Intelligent generation and recommendation system and method for vocational college online open shared resources

Abstract

The invention relates to the technical field of educational informatization, in particular to a vocational college online open shared resource intelligent generation and recommendation system and a vocational college online open shared resource intelligent generation and recommendation method, which comprises a resource acquisition module, a resource generation module, a resource recommendation module, a user module and a resource management module, the acquisition module is used for acquiring user information and behavior data; sending user request information to the resource recommendation module; receiving the personalized resources recommended by the resource recommendation module; and the storage module is connected with the modules and is used for storing various data generated by the operation of the system. A high-efficiency, intelligent and safe vocational education resource generation and recommendation system is established by creatively integrating a plurality of advanced technologies. The system not only can significantly improve the generation efficiency and quality of educational resources, but also can provide more accurate and more personalized learning experience for learners.

Description

Technical Field

[0001] The present invention relates to the technical field of educational informatization, in particular to an intelligent generation and recommendation system and method for online open sharing resources in vocational colleges. Background Art

[0002] With the rapid development of information technology and the popularization of the Internet, online education has become an indispensable part of the modern education system. Especially in the field of vocational education, due to its strong practicality and fast update speed, the demand for high-quality and timely updated online education resources is particularly urgent. However, there are still many challenges in the intelligent generation and recommendation of online open sharing resources in current vocational colleges.

[0003] Existing vocational education resource generation methods mainly rely on manual writing, which is not only inefficient but also difficult to respond promptly to the rapidly changing vocational skill requirements. Although some systems have begun to try to use artificial intelligence technology to assist in resource generation, most are limited to simple text generation or question generation and cannot meet the needs of vocational education for diverse and highly practical teaching resources.

[0004] In terms of resource recommendation, existing systems generally adopt recommendation algorithms based on collaborative filtering or content features. Although these methods perform well in general scenarios, they have obvious deficiencies in the special field of vocational education. For example, it is difficult for them to fully consider factors such as students' vocational development plans and changes in enterprise employment needs, resulting in a large deviation between the recommended results and actual needs. In addition, these systems often ignore the logical relationships between educational resources and are difficult to provide students with a systematic learning path.

[0005] Another prominent problem is the uneven quality of resources. Existing systems lack an effective quality control mechanism in resource collection and screening, resulting in a large number of low-quality or outdated resources being mixed in, seriously affecting the learning effect. At the same time, due to the lack of an effective resource update and optimization mechanism, the resources in the system are prone to aging problems and cannot reflect the latest developments in the industry in a timely manner.

[0006] In addition, existing systems also have obvious deficiencies in personalized recommendation. They usually adopt static user portraits and are difficult to capture the dynamic changes in learners' interests and needs. In the field of vocational education, learners' needs may change rapidly with the adjustment of their vocational plans, and existing systems are difficult to respond promptly to this.

[0007] Finally, existing systems also face challenges in data security and privacy protection. With the increasing strictness of personal information protection regulations, how to protect user privacy while providing personalized services has become an urgent problem to be solved. Summary of the Invention

[0008] In view of the above problems, the present invention proposes an intelligent generation and recommendation system and method for online open and shared resources in vocational colleges based on AIGC. The system aims to achieve efficient resource generation and precise personalized recommendation by innovatively combining AIGC technology with educational resource management, thereby significantly improving the teaching quality and learning effect of vocational education.

[0009] The present invention proposes an intelligent generation and recommendation system for online open and shared resources in vocational colleges, including:

[0010] A resource collection module for collecting open and shared educational resources from the Internet;

[0011] A resource generation module connected to the resource collection module for:

[0012] Receiving the collected resources sent by the resource collection module;

[0013] Generating high-quality educational resources based on the collected resources by using a deep neural network and reinforcement learning;

[0014] A resource recommendation module connected to the resource generation module for:

[0015] Obtaining user request information;

[0016] Performing personalized resource recommendation based on the user request information and the high-quality educational resources;

[0017] A user module connected to the resource recommendation module for:

[0018] Collecting user information and behavior data;

[0019] Sending user request information to the resource recommendation module;

[0020] Receiving the personalized resources recommended by the resource recommendation module;

[0021] A storage module connected to the above-mentioned modules for storing various types of data generated during the operation of the system.

