College student education teaching material aided design method and system based on knowledge sharing
By using a knowledge-sharing approach, word embedding models and hierarchical clustering are used to construct isomorphic graphs. Combined with learning data and multimodal resources, personalized textbook design schemes are generated, which solves the problem of insufficient logic and systematicity in traditional textbooks and improves learning efficiency and textbook adaptability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-03-10
AI Technical Summary
Traditional digital teaching materials are not perfect in terms of logic and system, lack personalized adjustments, have low learning efficiency, slow update response, and are difficult to meet the needs of different learners.
By acquiring multiple published textbooks, using word embedding models and hierarchical clustering to identify knowledge points, constructing isomorphic graphs, and combining learning data and multimodal resources, personalized textbook design schemes are generated.
It achieves a comprehensive and logically coherent learning path, improves learning efficiency, shortens the learning cycle, and promotes the iterative optimization of teaching materials and the sharing of resources.
Smart Images
Figure CN121636701A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of big data analytics, and in particular to a method and system for assisting in the design of educational materials for university students based on knowledge sharing. Background Technology
[0002] The development of digital textbooks aims to optimize the creation, editing, distribution, and use of teaching materials using digital technology. This approach seeks to improve the accessibility and interactivity of textbooks while ensuring content quality and educational effectiveness. In the digital age, this method emphasizes control over content quality and sensitivity to reader needs to provide more effective and engaging learning materials. Digital textbook development utilizes content management systems to organize and edit materials, leverages data analytics tools to analyze and predict user needs, and implements interactive and multimedia elements to enhance the learning experience. It emphasizes copyright protection and intellectual property management to ensure the rights of creators and publishers are protected, while also providing users with legal and high-quality educational resources. Through these means, the development of digital textbooks aims to create a more effective, interactive, and accessible learning environment.
[0003] Traditional digital textbook writing methods lack in-depth analysis of the complex relationships between knowledge points during the textbook production process, resulting in imperfections in the logic and systematic nature of the textbook content. The learning paths are relatively fixed, lacking personalized adjustments for different learners, leading to low learning efficiency and a diminished learning experience. Traditional methods are also slow to respond to textbook updates and assessments, making it difficult to adjust in a timely manner based on student feedback, which limits the real-time nature and adaptability of the textbook content.
[0004] Therefore, there is a need to provide a knowledge-sharing-based method and system for assisting in the design of university education textbooks, in order to improve the level of intelligence in assisting in the design of university education textbooks. Summary of the Invention
[0005] This invention provides a knowledge-sharing-based method for assisting in the design of educational textbooks for university students, comprising: acquiring multiple published textbooks for the subject to be designed; determining multiple knowledge points of the subject based on the multiple published textbooks; acquiring shared teaching feedback data for the subject; constructing an isomorphic graph of the multiple knowledge points based on the shared teaching feedback data; acquiring learning data of multiple students in related subjects; acquiring multimodal learning support resources for the multiple knowledge points; and generating a textbook design scheme for the subject based on the learning data of multiple students in related subjects, the isomorphic graph of the multiple knowledge points, and the multimodal learning support resources for the multiple knowledge points.
[0006] Furthermore, based on multiple published textbooks of a subject, multiple knowledge points of the subject to be designed are identified, including: for each published textbook, using a word embedding model, converting paragraphs in the published textbook into numerical vectors; based on the numerical vectors, clustering the paragraphs in the published textbook to generate clustering results; based on the clustering results, identifying multiple knowledge points of the published textbook to be merged; and based on the multiple knowledge points of each published textbook to be merged, identifying multiple knowledge points of the subject to be designed.
[0007] Furthermore, based on numerical vectors, the paragraphs in the published textbook are clustered to generate clustering results. Based on the clustering results, multiple knowledge points to be merged in the published textbook are identified, including: for any two paragraphs, calculating the cosine similarity of their numerical vectors; clustering the paragraphs in the published textbook based on the cosine similarity of their numerical vectors using hierarchical clustering to generate clustering results, where the clustering results include multiple paragraph clusters; for each paragraph cluster, based on a subject terminology database, counting multiple terms in each paragraph included in the cluster and the word frequency of each term; determining candidate terms for the paragraph cluster based on the multiple terms in each paragraph; calculating the correlation coefficient between any two candidate terms based on the word frequencies of any two candidate terms in each paragraph included in the cluster; and using the first optimized particle swarm optimization algorithm, determining multiple key terms for the knowledge points corresponding to the paragraph cluster based on the correlation coefficient between any two candidate terms.
[0008] Furthermore, based on the multiple knowledge points to be merged in each published textbook, multiple knowledge points of the subject to be designed are determined, including: based on the multiple key terms of the multiple knowledge points to be merged in each published textbook, the multiple knowledge points to be merged in each published textbook are deduplicated to determine the multiple knowledge points of the subject to be designed.
