Multi-source software-defined education resource recommendation method and system based on knowledge graph

Abstract

The application provides a multi-source software-defined science and education resource recommendation method and system based on a knowledge graph, comprising the following steps: constructing an artificial intelligence knowledge graph based on artificial intelligence knowledge entities and the relationships between the entities; obtaining science and education resource data of a science and education institution, extracting knowledge entities in the science and education resource data, connecting the knowledge entities to the artificial intelligence knowledge graph, and obtaining a science and education resource knowledge graph; extracting local resource features and user features based on a graph neural network, and obtaining resource features and user features of each local client; constructing a local knowledge graph based on the science and education resource knowledge graph, using the resource features of each local client as initial features for model training until the model converges; inputting the user and the resource into the graph neural network for prediction, and recommending resources with high scores to each user. The scheme effectively extracts knowledge entities and the relationships between the entities in the field of artificial intelligence, forms a structured knowledge graph, and improves the recommendation accuracy.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of resource recommendation, and particularly relates to a multi-source software-defined science and education resource recommendation method and system based on a knowledge graph. BACKGROUND

[0002] A large amount of science and education resources are generated in the process of digital education, which greatly improves the resource acquisition efficiency of learners, but there are problems of resource complexity and difficulty in management. In view of this, a software-defined science and education resource management model is proposed, the skill points and knowledge points contained in the science and education resources are mined through a knowledge graph, and a science and education resource knowledge semantic network is constructed by taking the knowledge graph as a medium, so as to overcome the management problem of multi-source massive science and education resources from a semantic level. However, the software-defined science and education resource model still faces a serious problem of "knowledge navigation" when processing massive science and education resources, that is, users have difficulty in obtaining the resources they need from massive, multi-source distributed science and education resources.

[0003] In view of the problem of knowledge navigation, a common solution is to construct a recommendation system based on a collaborative filtering method, which is a centralized scenario and assumes that the science and education resources and the access records of users to the resources are concentrated in the same institution. However, in a multi-source distributed scenario, different science and education resources and the interaction information of users and resources are stored in different institutions, and data sharing between different institutions is difficult due to privacy protection and knowledge security considerations, which brings difficulties to the science and education resource recommendation based on collaborative filtering.

[0004] In view of this, a recommendation scheme based on federated learning is proposed, the model runs in a decentralized federated learning environment, which can ensure that the data of users is completely retained on their personal devices, and at the same time helps to train an accurate model. On the other hand, in order to better adapt to the data heterogeneity in FL (for example, the data of users differs greatly in scale and quality), a first-order meta-learning method is introduced, which can realize fast device personalization using only a small number of data points. However, most of the schemes are aggregated from the model level, which is difficult to effectively integrate the user preference features existing in different institutions, resulting in low recommendation accuracy. SUMMARY

[0005] The application aims to provide a multi-source software-defined science and education resource recommendation method and system based on a knowledge graph to solve the above problems.

[0006] To achieve the above purpose, the application adopts the following technical scheme:

[0007] In a first aspect, the application provides a multi-source software-defined science and education resource recommendation method based on a knowledge graph, comprising:

[0008] Based on the artificial intelligence knowledge entity and the relationship between entities, an artificial intelligence knowledge graph is constructed; the scientific and educational resource data of the scientific and educational institutions are acquired, the knowledge entities in the scientific and educational resource data are extracted, and the scientific and educational resource knowledge graph is obtained by connecting to the artificial intelligence knowledge graph;

[0009] Based on the graph neural network, the local resource features and user features are extracted to obtain the resource features and user features of each local client;

[0010] Based on the scientific and educational resource knowledge graph, a local knowledge graph is constructed, and the resource features of each local client are used as initial features for model training until the model converges;

[0011] The user and the resource are input into the graph neural network for prediction, and the resources with high scores are recommended to each user.

[0012] Optionally, based on the artificial intelligence knowledge entity and the relationship between entities, an artificial intelligence knowledge graph is constructed, including:

[0013] A knowledge corpus in the field of artificial intelligence is constructed, and entity recognition algorithms and relationship extraction algorithms are used to extract knowledge entities and relationships between entities in the field of artificial intelligence to form a knowledge graph in the field of artificial intelligence;

[0014] Specifically:

[0015] Text data is collected and labeled, and is divided into a training set, a validation set and a test set, and the text is segmented, stop words are removed and encoded;

[0016] On a given corpus, the BERT language processing model is used to extract text features, then the trained model is applied to the test set for named entity recognition by evaluating the performance on the validation set and adjusting the trained model.

[0017] Relevant features are extracted, and a recurrent neural network is used for training, and the model performance is evaluated on the validation set.