[0022] Preferably, the resource collection module includes:

[0023] A resource crawling sub-module for crawling open and shared resources from the Internet;

[0024] A resource identification sub-module connected to the resource crawling sub-module for identifying and verifying the validity of the open and shared resources;

[0025] A resource classification and marking sub-module connected to the resource identification sub-module for classifying and labeling the open and shared resources that have been verified as valid;

[0026] A resource deduplication sub-module, connected to the resource classification and marking sub-module, for filtering duplicate resources;

[0027] A resource filtering sub-module, connected to the resource deduplication sub-module, for filtering resources that do not meet the screening rules;

[0028] A resource fusion sub-module, connected to the resource filtering sub-module, for intelligently splicing and fusing similar resources.

[0029] Preferably, the resource generation module includes:

[0030] A user information collection sub-module, for obtaining user personal information and behavior data;

[0031] A resource generation sub-module, connected to the user information collection sub-module, for:

[0032] Receiving the user data sent by the user information collection sub-module;

[0033] Generating personalized educational resources based on the user data and a deep neural network model;

[0034] A resource evaluation sub-module, connected to the resource generation sub-module, for evaluating the quality and compliance of the personalized educational resources;

[0035] A resource security sub-module, connected to the resource evaluation sub-module, for performing security detection on the personalized educational resources.

[0036] Preferably, the resource recommendation module includes:

[0037] A user behavior data collection and analysis sub-module, for collecting and analyzing user behavior data;

[0038] A resource retrieval sub-module, connected to the user behavior data collection and analysis sub-module, for:

[0039] Receiving the analysis result sent by the user behavior data collection and analysis sub-module;

[0040] Retrieving matching educational resources based on the analysis result;

[0041] A resource matching sub-module, connected to the resource retrieval sub-module, for:

[0042] Receiving the educational resources retrieved by the resource retrieval sub-module;

[0043] Matching the best resources for the user based on a deep reinforcement learning algorithm.

[0044] Preferably, it further includes a data security module, and the data security module includes:

[0045] An intrusion detection unit for real-time intrusion detection of the system;

[0046] A data backup unit for periodically backing up system data;

[0047] An anti-virus processing unit for virus scanning and processing of the system;

[0048] An identity authentication unit for authenticating user identities;

[0049] An access permission control unit for controlling users' access permissions to resources.

[0050] Preferably, the resource generation module further includes:

[0051] A multi-modal course resource generation sub-module for generating course resources in various forms such as text, images, and videos;

[0052] A course resource optimization sub-module, connected to the multi-modal course resource generation sub-module, for optimizing the generated multi-modal course resources.

[0053] Preferably, the resource generation module further includes:

[0054] A knowledge graph construction sub-module for constructing a professional knowledge framework structure diagram;

[0055] A resource generation sub-module, connected to the knowledge graph construction sub-module, for:

[0056] Receiving the knowledge graph constructed by the knowledge graph construction sub-module;

[0057] Generating logical and systematic course resources based on the knowledge graph.

[0058] Preferably, the resource generation module includes:

[0059] A teacher-side personalized resource generation sub-module for generating personalized teaching resources according to the teacher's teaching style;

[0060] A student-side personalized resource generation sub-module for generating personalized learning resources according to the students' learning characteristics.

[0061] Preferably, the resource recommendation module further includes:

[0062] A context awareness sub-module for obtaining the user's actual context and emotional information through natural language processing;

[0063] A personalized recommendation sub-module, connected to the context awareness sub-module, for:

[0064] Receive the user scenario and emotional information sent by the scenario awareness sub-module;

[0065] Based on the user scenario and emotional information, perform accurate resource recommendations.