[0009] Furthermore, the shared teaching feedback data of the subject to be designed includes the mastery levels of multiple students on multiple knowledge points of the subject; based on the shared teaching feedback data of the subject, an isomorphic graph of multiple knowledge points is constructed, including: for any two knowledge points, calculating the correlation coefficient between the two knowledge points based on the mastery levels of multiple students on the two knowledge points; for each knowledge point, determining the associated knowledge points based on the correlation coefficient between any two knowledge points; based on the associated knowledge points of each knowledge point, constructing an isomorphic graph of multiple knowledge points, wherein the isomorphic graph includes multiple nodes and multiple edges, each node represents a knowledge point, the edges are used to connect two nodes, and the weight of the edge represents the correlation coefficient between the two knowledge points corresponding to the two connected nodes.
[0010] Furthermore, the related subjects are identified, including: identifying multiple subjects to be screened; obtaining the mastery of multiple knowledge points of multiple subjects to be screened by multiple students; calculating the correlation coefficient between the subjects to be screened and the subjects to be designed based on the mastery of multiple knowledge points of multiple subjects to be designed by multiple students and the mastery of multiple knowledge points of the subjects to be screened; and determining the related subjects based on the correlation coefficient between the subjects to be screened and the subjects to be designed.
[0011] Furthermore, based on the mastery of multiple knowledge points in the design discipline and the mastery of multiple knowledge points in the subject to be screened by multiple students, the correlation coefficient between the subject to be screened and the subject to be designed is calculated, including: based on the mastery of multiple knowledge points in the design discipline and the mastery of multiple knowledge points in the subject to be screened by multiple students, the correlation coefficient between any knowledge point in the subject to be designed and any knowledge point in the subject to be screened is calculated; based on the correlation coefficient between any knowledge point in the subject to be designed and any knowledge point in the subject to be screened is calculated, the correlation coefficient between the subject to be screened and the subject to be designed is calculated.
[0012] Furthermore, based on the learning data of multiple students in related subjects, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points, a teaching material design scheme for the subject is generated, including: determining the weights of multiple knowledge points based on the learning data of multiple students in related subjects; and generating a teaching material design scheme for the subject based on the weights of multiple knowledge points, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points through a second optimized particle swarm optimization algorithm.
[0013] Furthermore, through a second optimized particle swarm optimization algorithm, based on the weights of multiple knowledge points, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points, a subject-specific educational textbook design scheme is generated. This includes: determining the utilization characteristics of the multimodal auxiliary learning resources of knowledge points based on the weights of multiple knowledge points; initializing the particle swarm, where each particle represents a learning path; establishing a second optimized fitness function, where the second optimized fitness function is related to the isomorphic graph of multiple knowledge points, the utilization characteristics of the multimodal auxiliary learning resources of knowledge points, and the weights of multiple knowledge points; and optimizing the particle swarm based on the second optimized fitness function to generate the subject-specific educational textbook design scheme.
[0014] This invention provides a knowledge-sharing-based textbook design assistance system for university students, applying the aforementioned knowledge-sharing-based textbook design assistance method. The system includes: a textbook acquisition module for acquiring multiple published textbooks for a subject; a knowledge point analysis module for identifying multiple knowledge points based on these textbooks; a data acquisition module for acquiring shared teaching feedback data for the subject; the knowledge point analysis module for determining isomorphic graphs of multiple knowledge points based on the shared teaching feedback data; the data acquisition module for acquiring learning data of multiple students in related subjects; the data acquisition module for acquiring multimodal supplementary learning resources for multiple knowledge points; and an assistance design module for generating a textbook design scheme for the subject based on the learning data of multiple students in related subjects, the isomorphic graphs of multiple knowledge points, and the multimodal supplementary learning resources for multiple knowledge points.
[0015] Compared to existing technologies, the knowledge-sharing-based auxiliary design method and system for university student educational materials provided in this specification has at least the following beneficial effects: 1. By integrating multiple published textbooks and shared teaching feedback data, we ensure comprehensive coverage of knowledge points and meet actual teaching needs, avoiding subjective experience biases and improving the scientific and practical nature of textbook design. The isomorphic diagrams constructed based on shared feedback data can intuitively reflect the strength of connections between knowledge points and teaching difficulties, assisting in the design of logically coherent and hierarchical learning paths, strengthening the systematic nature of the knowledge system. Combining student learning data (such as learning progress and ability level) with multimodal resources (videos, simulations, interactive tools), we can customize differentiated textbook content for different student groups, meet diverse learning needs, and improve learning efficiency. The introduction of multimodal resources enriches teaching methods, while the selection of suitable resources through learning data avoids resource redundancy, shortens the learning cycle, reduces cognitive load, and integrates teaching feedback across textbooks and disciplines, promoting the sharing and reuse of educational resources, driving iterative optimization of textbook design, and forming a virtuous cycle of "feedback-improvement". 2. The semantic association of paragraphs is quantified by measuring cosine similarity of numerical vectors, and paragraph clusters are automatically divided using hierarchical clustering, avoiding the subjectivity of manual division and ensuring the objectivity and accuracy of knowledge point merging. Based on a subject-specific terminology database, statistical term frequencies are used to focus on core domain concepts and avoid interference from common vocabulary; correlation coefficients are used to quantify the strength of associations between terms, selecting closely related key terms to improve the professionalism and coherence of knowledge point representation. The first optimized particle swarm optimization algorithm (combining term relevance and average term frequency) dynamically balances the internal relevance and external saliency of terms, avoiding local optima and generating an optimal set of key terms, enhancing the core nature and interpretability of knowledge points. 3. By calculating the correlation coefficients and association coefficients between the knowledge points of the subject to be designed and the subjects to be selected, the implicit connections between subjects are transformed into quantifiable numerical indicators, avoiding subjective judgment bias and improving the scientific rigor of the selection of related subjects. By analyzing the differences in students' mastery of knowledge points in different subjects, related subjects with strong complementarity to the subject to be designed are identified, providing a basis for personalized textbook design and improving learning outcomes. Attached Figure Description
[0016] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1 This is a flowchart illustrating a knowledge-sharing-based auxiliary design method for university student educational materials in one embodiment of this application; Figure 2 This is a module diagram of a knowledge-sharing-based auxiliary design system for university student educational materials, as shown in one embodiment of this application. Detailed Implementation
[0017] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below.
