and input into the K-means clustering algorithm, with the cluster number set to 5, to divide the content resource set into five categories: primary, intermediate, advanced, intensive, and expanded.

[0008] Preferably, in step 2, the knowledge point coverage is calculated by comparing the matching rate of resource tags and course standard knowledge graph nodes, the interactivity requirement is determined by the number of embedded interactive elements in the resource and the response delay threshold, and the matching degree threshold is set to 85%. When the threshold is lower than this, the system retrieves supplementary resources with a theme relevance score greater than 0.9 and a diversity index greater than 0.7 from the cloud content resource library for injection.

[0009] Preferably, in step 3, the difficulty level is determined by the word frequency, sentence length average, and syntax tree depth, the time length distribution attribute limits the single resource duration to no more than 120 seconds, the sequence generation method uses a path planning algorithm based on Markov decision process, the state space is the knowledge point node, the action space is the resource selection, and the reward function integrates the difficulty gradient smoothness and the time length balance to ensure that the preliminary presentation structure conforms to the cognitive law of from simple to deep.

[0010] Preferably, in step 4, the difficulty level distribution uniformity is determined by calculating the standard deviation of the difficulty difference between adjacent resources. If the standard deviation is greater than 0.3, the local rearrangement mechanism is started to exchange resource positions in a sliding window manner within a window length of 3, and the logical coherence is verified through the knowledge point dependency relationship graph. Only when the pre-knowledge point has been fully covered, the resource position is allowed to be fixed.

[0011] Preferably, in step 5, the visual complexity is calculated by the number of objects in the image, the color saturation variance, and the texture energy, the speech speed is quantified in terms of the number of syllables per minute, the optimization model is a multi-objective constraint satisfaction problem solver, the objective function minimizes the covariance of visual complexity and speech speed, and the constraint conditions are that the speech speed is not less than 120 syllables per minute and not more than 200 syllables per minute, and the number of image objects is not more than 8.

[0012] Preferably, in step 6, the learning needs are dynamically modeled by user historical interaction logs and real-time feedback signals, the personalized progress is inferred by the proportion of completed knowledge points and error rate trends, the interest preferences are calculated by the resource click frequency and dwell time, and the dynamic replacement mechanism uses a sliding buffer strategy to replace low-relevance resources with a proportion not exceeding 20% while maintaining the overall difficulty curve and modal balance.

[0013] Preferably, in step 7, the memory occupation is obtained by static analysis of resource file size and dynamic monitoring of runtime stack peak value, the processing speed is measured according to resource decoding and rendering time consumption, the loading sequence is adjusted by adopting a hierarchical preloading strategy, high-priority resources (such as first-screen content and key knowledge points) are placed in a memory resident area, the remaining resources are sorted according to usage probability, and the delay loading threshold is set to complete loading within 500 milliseconds after user operation.

[0014] Preferably, the method deploys a lightweight feature extraction module in an embedded system, the text density calculation module occupies less than 2 megabytes of memory, the speech rhythm analysis module adopts fixed-point number operation, the processing time of a single audio segment is less than 800 milliseconds, the image resolution evaluation module is based on edge detection and region segmentation, the single-frame processing delay is less than 30 milliseconds, and the overall system can stably run on an ARM Cortex-A7 processor with a main frequency of 800 megahertz.

[0015] Preferably, the content resource library contains more than 50,000 labeled resources, each resource is associated with knowledge point labels, difficulty levels, modal attributes and teaching target mapping relationships, the system supports offline caching of commonly used resource sets, the cache capacity can be configured to 512 megabytes to 2 gigabytes, and the system can still complete the multi-modal optimization process in a network limited environment.