[0066] An intelligent generation and recommendation method for online open sharing resources in vocational colleges, including the following steps:

[0067] S1. Collect shared open resources:

[0068] Crawl open shared resources from the Internet;

[0069] Identify, classify, deduplicate, and filter the crawled resources;

[0070] Intelligently splice and integrate similar resources;

[0071] S2. Generate high-quality resources:

[0072] Collect user personal information and behavior data;

[0073] Use deep neural networks and reinforcement learning models to generate high-quality personalized educational resources based on the collected user data and shared open resources;

[0074] Evaluate the quality and compliance of the generated resources;

[0075] Conduct security detection on the generated resources;

[0076] S3. Build a knowledge graph:

[0077] Build a knowledge graph based on the professional knowledge framework;

[0078] Use the knowledge graph to guide resource generation and improve the logic and systematicness of resources;

[0079] S4. Generate multi-modal resources:

[0080] Generate course resources in various forms such as text, images, and videos;

[0081] Optimize the generated multi-modal resources;

[0082] S5. Generate personalized resources:

[0083] Generate personalized teaching resources according to the teaching styles of teachers;

[0084] Generate personalized learning resources according to the learning characteristics of students;

[0085] S6. Resource recommendation:

[0086] Collect and analyze user behavior data;

[0087] Retrieve matching educational resources based on user behavior data;

[0088] Use the deep reinforcement learning algorithm to match the best resources for users;

[0089] S7. Situation-aware recommendation:

[0090] Obtain the actual situation and emotional information of users through natural language processing;

[0091] Based on the user situation and emotional information, perform precise resource recommendation;

[0092] S8. Continuous optimization:

[0093] Collect user feedback information;

[0094] Based on user feedback, continuously optimize the resource generation and recommendation strategies;

[0095] S9. Data security protection:

[0096] Perform real-time intrusion detection on the system;

[0097] Regularly perform data backup;

[0098] Perform virus scanning and processing;

[0099] Implement user identity authentication and access control;

[0100] S10. Resource sharing and update:

[0101] Add high-quality generated resources to the shared resource library;

[0102] Regularly update and optimize the shared resource library.

[0103] Through modular design and the application of advanced algorithms, the system of the present invention has achieved breakthrough innovations in multiple aspects. First, in terms of resource generation, this system uses a method that combines deep neural networks and reinforcement learning, which can automatically generate high-quality and diverse educational resources, greatly improving the efficiency and quality of resource generation. This not only reduces the workload of teachers but also enables educational resources to respond more quickly to the needs of industry development.

[0104] Secondly, in terms of resource recommendation, this system innovatively introduces a recommendation algorithm based on knowledge graphs and situation-aware technology. This enables the system to more comprehensively understand the needs and learning environment of learners and provide more accurate and personalized recommendation services. Especially in the field of vocational education, this method can effectively combine learning content with career development needs and plan a more reasonable learning path for learners.

[0105] Furthermore, the system also has important breakthroughs in resource quality control. By introducing a multi-level resource screening and optimization mechanism, the system can effectively filter out low-quality resources and continuously optimize existing resources through continuous learning. This not only improves the quality of the overall resource library but also ensures the timeliness and practicality of the resources.

[0106] In addition, the system introduces dynamic user profiling technology in personalized services. By real-time analyzing user behavior and learning progress, the system can timely adjust the recommendation strategy to better adapt to the changing needs of learners. This is particularly important in the field of vocational education because learners' career goals and learning needs may change over time.

[0107] Finally, the system also has innovative designs in data security and privacy protection. By adopting advanced encryption technologies and access control mechanisms, the system can effectively protect users' privacy information while providing personalized services.

[0108] Generally speaking, the present invention innovatively integrates a variety of advanced technologies to establish an efficient, intelligent, and secure vocational education resource generation and recommendation system. This system can not only significantly improve the generation efficiency and quality of educational resources but also provide learners with a more accurate and personalized learning experience. Its application will strongly promote the informatization and intelligentization process of vocational education and provide strong support for cultivating high-quality skilled talents. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 It is the overall block diagram of the system of the present invention.

[0110] Figure 2 It is the logical block diagram of the resource collection module 1 of the present invention.

[0111] Figure 3 It is the logical block diagram of the resource generation module 2.

[0112] Figure 4 It is the logical block diagram of the resource recommendation module 3 of the present invention.

[0113] Figure 5 It is the logical block diagram of the data security module 6 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0114] Please refer to Figures 1-5 The present invention provides an intelligent generation and recommendation system and method for online open and shared resources in vocational colleges based on AIGC. The system aims to solve problems such as low generation efficiency of online education resources and insufficient accuracy of personalized recommendations in current vocational colleges. By innovatively combining AIGC technology with educational resource management, it realizes efficient resource generation and accurate personalized recommendations.