[0018] Figure 1 This is a flowchart illustrating a knowledge-sharing-based method for assisting in the design of university education textbooks, as shown in one embodiment of this application. Figure 1 As shown, a knowledge-sharing-based method for assisting in the design of university educational textbooks may include the following process. This knowledge-sharing-based method for assisting in the design of university educational textbooks belongs to the field of big data analytics.
[0019] S101. Obtain multiple published textbooks for the subject to be designed.
[0020] The discipline to be designed can be a science and engineering discipline. For example, the discipline to be designed could be mathematics and applied mathematics, and corresponding published textbooks could include *Higher Algebra* (Higher Education Press, edited by Yang Yichuan and Zhou Meng) and *Mathematical Analysis (Volume 2)* (Beijing University of Aeronautics and Astronautics Press, edited by Gao Zongsheng et al.). Another example is physics, and corresponding published textbooks could include *General Course in University Physics: Mechanics* (Peking University Press, edited by Zhong Xihua and Zhou Yueming) and *New Concept Physics Course: Quantum Physics* (Higher Education Press, edited by Zhao Kaihua).
[0021] Multiple textbooks for design disciplines can be identified through manual or big data analysis.
[0022] S102. Determine multiple knowledge points of the discipline to be designed based on multiple published textbooks of the discipline to be designed.
[0023] Specifically, it includes: For each published textbook, convert the paragraphs in the published textbook into numerical vectors through a word embedding model (such as Word2Vec, BERT, etc.), cluster the paragraphs in the published textbook based on the numerical vectors to generate a clustering result, and determine multiple knowledge points to be merged in the published textbook based on the clustering result; Determine multiple knowledge points of the discipline to be designed based on the multiple knowledge points to be merged in each published textbook.
[0024] Specifically, first segment the textbook paragraphs, remove stop words (such as "of", "is"), and retain professional terms (such as "quantum entanglement", "neural network"). For example, for the paragraph "Quantum entanglement is a phenomenon in quantum mechanics, referring to the state of two particles being correlated with each other", only after preprocessing, retain the keywords "quantum entanglement", "quantum mechanics", "particle state".
[0025] For example: Convert paragraph 1 (quantum entanglement) into a numerical vector [0.2, -0.5, 0.8,...] through a word embedding model, and convert paragraph 2 (quantum computing) into a numerical vector [0.3, -0.4, 0.9,...].
[0026] Preferably, based on the numerical vectors, cluster the paragraphs in the published textbook to generate a clustering result, and based on the clustering result, determine multiple knowledge points to be merged in the published textbook, including: For any two paragraphs, calculate the cosine similarity of the numerical vectors of the two paragraphs; Cluster the paragraphs in the published textbook through hierarchical clustering based on the cosine similarity of the numerical vectors of any two paragraphs to generate a clustering result, where the clustering result includes multiple paragraph clusters; For each paragraph cluster, based on the subject term library, count multiple terms in each paragraph included in the paragraph cluster and the word frequency of each term, determine the candidate terms of the paragraph cluster based on the multiple terms of each paragraph, calculate the correlation coefficient of any two candidate terms based on the word frequency of any two candidate terms in each paragraph included in the paragraph cluster, and determine multiple key terms of the knowledge point corresponding to the paragraph cluster through the first optimized particle swarm algorithm.
[0027] Specifically, clustering the paragraphs in the published textbook through hierarchical clustering based on the cosine similarity of the numerical vectors of any two paragraphs to generate a clustering result may include the following steps: 1. Initialization: Consider each paragraph as an independent cluster.
[0028] 2. Merging Strategy: Calculate pairwise similarity between all clusters and merge the two clusters with the highest similarity. Repeat until all paragraphs are merged into one cluster (or the preset number of clusters is reached).