[0016] Compared with the prior art, the present application has the following beneficial effects: The present application extracts feature parameters such as text density, speech rhythm and image resolution of multi-modal content systematically, and combines teaching targets and learner cognitive rules to realize intelligent grouping, dynamic matching and sequence optimization of English auxiliary teaching content in an embedded environment. This method effectively solves the problems of static content organization, logical discontinuity and cognitive mismatch in existing systems, significantly improving the level of hierarchy, coherence and individualization of teaching content. At the same time, through the coordinated optimization of visual complexity and speech speed, the cognitive load of learners is reduced, and the synergistic effect of multi-modal information is enhanced. In addition, for the low power consumption and memory limited characteristics of embedded systems, the present application designs a lightweight feature extraction and hierarchical loading mechanism to ensure the optimization effect while ensuring the real-time and stability of system operation, providing efficient, scientific and device-friendly technical support for mobile English teaching. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is the overall technical scheme architecture schematic diagram of the multi-modal optimization method of English auxiliary teaching software based on an embedded system according to the present application; Figure 2 is the core principle framework schematic diagram of multi-modal feature parameter extraction and intelligent grouping in the present application; Figure 3is a content screening logic flow framework based on teaching target matching and resource optimization in the application; Figure 4 is an initial content sequence generation and rearrangement logic flow framework combined with cognitive rules and difficulty gradient in the application; Figure 5 is a multi-level interaction relationship and data flow diagram for visual complexity and speech speed cooperative optimization and embedded adaptation in the application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below in combination with specific embodiments.

[0019] Currently, under the background of the in-depth promotion of educational informatization, English auxiliary teaching software, as the core carrier of the digital transformation of language education, is gradually integrated into various intelligent terminals and embedded devices, providing a flexible and convenient language training environment for learners. Embedded systems, with the advantages of low power consumption, high integration and portability, have become an important platform for mobile English teaching and are widely used in students' daily learning scenarios. However, due to the limitations of hardware resources and algorithm efficiency, the existing embedded English auxiliary teaching software has obvious shortcomings in content organization logic and multi-modal presentation strategy, making it difficult to balance the dual needs of teaching scientificity and user experience. In view of the above technical problems, the application proposes to extract feature parameters such as text density, speech rhythm and image resolution of multi-modal content systematically, and combine teaching targets and learner cognitive rules to realize intelligent grouping, dynamic matching and sequence optimization of English auxiliary teaching content in embedded environment, effectively solving the problems of static content organization, logical discontinuity and cognitive mismatch of existing systems, and applying to a multi-modal optimization method of English auxiliary teaching software based on embedded system.

[0020] REFERENCE Figure 1 The overall technical scheme architecture of the multi-modal optimization method of English auxiliary teaching software based on embedded system proposed by the application includes a multi-modal feature extraction module, a teaching target matching module, a sequence generation and rearrangement module, a multi-modal cooperative optimization module, a dynamic replacement module and an embedded adaptation module. Each module works cooperatively to form a closed-loop optimization process, ensuring efficient, scientific and device-friendly content organization on resource-constrained embedded devices.

[0021] In the above method, the step 1 of extracting the feature parameters of the multi-modal content resources includes text density, speech rhythm and image resolution. The feature parameters are grouped by data grouping method to obtain a classified content resource set. Specifically, in the step 1, the text density is calculated by the number of effective words in a unit area and the syntax complexity. The number of effective words refers to the number of meaningful words after removing stop words, and the syntax complexity is evaluated based on the average branch factor and nesting depth of the dependency syntax analysis tree. The speech rhythm is quantified based on syllable interval time and stress distribution entropy. The syllable interval time is obtained by the time difference between the center points of adjacent syllables through forced alignment algorithm, and the stress distribution entropy is calculated according to the acoustic features (such as fundamental frequency and energy envelope) to determine the uncertainty of stress position. The image resolution is evaluated by combining pixel density and edge gradient information. The pixel density refers to the number of effective pixels in a unit area, and the edge gradient information is calculated by the mean and variance of the image gradient amplitude through Sobel operator. The above three types of feature parameters are normalized and input into the K-means clustering algorithm. The number of clustering clusters is set to 5, which is used to divide the primary, intermediate, advanced, intensive and expanded content resource sets. The normalization uses the min-max scaling method to map each parameter to the interval of 0 to 1, ensuring the comparability of different dimensional features. The initial centroid of K-means clustering is determined by the K-means++ algorithm, and the iteration termination condition is that the centroid change is less than 1e-4 or the maximum iteration number reaches 100 times. The process is executed by the lightweight feature extraction module in the embedded system. The text density calculation module occupies less than 2 megabytes of memory, the speech rhythm analysis module uses fixed-point arithmetic, and the processing time of a single audio segment is less than 800 milliseconds. The image resolution evaluation module is based on edge detection and region segmentation, and the single-frame processing delay is less than 30 milliseconds. The overall system can run stably on an ARM Cortex-A7 processor with a main frequency of 800 megahertz. Figure 2 The figure shows the core principle framework of multi-modal feature parameter extraction and intelligent grouping, including text, speech, image feature extraction channels and their fusion clustering process.