[0115] AsFigure 1 As shown in the figure, the system of the present invention includes a resource collection module 1, a resource generation module 2, a resource recommendation module 3, a user module 4, and a storage module 5. These modules work together to form a complete closed-loop system for resource generation and recommendation.

[0116] The resource collection module 1 is used to collect open and shared educational resources from the Internet. Preferably, this module uses a distributed crawler technology, which can efficiently collect data from multiple educational resource websites simultaneously. For example, the concurrency number of the crawler task can be set to 50 - 100, which can avoid putting too much pressure on the target website while ensuring the collection efficiency.

[0117] The resource generation module 2 is connected to the resource collection module 1 and is used to receive the collected resources and generate high-quality educational resources. In an embodiment of the present invention, the resource generation module 2 adopts a method combining a deep neural network and reinforcement learning. Specifically, the following algorithm can be used:

[0118]

[0119] Among them, Q(s,a) represents the value function of taking action a in state s, α is the learning rate (usually taking values from 0.1 to 0.3), r is the immediate reward, γ is the discount factor (usually taking values from 0.9 to 0.99), and s ′ is the next state. This algorithm can generate high-quality resources that better meet the teaching needs through continuous learning and optimization.

[0120] The resource recommendation module 3 is connected to the resource generation module 2 and is used to obtain user request information and perform personalized resource recommendations. This module adopts a collaborative filtering algorithm based on deep learning, and its core formula is as follows:

[0121]

[0122] Among them, is the predicted score of user u for resource i, μ is the global average score, b u and b i are the bias terms of the user and the resource respectively, q i and p u are the latent vectors of the resource and the user respectively, N(u) is the set of resources interacted by user u, and y j is the latent vector of resource j. This algorithm can effectively capture user preferences and resource characteristics and improve the recommendation accuracy.

[0123] The user module 4 is connected to the resource recommendation module 3, mainly responsible for collecting user information and behavior data, and receiving personalized resources recommended. In a preferred embodiment of the present invention, the user module 4 adopts a user behavior analysis algorithm based on the attention mechanism, which can more accurately capture the changes in user interest points.

[0124] The storage module 5 is connected to the above-mentioned modules, and is used to store various types of data generated during the operation of the system. Preferably, this module adopts a distributed storage architecture, such as the Hadoop Distributed File System (HDFS), which can improve the efficiency of data storage and reading.

[0125] Furthermore, the resource collection module 1 includes a resource crawling sub-module 11, a resource identification sub-module 12, a resource classification and marking sub-module 13, a resource deduplication sub-module 14, a resource filtering sub-module 15, and a resource fusion sub-module 16. The collaborative work of these sub-modules ensures the high quality and diversity of the collected resources.

[0126] The resource crawling sub-module 11 is responsible for crawling open and shared resources from the Internet. In an embodiment of the present invention, this sub-module is implemented using the Scrapy framework, which can efficiently crawl multiple websites concurrently. In order to avoid causing excessive pressure on the target website, the crawling interval can be set to 1 - 3 seconds.

[0127] The resource identification sub-module 12 is connected to the resource crawling sub-module 11, and is used to identify and verify the validity of the crawled resources. This sub-module adopts an image recognition algorithm based on the Convolutional Neural Network (CNN) and a text classification algorithm based on BERT, which can effectively identify invalid or low-quality resources.

[0128] The resource classification and marking sub-module 13 classifies and labels the resources that have been verified as valid. The present invention preferably adopts a hierarchical classification system, such as labeling according to the hierarchy of "subject - major - course - knowledge point", which can organize resources more finely and facilitate subsequent retrieval and recommendation.

[0129] The resource deduplication sub-module 14 is responsible for filtering duplicate resources. This sub-module adopts a fast deduplication algorithm based on MinHash and Locality-Sensitive Hashing (LSH). Its core idea is to map documents with high similarity to the same bucket, thereby greatly reducing the number of document pairs that need to be compared and improving the deduplication efficiency.