[0029] 3. Stopping conditions: Threshold method: Stop when the minimum similarity between clusters is lower than a threshold (e.g., 0.3). Cluster number method: Directly specify the number of clusters (e.g., determined by the elbow rule).
[0030] 4. Output: Multiple paragraph clusters, each cluster representing a candidate knowledge point.
[0031] For example, paragraph cluster 1 contains paragraphs such as "quantum entanglement" and "quantum superposition," corresponding to the candidate knowledge point "fundamentals of quantum mechanics." Paragraph cluster 2 contains paragraphs such as "neural network structure" and "backpropagation algorithm," corresponding to the candidate knowledge point "principles of deep learning."
[0032] The subject-specific terminology database can include multiple professional terms for the subject to be designed (such as "quantum entanglement" and "neural network"). The non-professional terms in the paragraph are filtered out using the subject-specific terminology database, and the remaining terms are used as multiple terms in the paragraph.
[0033] The correlation coefficient is calculated using correlation coefficient formulas (e.g., Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) based on the word frequency of any two candidate terms in each paragraph included in the paragraph cluster.
[0034] Using the first optimized particle swarm optimization algorithm, multiple key terms for the knowledge points corresponding to paragraph clusters are determined based on the correlation coefficient between any two candidate terms. This can include the following steps: S11. Particle encoding: Each particle is a binary vector with a length equal to the total number of candidate terms, N. A value of 1 indicates that the term is selected, and 0 indicates that it is not selected. For example, if the candidate terms are [quantum entanglement, quantum computing, Bell's inequality, superposition], then the particle [1,1,0,1] represents the selection of the first three terms. Specifically, M particles are randomly generated, ensuring that each particle contains at least one term (avoiding empty combinations). High-probability selection of high-frequency terms is initialized based on the median term frequency in the paragraph cluster (accelerating convergence).
[0035] S12. Establish the first optimized fitness function, wherein the first optimized fitness function is related to the correlation coefficient of any two candidate terms included in the particle and the mean word frequency of the candidate terms; S13. Recording Individual and Global Optimals: Individual Optimal (pbest): The position with the best fitness in the history of each particle. Global Optimal (gbest): The position with the best fitness among all particles; S14. Determine if the termination condition is met, for example, reaching the maximum number of iterations (e.g., 100 times) or the fitness value changing less than a threshold for 10 consecutive iterations (e.g., ...). If not, determine multiple key terms for the knowledge points corresponding to the paragraph cluster based on the particle with the highest fitness; if not, proceed to S15. S15, Update particle position and velocity, then execute S13.
[0036] For example, the fitness function for the first optimization is:
[0037] in, The fitness function is the first optimized one. and To preset weights, and Greater than 0, ,For example, It is 0.6. It is 0.4. Let be the correlation coefficient between the i-th candidate term and the j-th candidate term included in the particle. The total number of candidate terms included in the particle. Let be the average word frequency of the i-th candidate term included in the particle, which is the average word frequency of the i-th candidate term included in the particle cluster in each paragraph.
[0038] Understandably, the above formula, through Calculate the average correlation coefficient among all candidate term pairs within a particle to measure the overall correlation of candidate terms within the particle. A higher value indicates a stronger correlation between candidate terms. Calculate the average word frequency of each term in the particle to measure the general importance of the terms. Maximize This is equivalent to simultaneously optimizing the internal tightness (high relevance) and external salience (high word frequency) of terms.
[0039] The semantic associations of paragraphs are quantified by measuring cosine similarity of numerical vectors, and paragraph clusters are automatically divided by hierarchical clustering, avoiding the subjectivity of manual division and ensuring the objectivity and accuracy of knowledge point merging. Based on the statistical frequency analysis of terminology in a subject-specific terminology database, core concepts in the field are focused on, avoiding interference from common vocabulary; the correlation coefficient is used to quantify the strength of associations between terms, selecting closely related key terms to improve the professionalism and coherence of knowledge point representation. Utilizing the first optimized particle swarm optimization algorithm (combining terminology relevance and mean word frequency), the internal relevance and external saliency of terms are dynamically balanced to avoid local optima and generate an optimal set of key terms, enhancing the core nature and interpretability of knowledge points.
[0040] As a preferred approach, based on multiple knowledge points to be merged from each published textbook, multiple knowledge points for the subject to be designed are identified, including: Based on multiple key terms of multiple knowledge points to be merged in each published textbook, duplicates of multiple knowledge points to be merged in each published textbook are removed to determine multiple knowledge points of the subject to be designed.