[0022] In the above method, in the step 2, the content resource set after classification is matched with the preset teaching target, the teaching target includes knowledge point coverage and interaction requirement, if the matching degree is higher than the preset threshold, the set is reserved, otherwise, supplementary resources are obtained from the content resource library, the supplementary resources meet the theme relevance and diversity requirements, and the optimized content resource set is determined. Specifically, in the step 2, the knowledge point coverage is calculated by comparing the resource label with the node matching rate of the curriculum standard knowledge graph, the curriculum standard knowledge graph is stored in the form of directed acyclic graph, the node represents the knowledge point, the edge represents the dependency relationship, and the matching rate is equal to the ratio of the number of knowledge points covered by the resource label to the total number of knowledge points required by the teaching target; the interaction requirement is determined according to the number of interactive elements in the resource and the response delay threshold, the interactive elements include multiple-choice questions, drag-and-drop questions, voice follow-up feedback, etc., and the response delay threshold is set to 300 milliseconds, that is, the system feedback time after user operation should not exceed the threshold. The matching degree threshold is set to 85%, when it is lower than the threshold, the system retrieves the supplementary resources with theme relevance score greater than 0.9 and diversity index greater than 0.7 from the cloud content resource library for injection. The theme relevance score is obtained by calculating the cosine similarity of the resource text and the teaching theme keywords, and the diversity index is calculated based on the resource modal type, difficulty level and source Shannon entropy. The content resource library contains more than 50,000 annotated resources, each resource is associated with knowledge point label, difficulty level, modal attribute and teaching target mapping relationship, the system supports offline caching of common resource set, the cache capacity can be configured to 512 megabytes to 2 gigabytes, ensuring that the multi-modal optimization process can still be completed in a network limited environment. Reference Figure 3 The figure shows the content screening logic flow framework based on teaching target matching and resource optimization, including the complete process of matching degree calculation, threshold judgment, supplementary resource retrieval and injection.

[0023] In the above method, the step 3, obtaining the difficulty level and duration distribution attribute of each resource in the optimized content resource set, the attribute is based on the learner's cognitive level and content duration evaluation, an initial content presentation structure is constructed by using a sequence generation method combined with the progressive learning principle to obtain a preliminary presentation structure. Specifically, in the step 3, the difficulty level is determined by the word frequency, the average sentence length and the syntax tree depth, the word frequency is mapped to the difficulty value from 0 to 1 according to the COCA corpus word frequency ranking, the average sentence length refers to the average number of words in a sentence, and the syntax tree depth is obtained by dependency syntax analysis to obtain the maximum nesting level of the syntax structure; the duration distribution attribute limits the single resource duration to be no more than 120 seconds, so as to avoid the learner's attention dispersion. The sequence generation method adopts a path planning algorithm based on Markov decision process, the state space is the knowledge point node, the action space is the resource selection, the reward function integrates the difficulty gradient smoothness and the duration balance, and the preliminary presentation structure is ensured to meet the cognitive law from shallow to deep. The state transition probability of Markov decision process is constructed based on the knowledge point dependency relationship diagram, and only the transition from the prerequisite knowledge point to the post-knowledge point is allowed; the reward function is defined as:

[0024] Wherein, di is the difficulty level of the i-th resource, ti is the duration of the i-th resource, and a and β are weight coefficients, which are set to 0.6 and 0.4 respectively, to ensure the comprehensive optimization of difficulty gradient smoothness and duration balance. The algorithm is solved by Q-learning, the learning rate is set to 0.1, the discount factor is set to 0.9, the exploration strategy adopts -greedy, The initial value is 0.3 and decays with the number of iterations.