[0130] The resource filtering sub-module 15 is used to filter resources that do not meet the screening rules. In an embodiment of the present invention, multi-dimensional screening rules are set, including resource integrity, timeliness, readability, etc. For example, specific indicators such as the resource update time not exceeding 2 years and the text readability score (such as the Flesch-Kincaid score) not being lower than 60 points can be set.

[0131] The resource integration sub-module 16 intelligently splices and integrates similar resources. This sub-module adopts a resource integration algorithm based on semantic similarity, which can integrate multiple fragmented resources with related content into a complete knowledge unit, improving the systematicness and integrity of the resources.

[0132] Through the collaborative work of the above modules, the system of the present invention can efficiently collect, generate and recommend high-quality vocational education resources, providing strong support for online education in vocational colleges. The innovation of this system is mainly reflected in the following aspects:

[0133] 1. The use of AIGC technology to realize the intelligent generation of educational resources, greatly improving the efficiency and quality of resource generation.

[0134] 2. Introducing a deep reinforcement learning algorithm enables the system to continuously learn and optimize to adapt to the needs of different users.

[0135] 3. Utilizing multi-modal resource generation technology can generate educational resources in various forms such as text, images, and videos to meet the needs of different learning scenarios.

[0136] 4. The resource generation method based on the knowledge graph ensures the logic and systematicness of the generated resources.

[0137] 5. Adopting a scenario-aware personalized recommendation algorithm improves the accuracy of resource recommendation.

[0138] In summary, the system and method provided by the present invention provide an innovative solution for the intelligent generation and recommendation of online open and shared resources in vocational colleges, and have important theoretical value and practical application prospects.

[0139] Continue to describe other embodiments of the present invention. As Figure 3 shown, the resource recommendation module 3 includes a user behavior data collection and analysis sub-module 31, a resource retrieval sub-module 32, and a resource matching sub-module 33. The collaborative work of these sub-modules realizes accurate personalized resource recommendation.

[0140] The user behavior data collection and analysis sub-module 31 is responsible for collecting and analyzing user behavior data. In a preferred embodiment of the present invention, this sub-module adopts a user behavior analysis algorithm based on time series. Specifically, a long short-term memory network (LSTM) can be used to capture the temporal characteristics of user behavior. The core formula of LSTM is as follows:

[0141] f t =σ(W f ·[h t-1 ,x t +b f ),

[0142] it = σ(W i · [h t-1 , x t + b i ),

[0143]

[0144]

[0145] o t = σ(W o · [h t-1 , x t + b o ),

[0146] h t = o t * tanh(C t ),

[0147] Among them, f t , i t , o t are the forget gate, input gate, and output gate respectively, C t is the cell state, h t is the hidden state, σ is the sigmoid function, and W and b are the weight and bias parameters. In this way, the system can better understand the long-term interests and short-term behavior characteristics of users, so as to provide more accurate recommendations.

[0148] The resource retrieval sub-module 32 is connected to the user behavior data collection and analysis sub-module 31, and is responsible for retrieving matching educational resources based on the analysis results. The system of the present invention adopts a semantic-based retrieval algorithm, and its core is to use word embedding technology to map user queries and resource descriptions into the same semantic space. Preferably, pre-trained word vector models such as Word2Vec or GloVe can be used, so as to make full use of the semantic information in the large-scale corpus and improve the accuracy of retrieval.

[0149] The resource matching sub-module 33 is connected to the resource retrieval sub-module 32, and is used to match the best resources for users based on the deep reinforcement learning algorithm. In one embodiment of the present invention, a reinforcement learning algorithm based on policy gradient is adopted. Its core idea is to optimize the recommendation strategy by maximizing the long-term cumulative reward. The update formula of the policy gradient is as follows:

[0150]

[0151] Among them, θ is the parameter of the policy network, π θ is the current policy, The action value function for taking action a in state s. In this way, the system can continuously optimize the recommendation strategy and improve the long-term satisfaction of users.

[0152] Furthermore, the system of the present invention further includes a data security module 6. This module includes an intrusion detection unit 61, a data backup unit 62, an anti-virus processing unit 63, an identity authentication unit 64, and an access permission control unit 65. The collaborative work of these units ensures the security and stability of the system.