[0041] Specifically, through deduplication, repetitive or highly similar knowledge points from multiple textbooks are integrated into unique, independent knowledge points, ultimately forming a standardized knowledge point system for the subject to be designed. The same knowledge point may be split into multiple knowledge points in multiple textbooks due to differences in expression or contextual expansion (e.g., "Newton's First Law" and "Law of Inertia"). By matching and merging key terms, the uniqueness and consistency of subject knowledge points are ensured, providing a basic framework for subsequent course design and syllabus development. Key terms for each knowledge point are vectorized (e.g., using TF-IDF, Word2Vec, or BERT embedding), converting the term set into numerical vectors to facilitate quantifying similarity. The cosine of the angle between the numerical vectors of two knowledge points to be merged is calculated to measure their semantic similarity. If the cosine of the angle between the numerical vectors of two knowledge points to be merged is greater than a threshold (e.g., 0.8), one of them is retained.
[0042] S103. Obtain shared teaching feedback data for the subject to be designed.
[0043] Among them, the shared teaching feedback data of the subject includes the mastery of multiple knowledge points of the design subject by multiple students.
[0044] Specifically, it can acquire learning data from multiple students regarding design subjects, covering different teaching scenarios (classroom tests, assignments, exams, online learning platforms, etc.) and formats (objective questions, subjective questions, practical assessments, etc.), including students' accuracy rate and completion time for each knowledge point. By weighting the students' accuracy rate and completion time for each knowledge point, the system can calculate the students' mastery of multiple knowledge points in the design subject.
[0045] For knowledge points that were not assessed or had missing data, the following strategy was adopted: If a student did not participate in the assessment related to that knowledge point, it was marked as "not assessed". If some data was missing (such as completion time), it was filled in with the mean or median of other students for the same knowledge point.
[0046] S104. Based on the shared teaching feedback data of the subject to be designed, construct isomorphic diagrams of multiple knowledge points.
[0047] Specifically, it includes: For any two knowledge points, calculate the correlation coefficient between the two knowledge points based on the mastery of the two knowledge points by multiple students; For each knowledge point, the related knowledge points are determined based on the correlation coefficient between any two knowledge points; Based on the associated knowledge points of each knowledge point, an isomorphic graph of multiple knowledge points is constructed. The isomorphic graph includes multiple nodes and multiple edges. Each node represents a knowledge point, and the edges are used to connect two nodes. The weight of the edge represents the correlation coefficient between the two knowledge points corresponding to the two connected nodes.
[0048] Specifically, the correlation coefficient between two knowledge points can be calculated using correlation coefficient formulas (e.g., Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) based on the degree of mastery of two knowledge points by multiple students.
[0049] If the correlation coefficient between two knowledge points is greater than the correlation coefficient threshold (e.g., 0.5), then the nodes corresponding to the two knowledge points are connected by an edge.
[0050] S105. Obtain the learning data of multiple students corresponding to related subjects.
[0051] As a preferred option, related disciplines are identified, including: Identify multiple disciplines to be screened; To obtain the mastery of multiple knowledge points of the selected subject by multiple students; Based on the mastery of multiple knowledge points of the design discipline by multiple students and the mastery of multiple knowledge points of the selected discipline, the correlation coefficient between the selected discipline and the design discipline is calculated. Based on the correlation coefficient between the disciplines to be screened and the disciplines to be designed, the related disciplines are determined.
[0052] Specifically, multiple candidate disciplines can be identified through manual methods or big data analysis. For example, if the subject to be designed is computer science, multiple candidate disciplines could include statistics, advanced mathematics, etc.
[0053] As a preferred approach, based on the mastery of multiple knowledge points in the design discipline by multiple students and the mastery of multiple knowledge points in the subject to be selected, the correlation coefficient between the subject to be selected and the subject to be designed is calculated, including: Based on the mastery of multiple knowledge points in the design discipline and the mastery of multiple knowledge points in the selected discipline by multiple students, the correlation coefficient between any knowledge point in the design discipline and any knowledge point in the selected discipline is calculated. Based on the correlation coefficient between any knowledge point of the subject to be designed and any knowledge point of the subject to be screened, calculate the correlation coefficient between the subject to be screened and the subject to be designed.
[0054] The correlation coefficient can be calculated using correlation coefficient formulas (e.g., Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) based on the mastery of a knowledge point in the design subject and a knowledge point in the subject to be screened by multiple students.
[0055] For any knowledge point in the subject to be designed, the weight of the knowledge point can be determined based on the isomorphism graph of multiple knowledge points, and the maximum value of the correlation coefficient between the knowledge point and any knowledge point in the subject to be screened can be determined. Specifically, the weight of the knowledge point can be calculated based on the average of the edge connections between the knowledge point and the knowledge points corresponding to the connected nodes.
[0056] Based on the weight of each knowledge point in the subject to be designed and the maximum value of the correlation coefficient between the knowledge point and any knowledge point in the subject to be screened, the correlation coefficient between the subject to be screened and the subject to be designed is calculated.
