Convolutional neural network-based online educational resource AI recommendation system

Abstract

An online education resource AI recommendation system based on a convolutional neural network, comprises: a user terminal; an information collection module; an information analysis module; and an educ

Description

Technical Field The present invention belongs to the field of online education technology, more specifically, it relates to an Al recommendation system for online educational resources based on convolutional neural networks. Background Technology With the rapid development and popularization of digital technology and Al, the education industry is gradually undergoing digitalization and intelligence transformation. An important component of modern education is online education platforms. These platforms offer a wide range of learning resources, providing users with flexible learning methods. However, as online educational resources become increasingly abundant, users face the challenge of finding suitable learning resources amidst information overload and resource redundancy. Therefore, more efficient Al algorithms, such as recommendation systems, are needed to help users quickly extract relevant information and resources from the platform. Traditional education models primarily rely on face-to-face classrooms, but learning is a continuous process, and people have learning needs at any stage and time. Online education resource recommendation Al platforms provide a convenient learning channel. With the improvement of education levels and diversification of learning needs, online educational resources are continuously increasing globally, offering a wide variety of course content and formats covering various subjects, levels, and types. However, users face challenges of resource dispersion and difficulty in filtering, requiring them to spend a lot of time and effort to find suitable learning content. Furthermore, with the growing trend of internet educational resources and their impact on the education sector, the demand for personalized and Al-driven educational resource recommendations will become increasingly prominent. Therefore, an efficient and accurate method is needed to deeply analyze and process educational resources, achieve a deep understanding and precise classification of educational resource content, and effectively capture users' interests and learning needs to provide users with more precise, personalized, and intelligent learning recommendations. Existing education system recommendation systems are based on content recommendations. Content-based recommendation algorithms primarily analyze students' interests and course characteristics to determine if a student is interested in a particular course. However, these recommendation algorithms often lead to a large number of repetitive and limited educational resource suggestions, potentially causing users to seek resources on other platforms, resulting in user attrition. Summary of the Invention Addressing the shortcomings of existing content-based recommendation systems, the present invention proposes an Al recommendation system for online educational resources based on convolutional neural networks. This recommendation system, based on user information analysis, inputs the analysis results into a convolutional neural network model to recommend a certain number of educational resources to the user. The invention also introduces a supplementary recommendation method. This method involves analyzing user behavior information and comparing it with the behavior patterns of other users to find users with similar interests. It then recommends educational resources used by these similar-interest users, providing the user with slightly different educational resources that align with their learning interests, enriching their learning experience and expanding the breadth of their learning. The present invention adopts the following technical solution: An Al recommendation system for online educational resources based on convolutional neural networks, comprising: User terminal: obtains editing content information entered by the user on the user interface and synchronizes it in real-time to the information collection module; Information collection module: collects user feature information, behavior information, and user input information, splits the text content in the obtained editing content information into keywords, obtains a set of recommended keywords for the user terminal, and sends the user's historical keywords, feature information, behavior information, and the current keyword Al recommendation set to the information analysis module; Information analysis module: receives relevant data information sent by the information collection module, preprocesses it, assigns weights, and obtains user feature vectors; Educational resource recommendation module: uses the user feature vectors analyzed by the information analysis module as input data, inputs them into the neural network model, filters out the educational resources to be recommended, and forms a list of recommended educational resources for users to choose from; Determines whether the recommended number of educational resources meets the required recommendation quantity; If yes, the recommended educational resources are sent to the user terminal for display; If no, calculates the difference between the required recommendation quantity and the number of educational resources recommended to the user, and based on the supplementary recommendation method for educational resources, selects the corresponding additional educational resources from the educational resource library to intelligently recommend to the user. In a further improved embodiment of this technical solution, the neural network model is a convolutional neural network model. In a further improved embodiment of this technical solution, the construction steps of the convolutional neural network model are as follows: Obtaining the user feature dataset and divide it into training and testing sets; Building a deep learning Al convolutional neural network model; Using the training set to iteratively train the convolutional neural network model, test the model using the testing set, evaluate the prediction effect of the convolutional neural network model based on Mean Squared Error (MSE), adjust the model parameters until convergence is achieved, and obtain the required convolutional neural network model. In a further improved embodiment of this technical solution, the convolutional neural network model includes input layer, convolutional layer, fully connected layer, and output layer. In a further improved embodiment of this technical solution, the structure of the convolutional neural network model is as follows: input layer, the first convolutional layer, the second convolutional layer, the first pooling layer, the third convolutional layer, the second pooling layer, the first fully connected layer, the second fully connected layer, and output layer. In a further improved embodiment of this technical solution, the user feature information includes name, gender, age, education level, and language category. In a further improved embodiment of this technical solution, the steps of the supplementary recommendation method are as follows: Analyzing the current user's behavior information and compare it with the behavior patterns of other users to infer the user's interests and preferences; Finding other users with similar interests to the current user; Based on the behavior of similar users, recommending additional educational resources to the current user to fill the gap. In a further improved embodiment of this technical solution, the current user's behavior information includes educational resources browsed, collected, and rated by the user. In a further improved embodiment of this technical solution, the method of recommending additional educational resources to the current user based on the behavior of similar users includes: Obtaining all educational resources browsed by similar users; After excluding the recommended educational resources, sorting the remaining educational resources in descending order based on the number of views; Selecting the top-ranked educational resources according to the difference in quantity and recommend them as supplementary resources to the user. The present invention has following advantages over the prior art. The present invention introduces an Al recommendation system for online educational resources based on convolutional neural networks, which recommends educational resources that meet user needs through user analysis. Different data information is given different weights, resulting in different recommendation results that cater to users' personalized needs. For cases where resources are limited, by analyzing user behavior information, the system identifies other users with similar interests and recommends educational resources they have browsed, helping users expand their knowledge and broaden their learning horizons. Description of the Drawings Figure 1 illustrates the schematic diagram of the Al recommendation system for online educational resources based on convolutional neural