[0153] The intrusion detection unit 61 adopts an anomaly detection algorithm based on deep learning. Preferably, an autoencoder can be used to learn the characteristics of normal network traffic, and then the anomaly traffic can be detected by comparing the reconstruction error. The loss function of the autoencoder can be expressed as:

[0154]

[0155] where x i is the input data, is the reconstructed data, W l is the weight matrix of the l-th layer, and λ is the regularization coefficient. By minimizing this loss function, the system can effectively learn the characteristics of normal traffic and thus accurately detect abnormal behaviors.

[0156] The data backup unit 62 is responsible for regularly backing up the system data. The system of the present invention adopts a strategy combining incremental backup and full backup. Preferably, incremental backup can be performed daily, and full backup can be performed once a week. This can reduce the time and storage space required for backup while ensuring data security.

[0157] The anti-virus processing unit 63 adopts a virus detection algorithm based on machine learning. Specifically, a support vector machine (SVM) can be used to classify normal files and virus files. The decision function of SVM can be expressed as:

[0158]

[0159] where x i is the training sample, y i is the sample label, α i is the Lagrange multiplier, K(x i , x) is the kernel function, and b is the bias term. By selecting an appropriate kernel function (such as the RBF kernel), SVM can effectively separate normal files and virus files.

[0160] The identity authentication unit 64 and the access permission control unit 65 work together to ensure that only authorized users can access system resources. The system of the present invention adopts a role-based access control (RBAC) model and combines multi-factor authentication technology. Preferably, biometric recognition (such as fingerprint or face recognition) can be used as the second authentication factor to further improve the security of the system.

[0161] Through the collaborative work of the above-mentioned modules and units, the system of the present invention not only realizes efficient resource recommendation, but also ensures the security of data and the stability of the system. This all-round design makes the system particularly suitable for application in the online education platform of vocational colleges, and can provide students with a safe, efficient and personalized learning experience.

[0162] The resource generation module 2 further includes a teacher-side personalized resource generation sub-module 21 and a student-side personalized resource generation sub-module 22. The design of these two sub-modules fully reflects the personalized characteristics of the system of the present invention and can generate customized educational resources according to the needs of different user groups.

[0163] The teacher-side personalized resource generation sub-module 21 mainly generates personalized teaching resources according to the teaching style of teachers. In a preferred embodiment of the present invention, this sub-module adopts a sequence-to-sequence (Seq2Seq) model based on the attention mechanism. Specifically, its core mechanism can be described by the following formula:

[0164]

[0165] Among them, c i is the context vector, α ij is the attention weight, h j is the encoder hidden state, s i -1 is the hidden state of the decoder at the previous moment, and a(·) is the alignment model. Through this mechanism, the system can generate teaching resources that are more in line with the personal characteristics of teachers according to their teaching style and key concerns.

[0166] The student-side personalized resource generation sub-module 22 generates personalized learning resources according to the learning characteristics of students. The system of the present invention innovatively introduces meta-learning technology here. Preferably, the model-agnostic meta-learning (MAML) algorithm can be used. Its core idea is to learn an initialization of the model that can quickly adapt to new tasks. The objective function of MAML can be expressed as:

[0167]

[0168] Among them,

[0169]

[0170] Here, θ is the model parameter, is the task, and α is the inner loop learning rate. By this method, the system can quickly adapt to the learning characteristics of different students and generate more personalized learning resources.

[0171] Furthermore, the resource recommendation module 3 further includes a scenario awareness sub-module 34 and a personalized recommendation sub-module 35. The introduction of these two sub-modules enables the system of the present invention to more accurately grasp the actual needs of users and provide a more considerate resource recommendation service.

[0172] The scenario awareness sub-module 34 mainly obtains the actual scenario and emotional information of the user through natural language processing technology. In an embodiment of the present invention, a multi-task learning model based on BERT is adopted. This model can perform scenario classification and sentiment analysis simultaneously, and its loss function can be expressed as:

[0173] L = λ 1 L scenario + λ 2 L emotion ,

[0174] where, L scenario and L emotion are the loss functions of scenario classification and sentiment analysis respectively, and λ 1 and λ 2 are weight coefficients. By adjusting these two weight coefficients, the importance of the two tasks can be balanced, so as to obtain more comprehensive user scenario information.