[0057] For example, the correlation coefficient between the subjects to be screened and the subjects to be designed can be calculated using the following formula:
[0058] in, Let be the correlation coefficient between the m-th subject to be screened and the subject to be designed. The total number of knowledge points in the subject to be designed. Let k be the weight of the k-th knowledge point in the subject to be designed. Let k be the maximum value of the k-th knowledge point in the subject to be designed. Let be the total number of edges connecting the node corresponding to the k-th knowledge point of the subject to be designed. This represents the total number of edges in an isomorphic graph containing multiple knowledge points. Let be the correlation coefficient between the k-th knowledge point and the q-th knowledge point in the subject to be designed. The total number of knowledge points included in the subject to be designed. Let be the correlation coefficient between the k-th knowledge point of the subject to be designed and the p-th knowledge point of the m-th subject to be screened. Let m be the total number of knowledge points included in the m-th subject to be screened. To obtain the maximum value.
[0059] Understandably, in the above formula, the weight of knowledge point k is proportional to the number of connected edges in the isomorphic graph, reflecting its coreness in the subject knowledge network. The average correlation coefficient between knowledge point k and other knowledge points within the subject reflects the strength of its content association with other knowledge points within the subject; the higher the average correlation coefficient, the closer the logical association of the knowledge point within the subject. Combining structural and content contributions ensures that the weight reflects both the position of the knowledge point in the knowledge network and its inherent logical relationship with other knowledge points within the subject. The maximum correlation coefficient between knowledge point k and all knowledge points in the subject to be screened, m, reflects the strongest association direction between knowledge point k and the subject to be screened. The association coefficient between the subject to be screened and the subject to be designed is the weighted sum of the weights of each knowledge point and their corresponding maximum correlation coefficients. By comprehensively considering the structural importance (weight) of all knowledge points within the subject and the strongest association with the subject to be screened, the overall association strength between the two subjects is quantified. The weight simultaneously considers the position (structure) of the knowledge point in the knowledge network and its logical relationship (content) with other knowledge points within the subject, avoiding single-dimensional bias. For example, if knowledge point k has many connecting edges (structural core) but low correlation coefficients with other knowledge points (loose content), its weight will be adjusted appropriately to prevent overestimation. By taking the maximum value, weak correlations between knowledge points and the subjects to be screened are ignored, highlighting the most representative correlation directions. For example, if knowledge point k has a strong correlation with "regression analysis" in "statistics" but a weak correlation with "probability theory," only the strong correlation signal of "regression analysis" will be retained. Transforming subject correlation into objective calculations based on student mastery data and knowledge networks enhances the scientific rigor and interpretability of interdisciplinary correlation analysis.
[0060] S106. Obtain multimodal auxiliary learning resources for multiple knowledge points.
[0061] As an example only, the types of multimodal auxiliary learning resources for knowledge points can be shown in Table 1.
[0062] Table 1 Resource Type Form Examples Core Role Text Resources Academic Papers, Blog Articles Provide Systematic Knowledge Framework with Detailed Explanations Video Resources Micro-Lessons, Animation Demonstrations, Expert Lectures Enhance Understanding and Memory through Audio-Visual Combination Interactive Simulation Resources Virtual Labs, Programming Sandboxes, 3D Model Operations Deepen Cognitive and Application Abilities through Hands-On Experience Graphical Image Resources Mind Maps, Information Diagrams, Conceptual Diagrams Intuitively Display Relationships, Assist Memory and Reasoning Audio Resources Podcast Explanations, Voice Notes Utilize Commuting and Other Scenarios for Knowledge Input S107. Based on the learning data of multiple students in the corresponding related subjects, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points, generate the teaching material design scheme of the subject.
[0063] Specifically, it includes: Based on the learning data of multiple students in the corresponding related subjects, the weights of multiple knowledge points are determined. The second optimized particle swarm optimization algorithm generates a subject-specific educational textbook design scheme based on the weights of multiple knowledge points, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points.
[0064] Specifically, the learning data for multiple students in a related subject can include the students' mastery of each knowledge point in the related subject.
[0065] For each knowledge point in the subject to be designed, the related knowledge points are determined based on the correlation coefficient between the knowledge point and each knowledge point in the related subjects. For example, if the correlation coefficient between the knowledge point and a certain knowledge point in the related subject is greater than the correlation coefficient threshold (e.g., 0.5), then the knowledge point in the related subject is considered a related knowledge point.
[0066] For each knowledge point in the subject to be designed, the initial weight of that knowledge point is determined based on the level of mastery of related knowledge points among multiple students. For example, the lower the level of mastery of related knowledge points among multiple students, the higher the initial weight of that knowledge point in the subject to be designed.
[0067] The initial weights of each knowledge point in the design discipline are normalized to generate the weights of each knowledge point in the design discipline.