[0025] In the above method, the step 4, for the difficulty level distribution in the preliminary presentation structure, the distribution reflects the gradient of resources from simple to complex, if the distribution is uneven, then reorder by adjusting the adjacent resource position, judge the logical coherence of the presentation structure to ensure smooth transition of knowledge points. Specifically, in the step 4, the difficulty level distribution uniformity is determined by calculating the standard deviation of the difficulty difference of adjacent resources, if the standard deviation is greater than 0.3, then start the local rearrangement mechanism, exchange resource positions in the window length of 3 in a sliding window manner, and verify the logical coherence through the knowledge point dependency graph, only when the pre-knowledge point has been covered, the resource position is allowed to be fixed. The sliding window starts from the sequence starting position, moves 1 resource position each time, and the resources in the window are arranged in ascending order of difficulty level, but need to meet the knowledge point dependency constraint. The knowledge point dependency graph is stored in the form of adjacency list, and the query time complexity of the pre-knowledge point is O(1). The rearrangement process is iteratively executed until the standard deviation is reduced to below 0.3 or a complete scan is completed without improvement. This mechanism ensures that the content sequence meets the smoothness of the difficulty gradient and the logical dependence of the knowledge points, avoiding cognitive jumps or knowledge gaps.

[0026] In the above method, the step 5, analyze the visual complexity and speech speed attributes from the reordered presentation structure, the attributes involve image element density and pronunciation unit rate, adjust the collocation of the attributes using an optimization model to get the enhanced content presentation structure. Specifically, in the step 5, the visual complexity is calculated by the number of objects in the image, the variance of color saturation, and the texture energy, the number of objects is obtained by a lightweight object detection model (such as MobileNet-SSD), the variance of color saturation is calculated based on the HSV color space, and the texture energy is obtained by weighting the contrast and correlation indicators of the gray level co-occurrence matrix; the speech speed is quantified in units of syllables per minute, and the number of syllables per unit time is counted by a syllable segmentation algorithm. The optimization model is a multi-objective constraint solver, the objective function minimizes the covariance of visual complexity and speech speed, the constraint conditions are that the speech speed is not less than 120 syllables per minute and not more than 200 syllables per minute, and the number of image objects is not more than 8. The covariance calculation formula is:

[0027] where vi is the visual complexity of the i-th resource, si is the speech speed of the i-th resource, and are the mean values respectively. The solver uses genetic algorithm, the population size is 50, the crossover probability is 0.8, the mutation probability is 0.1, and the maximum number of iterations is 200 generations. This optimization ensures the balanced load of visual and auditory modalities, avoiding cognitive fatigue caused by overload of a single modality. Reference Figure 5The figure shows the multi-level interaction and data flow of visual complexity and speech speed co-optimization and embedded adaptation, including attribute analysis, optimization model solving and result feedback.

[0028] In the above method, in step 6, according to the relevance of the enhanced content presentation structure and learning needs, including personalized progress and interest preferences, if the needs change, dynamically replace the resources in the presentation structure, the replacement maintains overall balance, and determines the final content presentation structure. Specifically, in step 6, the learning needs are dynamically modeled through user historical interaction logs and real-time feedback signals, the personalized progress is jointly inferred by the proportion of completed knowledge points and error rate trend, the proportion of completed knowledge points is calculated based on the number of knowledge points with correct answer rate exceeding 80%, and the error rate trend is determined by the linear regression slope of the sliding window; interest preferences are calculated based on the weighted sum of resource click frequency and dwell time, click frequency refers to the number of resource accesses per unit time, dwell time refers to the average dwell time of users on resource pages, and weight coefficients are 0.4 and 0.6 respectively. The dynamic replacement mechanism adopts a sliding buffer strategy, which replaces low-relevance resources at a rate not exceeding 20% while maintaining overall difficulty curve and modal balance. Low-relevance resources are defined as resources with interest preference scores lower than the current sequence mean by 0.5 standard deviations. When replacing, resources with similar difficulty levels (difference not exceeding 0.2), same modal types and highest interest preference scores are selected from the cache resource library to ensure the stability of the overall sequence structure.