[0175] The personalized recommendation sub-module 35 is connected to the scenario awareness sub-module 34 and performs accurate resource recommendation based on the obtained user scenario and emotional information. The system of the present invention innovatively adopts graph neural network (Graph Neural Network, GNN) technology here. Preferably, a graph attention network (Graph Attention Network, GAT) can be used. Its core operation can be expressed as:

[0176]

[0177] where, h i is the feature of node i, is the neighbor set of node i, W is the weight matrix, a is the attention vector, and σ is the activation function. In this way, the system can make full use of the structural information in the user-resource interaction graph to provide more accurate personalized recommendations.

[0178] Finally, the present invention also provides an intelligent generation and recommendation method for online open and shared resources in vocational colleges based on AIGC. The method includes the following steps:

[0179] S1. Collect shared open resources: Crawl open and shared resources from the Internet, identify, classify, de-duplicate, and filter the resources, and finally perform intelligent splicing and fusion on resources of the same type. Preferably, in this step, a distributed crawler technology can be adopted, and the concurrency number of crawler tasks is set to 50-100, and the crawling interval is 1-3 seconds to balance the collection efficiency and the impact on the target website.

[0180] S2. Generate high-quality resources: Collect user personal information and behavior data, and use deep neural networks and reinforcement learning models to generate high-quality personalized educational resources. In an embodiment of the present invention, a generative adversarial network (GAN) can be used to improve the quality and diversity of the generated resources. The objective function of GAN can be expressed as:

[0181]

[0182] where G is the generator, D is the discriminator, x is the real data, and z is the random noise. Through this adversarial learning method, the system can generate more real and high-quality educational resources.

[0183] S3. Construct a knowledge graph: Based on the professional knowledge framework, construct a knowledge graph, and use the knowledge graph to guide resource generation to improve the logic and systematicness of the resources. The system of the present invention adopts a knowledge graph construction method based on distant supervision, which can automatically extract entity relationships from large-scale texts and quickly construct a domain knowledge graph.

[0184] S4-S10. These steps respectively correspond to the processes of multi-modal resource generation, personalized resource generation, resource recommendation, scenario-aware recommendation, continuous optimization, data security protection, and resource sharing and update described above. Each step makes full use of advanced artificial intelligence technologies such as deep learning, reinforcement learning, and natural language processing to ensure the intelligence and efficiency of the entire system.

[0185] Through the above method, the system of the present invention can continuously generate high-quality personalized educational resources and make accurate recommendations according to the actual needs of users. This method is particularly suitable for application in the online education platform of vocational colleges and can significantly improve the teaching effect and learning efficiency.

[0186] It should be noted that: The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. Intelligent generation and recommendation system for online open shared resources in vocational schools, characterized by ,include: Resource collection module, used to collect open and shared educational resources from the Internet; A resource generation module, connected to the resource acquisition module, is used to: Receiving the collection resources sent by the resource collection module; Based on the collected resources, high-quality educational resources are generated using deep neural networks and reinforcement learning; A resource recommendation module, connected to the resource generation module, is used to: Get user request information; Based on the user request information and the high-quality educational resources, personalized resource recommendations are made; The user module is connected to the resource recommendation module and is used to: Collect user information and behavior data; Sending user request information to the resource recommendation module; Receiving personalized resources recommended by the resource recommendation module; The storage module is connected with the above modules and is used to store various data generated by the operation of the system.

2. The system according to claim 1, characterized in that , the resource acquisition module includes: Resource crawling submodule, used to crawl open shared resources from the Internet; A resource identification submodule, connected to the resource crawling submodule, for identifying and verifying the validity of the open shared resources; A resource classification and marking submodule, connected to the resource identification submodule, is used to classify and mark the open and shared resources that have been verified to be valid; A resource deduplication submodule, connected to the resource classification and marking submodule, for filtering duplicate resources; A resource filtering submodule, connected to the resource deduplication submodule, for filtering resources that do not meet the screening rules; The resource fusion submodule is connected to the resource filtering submodule and is used to intelligently splice and fuse similar resources.