[0068] Preferably, the second optimized particle swarm optimization algorithm generates a subject-specific educational textbook design scheme based on the weights of multiple knowledge points, the isomorphic graph of multiple knowledge points, and multimodal auxiliary learning resources for multiple knowledge points, including: Based on the weights of multiple knowledge points, the utilization characteristics of multimodal auxiliary learning resources for knowledge points are determined. Specifically, if the weight of a knowledge point is greater than the weight threshold (e.g., 0.2), it is determined that the multimodal auxiliary learning resources of the knowledge point will be used for that knowledge point; otherwise, the multimodal auxiliary learning resources of the knowledge point will not be used to reduce the learning cycle. Initialize the particle swarm, where each particle represents a learning path; A second optimized fitness function is established, wherein the second optimized fitness function is related to the isomorphic graph of multiple knowledge points, the utilization characteristics of multimodal auxiliary learning resources of knowledge points, and the weights of multiple knowledge points; Based on the second optimized fitness function, the particle swarm is optimized to generate a design scheme for educational teaching materials for the subject.
[0069] Specifically, the second optimized fitness function is related to the correlation coefficients of the edges included in the learning path and the weights of the knowledge points covered by the learning path. For example, the second optimized fitness function can be a weighted sum of the mean of the correlation coefficients of the edges included in the learning path and the mean of the weights of the knowledge points covered by the learning path. The particle position (path) is iteratively updated based on the second optimized fitness function, prioritizing the retention of paths with high coherence and high coverage of key knowledge points.
[0070] Resources are dynamically allocated using weight thresholds to avoid resource redundancy for low-weight knowledge points and shorten the learning cycle. The fitness function integrates edge weights from the isomorphic graph to ensure logically coherent paths that align with cognitive patterns. The average weight of knowledge points guides paths to prioritize high-weight content, improving learning efficiency.
[0071] The design of teaching materials for a subject can include learning paths and the utilization characteristics of multimodal auxiliary learning resources for each knowledge point.
[0072] Figure 2 This is a module diagram of a knowledge-sharing-based auxiliary design system for university student educational materials, as shown in one embodiment of this application. Figure 2 As shown, a knowledge-sharing-based textbook design assistance system for university students may include a textbook acquisition module, a knowledge point analysis module, a data acquisition module, and an assistance design module.
[0073] The textbook acquisition module is used to acquire multiple published textbooks for a particular subject. The knowledge point analysis module is used to identify multiple knowledge points based on multiple published textbooks in a subject area. The data acquisition module is used to acquire shared teaching feedback data for the subject. The knowledge point analysis module is used to determine the isomorphism diagram of multiple knowledge points based on shared teaching feedback data of the subject. The data acquisition module is also used to acquire learning data for multiple students in their respective related subjects; The data acquisition module is also used to acquire multimodal auxiliary learning resources for multiple knowledge points; The auxiliary design module is used to generate educational textbook design schemes for a subject based on the learning data of multiple students in the corresponding related subjects, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points.
[0074] A knowledge-sharing-based textbook design assistance system for university students can be used to implement a knowledge-sharing-based textbook design assistance method for university students, which will not be elaborated here.
[0075] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.
Claims
1. A knowledge sharing based college education textbook aided design method, characterized in that, The method comprises the following steps: obtain multiple published textbooks of a subject to be designed; determine multiple knowledge points of the subject to be designed based on the multiple published textbooks of the subject to be designed; obtain shared teaching feedback data of the subject to be designed; construct a homomorphism graph of the multiple knowledge points according to the shared teaching feedback data of the subject to be designed; obtain learning data of multiple students corresponding to related subjects; obtain multiple multi-modal auxiliary learning resources of the multiple knowledge points; generate an education textbook design scheme of the subject based on the learning data of the multiple students corresponding to the related subjects, the homomorphism graph of the multiple knowledge points, and the multiple multi-modal auxiliary learning resources of the multiple knowledge points.
2. The knowledge sharing based college education textbook aided design method according to claim 1, characterized in that, Based on the multiple published textbooks of the subject, the multiple knowledge points of the subject to be designed are determined, which comprises the following steps: for each published textbook, convert the paragraphs in the published textbook into numerical vectors through a word embedding model, cluster the paragraphs in the published textbook based on the numerical vectors to generate a clustering result, and determine multiple to-be-merged knowledge points of the published textbook based on the clustering result; determine the multiple knowledge points of the subject to be designed based on the multiple to-be-merged knowledge points of each published textbook.
3. The knowledge sharing based college education textbook aided design method according to claim 2, characterized in that, Based on the numerical vectors, the paragraphs in the published textbook are clustered to generate a clustering result, and the multiple to-be-merged knowledge points of the published textbook are determined based on the clustering result, which comprises the following steps: for any two paragraphs, calculate the numerical vector cosine similarity of the two paragraphs; cluster the paragraphs in the published textbook based on the numerical vector cosine similarity of any two paragraphs through hierarchical clustering to generate a clustering result, wherein the clustering result comprises multiple paragraph clusters; for each paragraph cluster, based on a subject terminology library, count the multiple terms of each paragraph included in the paragraph cluster and the term frequency of each term, determine the candidate terms of the paragraph cluster according to the multiple terms of each paragraph, and calculate the correlation coefficient of any two candidate terms based on the term frequency of any two candidate terms in each paragraph included in the paragraph cluster, and determine the multiple key terms of the knowledge point corresponding to the paragraph cluster through the first optimized particle swarm algorithm according to the correlation coefficient of any two candidate terms.