[0029] In the above method, in step 7, the running parameters of the final content presentation structure in the embedded system are obtained, including memory occupation and processing speed, the loading order of the presentation structure is adjusted for low-power characteristics, and the content organization structure suitable for mobile devices is obtained. Specifically, in step 7, memory occupation is obtained through static analysis of resource file size and dynamic monitoring of runtime stack peak value, static analysis is completed when resources are warehoused, and dynamic monitoring is realized through real-time sampling of the memory management unit of the embedded operating system; processing speed is measured according to resource decoding and rendering time, decoding time refers to the time from reading from storage medium to decompression completion in memory, and rendering time refers to the processing time from memory data to screen display. The loading order adjustment adopts a hierarchical preloading strategy, placing high-priority resources (such as first-screen content and key knowledge points) in the memory resident area, and sorting the remaining resources according to usage probability, with a delay loading threshold set to complete loading within 500 milliseconds after user operation. Usage probability is calculated based on a Markov chain prediction model, with state as the current knowledge point and transition probability as the knowledge point transition frequency in the historical sequence. This strategy significantly reduces memory peak occupation and improves system response speed, ensuring smooth operation on low-power embedded devices.

[0030] To verify the effectiveness of the present application, a specific application scenario is taken as an example: a second-year junior high school student uses a tablet computer equipped with the method of the present application to learn English grammar, and the teaching goal is to master the present perfect tense. The system first loads 500 multi-modal resources related to the present perfect tense from the local cache library, performs step 1 to extract features and group, and obtains five categories of sets: elementary (120), intermediate (180), advanced (100), reinforcement (60), and expansion (40). Step 2 calculates the knowledge point coverage as 82%, which is lower than the 85% threshold, and the system retrieves 30 supplementary resources from the cloud with a theme relevance score greater than 0.9 and a diversity index greater than 0.7, and the coverage is increased to 88%. Step 3 assesses the student's cognitive level as intermediate based on the historical error rate (45% for present perfect tense related questions), and generates a preliminary presentation structure containing 20 resources with difficulty levels ranging from 0.4 to 0.7, with a smooth increase in difficulty level and a single resource duration of less than 90 seconds. Step 4 detects a difficulty standard deviation of 0.35, and starts the rearrangement mechanism, adjusts the position of 3 resources, and the standard deviation decreases to 0.28, and all knowledge point dependencies are met. Step 5 optimizes, and the visual complexity and speech speed covariance decreases from 0.15 to 0.05, the speech speed is stable at 150 syllables per minute, and the number of image objects is not more than 6. Step 6 replaces 2 low-relevance text resources with animation resources based on the student's recent high click frequency (3 times per day) and long dwell time (average 120 seconds) on animation resources. Step 7 places the first screen's 3 key resources (present perfect tense basic structure, common verb changes, and typical example sentences) in the memory resident area, and the remaining resources are sorted by usage probability to ensure that the 500 millisecond loading is completed when sliding through the pages. The entire process takes 4.2 seconds on an 800 megahertz ARM Cortex-A7 processor, with a peak memory usage of 180 megabytes, significantly better than existing static arrangement schemes.

[0031] The above shows and describes the basic principles and main features of the present application and the advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A multimodal optimization method for English teaching software based on embedded systems, characterized in that: The specific steps include the following: Step 1: Extract feature parameters of multimodal content resources, including text density, speech rhythm and image resolution. Use data grouping method to group the feature parameters to obtain a classified set of content resources. Step 2: Match the categorized content resource set with the preset teaching objectives, which include knowledge point coverage and interactivity requirements. If the matching degree is higher than the preset threshold, the set is retained; otherwise, supplementary resources are obtained from the content resource library. The supplementary resources meet the requirements of topic relevance and diversity, thus determining the optimized content resource set. Step 3: Obtain the difficulty level and duration distribution attributes of each resource in the optimized content resource set. The attributes are evaluated based on the learner's cognitive level and the content duration. The initial content presentation structure is constructed by using a sequence generation method combined with the principle of progressive learning. Step 4: For the initial presentation structure, the distribution of difficulty levels reflects the gradient of resources from simple to complex. If the distribution is uneven, the adjacent resource positions are adjusted and reordered. The logical coherence of the presentation structure is judged to ensure smooth connection of knowledge points. Step 5: Analyze the visual complexity and speech speed attributes from the reordered presentation structure. These attributes involve the density of image elements and the rate of speech unit. Use an optimization model to adjust the combination of these attributes to obtain the enhanced content presentation structure. Step 6: Based on the relevance between the enhanced content presentation structure and learning needs, including personalized progress and interest preferences, if the needs change, the resources in the presentation structure are dynamically replaced, and the replacement maintains overall balance to determine the final content presentation structure. Step 7: Obtain the operating parameters of the final content presentation structure in the embedded system. The parameters include memory usage and processing speed. Adjust the loading order of the presentation structure for low power consumption characteristics to obtain a content organization structure adapted to mobile devices.

2. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 1, text density is calculated by weighting the number of effective words per unit area with syntactic complexity, speech rhythm is quantified based on syllable interval time and stress distribution entropy, and image resolution is comprehensively evaluated by combining pixel density and edge gradient information. After normalization, the three are input into the K-means clustering algorithm, with the number of clusters set to 5, which is used to divide the content resource sets into five categories: primary, intermediate, advanced, enhanced, and extended.

3. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 2, the knowledge point coverage is calculated by comparing the node matching rate between the resource tags and the knowledge graph of the curriculum standard. The interactivity requirement is determined based on the number of embedded interactive elements in the resource and the response latency threshold. The matching threshold is set to 85%. When it is lower than this threshold, the system retrieves supplementary resources from the cloud content resource library with a topic relevance score greater than 0.9 and a diversity index greater than 0.7 and injects them.

4. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 3, the difficulty level is determined by the frequency of words, the average sentence length, and the depth of the syntax tree. The duration distribution attribute limits the duration of a single resource to no more than 120 seconds. The sequence generation method adopts a path planning algorithm based on Markov decision process. The state space is the knowledge point node, the action space is the resource selection, and the reward function integrates the smoothness of the difficulty gradient and the balance of duration to ensure that the initial presentation structure conforms to the cognitive law of gradual deepening.

5. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 4, the uniformity of the difficulty level distribution is determined by calculating the standard deviation of the difficulty difference between adjacent resources. If the standard deviation is greater than 0.3, a local rearrangement mechanism is initiated, and resource positions are exchanged within a window length of 3 using a sliding window. Logical coherence is verified through a knowledge point dependency graph. The resource position is only allowed to be fixed when all the preceding dependency nodes of the subsequent knowledge point have been covered.

6. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 5, visual complexity is calculated by combining the number of objects in the image, color saturation variance, and texture energy. Speech speed is quantified in terms of syllables per minute. The optimization model is a multi-objective constraint satisfaction problem solver. The objective function minimizes the covariance between visual complexity and speech speed. The constraints are that the speech speed is not less than 120 syllables per minute and not more than 200 syllables per minute, and the number of image objects does not exceed 8.

7. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 6, learning needs are dynamically modeled using user historical interaction logs and real-time feedback signals. Personalized progress is inferred from the combined proportion of completed knowledge points and error rate trends. Interest preferences are calculated based on a weighted average of resource click frequency and dwell time. The dynamic replacement mechanism adopts a sliding buffer strategy, replacing low-relevance resources at a rate not exceeding 20% ​​while maintaining the overall difficulty curve and modal balance.

8. The multimodal optimization method for English teaching software based on an embedded system according to claim 1, characterized in that: In step 7, memory usage is obtained by statically analyzing the size of resource files and dynamically monitoring runtime stack peaks. Processing speed is based on the measurement of resource decoding and rendering time. Loading order adjustment adopts a layered preloading strategy, placing high-priority resources in the resident memory area and sorting the remaining resources according to their usage probability. The delayed loading threshold is set to complete loading within 500 milliseconds after the user operation.

9. The multimodal optimization method for English teaching software based on an embedded system according to claim 2, characterized in that: The text density calculation module occupies less than 2 megabytes of memory, the speech rhythm analysis module uses fixed-point arithmetic, and the processing time for a single audio segment is less than 800 milliseconds. The image resolution evaluation module is based on edge detection and region segmentation, and the single-frame processing latency is less than 30 milliseconds. The entire system can run stably on an ARM Cortex-A7 processor with a main frequency of 800 MHz.

10. The multimodal optimization method for English teaching software based on an embedded system according to claim 3, characterized in that: The content resource library contains more than 50,000 labeled resources. Each resource is associated with knowledge point tags, difficulty level, modal attributes, and teaching objective mapping relationships. The system supports offline caching of commonly used resource sets, and the cache capacity can be configured from 512 megabytes to 2 gigabytes.

Images