3. The system according to claim 1, characterized in that , the resource generation module includes: User information collection submodule, used to obtain user personal information and behavior data; The resource generation submodule is connected to the user information collection submodule and is used to: Receiving user data sent by the user information collection submodule; Generate personalized educational resources based on the user data and the deep neural network model; A resource evaluation submodule, connected to the resource generation submodule, for evaluating the quality and compliance of the personalized educational resources; The resource security submodule is connected to the resource evaluation submodule and is used to perform security detection on the personalized educational resources.

4. The system according to claim 1, characterized in that , the resource recommendation module includes: User behavior data collection and analysis submodule, used to collect and analyze user behavior data; The resource retrieval submodule is connected to the user behavior data collection and analysis submodule and is used to: Receiving the analysis results sent by the user behavior data collection and analysis submodule; Based on the analysis results, searching for matching educational resources; The resource matching submodule is connected to the resource retrieval submodule and is used to: Receiving educational resources retrieved by the resource retrieval submodule; Based on deep reinforcement learning algorithm, the best resources are matched to users.

5. The system according to claim 1, characterized in that , further comprising a data security module, the data security module comprising: Intrusion detection unit, used to perform real-time intrusion detection on the system; Data backup unit, used to regularly back up system data; Anti-virus processing unit, used to scan and process viruses on the system; An identity authentication unit, used to authenticate the user's identity; The access permission control unit is used to control the user's access permissions to resources.

6. The system according to claim 1, characterized in that , the resource generation module also includes: The multimodal course resource generation submodule is used to generate course resources in various forms such as text, images, and videos; The course resource optimization submodule is connected to the multimodal course resource generation submodule and is used to optimize the generated multimodal course resources.

7. The system according to claim 1, characterized in that , the resource generation module also includes: The knowledge graph construction submodule is used to construct the professional knowledge framework structure diagram; The resource generation submodule is connected to the knowledge graph construction submodule and is used to: Receive the knowledge graph constructed by the knowledge graph construction submodule; Based on the knowledge graph, logical and systematic course resources are generated.

8. The system according to claim 1, characterized in that , the resource generation module includes: The teacher-side personalized resource generation submodule is used to generate personalized teaching resources according to the teacher's teaching style; The student-side personalized resource generation submodule is used to generate personalized learning resources based on students' learning characteristics.

9. The system according to claim 1, characterized in that , the resource recommendation module also includes: The context-aware submodule is used to obtain the user's actual context and emotional information through natural language processing; The personalized recommendation submodule is connected to the context awareness submodule and is used to: Receiving user context and emotion information sent by the context awareness submodule; Based on the user context and sentiment information, accurate resource recommendations are made.

10. An intelligent generation and recommendation method for online open shared resources in vocational schools based on AIGC, characterized by , including the following steps: S1. Collect and share open resources: Crawl open shared resources from the Internet; Identify, classify, deduplicate and filter crawled resources; Intelligently splice and merge similar resources; S2. Generate high-quality resources: Collect user personal information and behavior data; Using deep neural networks and reinforcement learning models, we can generate high-quality personalized educational resources based on collected user data and shared open resources. Assess the quality and compliance of generated resources; Perform security checks on generated resources; S3. Build a knowledge graph: Build a knowledge graph based on the professional knowledge framework; Use knowledge graphs to guide resource generation and improve the logic and systematicness of resources; S4. Generate multimodal resources: Generate course resources in various forms such as text, images, and videos; Optimize the generated multimodal resources; S5. Personalized resource generation: Generate personalized teaching resources based on teachers' teaching styles; Generate personalized learning resources based on students' learning characteristics; S6.Resource recommendations: Collect and analyze user behavior data; Retrieve matching educational resources based on user behavior data; Use deep reinforcement learning algorithms to match users with the best resources; S7. Situational awareness recommendation: Obtain the user's actual situation and emotional information through natural language processing; Make accurate resource recommendations based on user context and sentiment information; S8.Continuous Optimization: Collect user feedback information; Continuously optimize resource generation and recommendation strategies based on user feedback; S9. Data security protection: Perform real-time intrusion detection on the system; Back up data regularly; Perform virus scanning and processing; Implement user authentication and access rights control; S10. Resource sharing and updating: Add high-quality generated resources to the shared resource library; Regularly update and optimize the shared resource library.

Images