4. The knowledge sharing based college education textbook aided design method according to claim 2, characterized in that, Based on the multiple to-be-merged knowledge points of each published textbook, the multiple knowledge points of the subject to be designed are determined, which comprises the following steps: based on the multiple key terms of the multiple to-be-merged knowledge points of each published textbook, de-duplicate the multiple to-be-merged knowledge points of each published textbook to determine the multiple knowledge points of the subject to be designed.
5. The knowledge sharing based college education textbook aided design method according to claim 2, wherein, The shared teaching feedback data of the subject to be designed comprises the mastery degrees of the multiple knowledge points of the subject to be designed by multiple students; According to the shared teaching feedback data of the subject, a homomorphism graph of the multiple knowledge points is constructed, which comprises the following steps: for any two knowledge points, calculate the correlation coefficient between the two knowledge points according to the mastery degrees of the two knowledge points by the multiple students; for each knowledge point, determine the related knowledge points of the knowledge point based on the correlation coefficient between any two knowledge points; based on the related knowledge points of each knowledge point, construct a homomorphism graph of the multiple knowledge points, wherein the homomorphism graph comprises multiple nodes and multiple edges, each node represents a knowledge point, an edge is used to connect two nodes, and the weight of the edge represents the correlation coefficient between the two knowledge points corresponding to the two connected nodes.
6. The knowledge sharing based college education textbook aided design method according to claim 2, characterized in that, determine the related subjects, which comprises the following steps: determine multiple to-be-screened subjects; To obtain the mastery of multiple knowledge points of the selected subject by multiple students; Based on the mastery of multiple knowledge points of the design discipline by multiple students and the mastery of multiple knowledge points of the selected discipline, the correlation coefficient between the selected discipline and the design discipline is calculated. Based on the correlation coefficient between the disciplines to be screened and the disciplines to be designed, the related disciplines are determined.
7. The knowledge sharing based college education textbook aided design method according to any one of claims 1-6, characterized in that, Based on the mastery levels of multiple knowledge points in the design discipline and the mastery levels of multiple knowledge points in the selected discipline by multiple students, the correlation coefficient between the selected discipline and the design discipline is calculated, including: Based on the mastery of multiple knowledge points in the design discipline and the mastery of multiple knowledge points in the selected discipline by multiple students, the correlation coefficient between any knowledge point in the design discipline and any knowledge point in the selected discipline is calculated. Based on the correlation coefficient between any knowledge point of the subject to be designed and any knowledge point of the subject to be screened, calculate the correlation coefficient between the subject to be screened and the subject to be designed.
8. The knowledge sharing based college education textbook aided design method according to any one of claims 1-6, characterized in that, Based on the learning data of multiple students in related subjects, isomorphic graphs of multiple knowledge points, and multimodal auxiliary learning resources for multiple knowledge points, a subject-specific educational textbook design scheme is generated, including: Based on the learning data of multiple students in the corresponding related subjects, the weights of multiple knowledge points are determined. The second optimized particle swarm optimization algorithm generates a subject-specific educational textbook design scheme based on the weights of multiple knowledge points, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points.
9. The knowledge sharing based college education textbook aided design method according to any one of claims 1-6, characterized in that, Using a second optimized particle swarm optimization algorithm, based on the weights of multiple knowledge points, the isomorphic graph of multiple knowledge points, and multimodal auxiliary learning resources for multiple knowledge points, a subject-specific educational textbook design scheme is generated, including: Based on the weights of multiple knowledge points, the utilization characteristics of multimodal auxiliary learning resources for each knowledge point are determined. Initialize the particle swarm, where each particle represents a learning path; A second optimized fitness function is established, wherein the second optimized fitness function is related to the isomorphic graph of multiple knowledge points, the utilization characteristics of multimodal auxiliary learning resources of knowledge points, and the weights of multiple knowledge points; Based on the second optimized fitness function, the particle swarm is optimized to generate a design scheme for educational teaching materials for the subject.
10. A knowledge sharing based college education textbook aided design system, characterized by, The method for assisting in the design of university education teaching materials based on knowledge sharing, as described in claim 1, includes: The textbook acquisition module is used to acquire multiple published textbooks for a particular subject. The knowledge point analysis module is used to identify multiple knowledge points based on multiple published textbooks in a subject area. The data acquisition module is used to acquire shared teaching feedback data for the subject. The knowledge point analysis module is used to determine the isomorphism diagram of multiple knowledge points based on shared teaching feedback data of the subject. The data acquisition module is also used to acquire learning data of multiple students in their corresponding related subjects; The data acquisition module is also used to acquire multimodal auxiliary learning resources for multiple knowledge points; The auxiliary design module is used to generate educational textbook design schemes for a subject based on the learning data of multiple students in the corresponding related subjects, the isomorphic graph of multiple knowledge points, and the multimodal auxiliary learning resources of multiple knowledge points.