networks; Figure 2 illustrates the schematic diagram of the structure of the convolutional neural network model. Detailed Description of the Embodiments To further explain various embodiments, the present invention provides accompanying drawings. These drawings are part of the disclosed content of the invention, primarily used to illustrate embodiments and can be used in conjunction with the relevant description in the specification to explain the operation principles of the embodiments. With reference to this content, those skilled in the art should be able to understand other possible embodiments and the advantages of the invention. The components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components. The present invention is further described in connection with the accompanying drawings and specific embodiments. Referring to Figure 1, it illustrates the schematic diagram of the Al recommendation system for online educational resources based on convolutional neural networks. The preferred embodiment of the online educational resource recommendation system includes: User terminal: obtains editing content information entered by the user on the user interface and synchronizes it in real-time to the information collection module. Users input the educational resource information they want to search for through keywords, sentences, or voice input methods. Voice input is also recognized as text and sent to the information collection module. Information collection module: collects user feature information, behavior information, and user input information. It splits the text content in the editing content information into keywords, obtains a set of recommended keywords for the user terminal, and sends the user's historical keywords, feature information, behavior information, and the current keyword recommendation set to the information analysis module. User feature information includes but is not limited to name, gender, age, education level, and language category. User behavior information includes but is not limited to educational resources browsed, collected, and rated by the user. Historical keywords of the user refer to the set of keywords obtained by analyzing the user's historical input information. Information analysis module: receives relevant data information sent by the information collection module, preprocesses it, assigns weights, and obtains user feature vectors. It combines historical keywords and current keywords, counts the occurrences of each keyword, assigns different weights based on the frequency of keyword occurrences, and assigns higher weights to keywords that appear more frequently, indicating higher user interest. Educational resource recommendation module: uses the user feature vectors analyzed by the information analysis module as input data, inputs them into the neural network model, filters out the educational resources to be recommended, and forms a list of Al-recommended educational resources for users to choose from. Determines whether the recommended number of educational resources meets the required recommendation quantity: If yes, the recommended educational resources are sent to the user terminal for display; If no, calculates the difference between the required recommendation quantity and the number of recommended educational resources, and based on the supplementary recommendation method for educational resources, selects the corresponding additional educational resources from the educational resource library to Al-recommend to the user. Convolutional neural networks can extract many features. They use convolutional kernels to extract features from data, which can be manually designed or learned through training. Convolutional neural networks use a design mechanism of local connections and weight sharing to reduce the parameter size in the network, giving it better feature extraction capabilities. The convolutional layer of a convolutional neural network typically uses a set of convolution kernels to extract features. These kernels share weight parameters during forward propagation, significantly reducing the computational load of the network. Therefore, the neural network model in this embodiment adopts a convolutional neural network model. Referring to Figure 2, it illustrates the schematic diagram of the structure of the convolutional neural network model. The convolutional neural network model includes an input layer, hidden layers, and an output layer. Generally, the hidden layers consist of convolutional layers, pooling layers, and fully connected layers, with varying numbers of these layers in different convolutional neural networks. The convolutional layer extracts features, the pooling layer reduces the number of parameters, and the fully connected layer acts as a "classifier" throughout the convolutional neural network. The structure of the convolutional neural network model in this embodiment includes an input layer, the first convolutional layer, the second convolutional layer, the first pooling layer, the third convolutional layer, the second pooling layer, the first fully connected layer, the second fully connected layer, and the output layer. The construction steps of the convolutional neural network model are as follows: Obtain a user feature dataset and divide the user feature dataset into a training set and a test set; Build a deep learning convolutional neural network model; Use the training set to iteratively train the convolutional neural network model, use the test set to test the convolutional neural network model, evaluate the prediction effect of the convolutional neural network model based on Mean Squared Error (MSE), adjust the model parameters until the model converges, and obtain the required convolutional neural network model. The steps of the education resource supplementary recommendation method in this embodiment are as follows: Analyze the current user's behavioral information and other users' behavioral patterns to infer the user's interests and preferences. Find other users with similar interests to the current user, compare the interests and hobbies of the current user with those of other users, rank them based on the overlap, where users with higher rankings have more similar interests to the current user. Recommend differential educational resources to the current user based on the behavior of similar users. Method for recommending differential educational resources to the current user based on the behavior of similar users: Obtain all educational resources browsed by similar users. Select the top five or top ten users with similar interests, collect the educational resources browsed by each user, then rank these educational resources from high to low based on the number of views. After excluding the already recommended educational resources, the remaining educational resources are sorted in descending order based on the number of views. Based on the differential quantity, select the top-ranked educational resources as supplementary recommendations for the user. Although the preferred embodiments have been specifically described and introduced in conjunction with the present invention, those skilled in the art should understand that various changes can be made to the present invention in form and detail without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. An online education resource Al recommendation system based on a convolutional neural network, characterized in that the system comprises:a user terminal, which acquires the editing content information input by the user on the user-end input interface and synchronizes it in real-time to the information collection module;an information collection module, which is used to collect the user's characteristic information, behavior information, and input information, and splits the text content in the acquired editing content information into keywords to obtain the keyword recommendation set of the user terminal, and sends the user's historical keywords, characteristic information, behavior information, and current keyword recommendation set to the information analysis module;an information analysis module, which receives the relevant data information sent by the information collection module, preprocesses it, and assigns weights to obtain the user feature vector;an education resource recommendation module, which takes the user feature vector analyzed by the information analysis module as input data, inputs it into the neural network model, and Al filters out the educational resources to be recommended and forms an education resource recommendation list for the user to choose from;the Al determines whether the number of recommended educational resources reaches the required recommendation quantity;if yes, the recommended educational resources are sent to the user terminal for display,if not, the Al calculates the difference between the required recommendation quantity and the number of educational resources recommended to the user, and based on the supplementary recommendation method of educational resources, the Al selects the corresponding number of supplementary educational resources from the educational resource library to recommend to the user.

2. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 1, characterized in that the neural network model is a convolutional neural network model.

3. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 2, characterized in that the construction steps of the convolutional neural network model are as follows:acquiring the user feature dataset and dividing the user feature dataset into a training set and a test set;constructing a deep learning convolutional neural network model;using the training set to perform multiple rounds of Al iterative training on the convolutional neural network model, using the test set to test the convolutional neural network model, and verifying the prediction effect of the convolutional neural network model based on the mean squared error (MSE) and adjusting the model parameters until the model converges to obtain the required convolutional neural network model.

4. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 3, characterized in that the convolutional neural network model comprises an input layer, convolutional layers, fully connected layers, and an output layer.

5. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 4, characterized in that the structure of the convolutional neural network model is an input layer, a first convolutional layer, a second convolutional layer, a first pooling layer, a third convolutional layer, a second pooling layer, a first fully connected layer, a second fully connected layer, and an output layer.

6. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 1, characterized in that the user characteristic information includes name, gender, age, education level, and language category.

7. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 1, characterized in that the steps of the supplementary recommendation method are as follows:analyzing the current user's behavior information and the behavior patterns of other users to infer the user's interests and preferences;finding other users with similar interests to the current user;recommending the differential educational resources to the current user based on the behavior of similar users.

8. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 7, characterized in that the current user's behavior information includes educational resources browsed by the user, educational resources collected by the user, and educational resources evaluated by the user.

9. The online education resource Al recommendation system based on a convolutional neural network as claimed in claim 8, characterized in that the method of recommending differential educational resources to the current user based on the behavior of similar users comprises:acquiring all the educational resources browsed by similar users;excluding the already recommended educational resources, and sorting the remaining educational resources in descending order based on the number of views;selecting the top-ranked educational resources according to the differential quantity as supplementary intelligent automated recommendations for the user.

Images