[0018] Optionally, the scientific and educational resource data of the scientific and educational institutions are acquired, the knowledge entities in the scientific and educational resource data are extracted, and the scientific and educational resource knowledge graph is obtained by connecting to the artificial intelligence knowledge graph, including:

[0019] The metadata, text, video, image and file data contained in the scientific and educational resources held by each scientific and educational institution are processed, the knowledge entities and their associations contained therein are extracted, and the software-defined scientific and educational resource knowledge graph integrating artificial intelligence knowledge points and scientific and educational resources is constructed by aligning and connecting with the constructed artificial intelligence knowledge graph:

[0020] The multimodal entity recognition process begins by collecting and labeling a dataset containing text and corresponding images. The text is then segmented and embedded to generate vector representations, while convolutional neural networks are used to extract features from the images. Separate text processing and image processing networks are constructed, and either Transformer or convolutional neural networks are used for processing. The processed results are then weighted and fused to generate a comprehensive feature representation, which is used to output the category of each entity.

[0021] The learned features are compared with the knowledge point features in the knowledge graph using cosine similarity measurement, and the closest result is selected for linking, thus completing the knowledge connection from science and education resources to entities in the knowledge graph.

[0022] Optionally, local resource features and user features are extracted based on graph neural networks to obtain the local resource features and user features of each client, including:

[0023] For each client, a user preference network is constructed by graph modeling the "user-resource" preference relationship; the cross-entropy loss function is constructed using the preference relationship to train the model; and when the model converges, the local resource features and user features of each client are obtained.

[0024] Optionally, a local knowledge graph is constructed based on the science and education resource knowledge graph. The local resource features of each client are used as initial features for model training until the model converges, including:

[0025] For science and education resource recommendation scenarios, a recommendation network is constructed based on graph convolutional neural networks;

[0026] Based on the constructed model, a loss function is built, and the data is input into the model to calculate the loss and update the model parameters through backpropagation until the model converges.

[0027] Optional model building includes:

[0028] Feature embedding module: At each layer, for each edge type Assign one Dimensional embedding And simultaneously use edge-type embeddings and node embeddings to compute attention scores:

[0029]

[0030] in Represents a node and nodes The type of the edges between them It is a learnable matrix used for transforming type embeddings;

[0031] Residual connection module: add pre-activation residual connection, used for cross-layer node representation, assuming is the attention weight of edge , is the activation function, when the feature dimension of the layer changes, the scheme uses a learnable linear transformation to transform the features, and the transformed features are aggregated by residual:

[0032]

[0033] Similarly, the learning of edge features also introduces a residual module, as follows:

[0034]

[0035] Multi-head attention module: introduce multi-head attention mechanism, with independent attention modules to learn features respectively, and the learned features are spliced to obtain the final feature representation, and the corresponding update rule is as follows:

[0036]

[0037]

[0038]

[0039] Final output layer The output result of the layer is:

[0040]

[0041] Preference information prediction: the output features are then input into the prediction module built based on the multi-layer perception mechanism. Specifically, for user features and resource features , the features are spliced and input into the multi-layer perception machine to obtain the predicted score as follows:

[0042] .

[0043] Optionally, model training includes:

[0044] Construct cross-entropy: for the output predicted label result set and the real label , construct cross-entropy loss as follows:

[0045]

[0046] Gradient propagation: then use the back propagation of the neural network to calculate the gradient and optimize the parameters, evaluate whether to reach the stopping condition, if the model does not converge, continue training until the model converges.

[0047] In a second aspect, the present application provides a multi-source software-defined science and education resource recommendation system based on a knowledge graph, comprising:

[0048] A science and education resource knowledge graph acquisition module is configured to construct an artificial intelligence knowledge graph based on artificial intelligence knowledge entities and the relationships between the entities, acquire science and education resource data of a science and education institution, extract knowledge entities in the science and education resource data, connect to the artificial intelligence knowledge graph, and obtain a science and education resource knowledge graph.

[0049] A feature extraction module is configured to extract local resource features and user features based on a graph neural network to obtain resource features and user features of each local client;

[0050] A model training module is configured to construct a local knowledge graph based on the science and education resource knowledge graph, and train the model with the resource features of each local client as initial features until the model converges.

[0051] A prediction output module is configured to input the user and the resource into the graph neural network for prediction, and recommend resources with high scores to each user.

[0052] In a third aspect, the present application provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the multi-source software-defined science and education resource recommendation method based on a knowledge graph.

[0053] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the steps of the multi-source software-defined science and education resource recommendation method based on a knowledge graph.

[0054] Compared with the prior art, the present application has the following technical effects:

[0055] The present application constructs a knowledge corpus in the field of artificial intelligence, and uses entity recognition algorithms and relationship extraction algorithms to effectively extract knowledge entities and relationships between entities in the field of artificial intelligence, forming a structured knowledge graph. This helps to efficiently organize and manage knowledge, facilitating subsequent retrieval and application.

[0056] The present application aligns and connects the science and education resources (including metadata, text, video, image, file, etc.) held by various science and education institutions with the constructed artificial intelligence knowledge graph, and constructs a software-defined science and education resource knowledge graph that fuses artificial intelligence knowledge points and science and education resources. This integration method not only realizes the sharing of resources, but also promotes the deep integration of knowledge and resources.

[0057] The present application models the preference relationship of "user-resource" through the graph neural network, and constructs a cross-entropy loss function using the preference relationship for model training. This scheme can accurately extract the features of users and resources. This helps to more accurately understand user needs and resource characteristics, providing strong support for subsequent recommendations. The scheme uses a knowledge graph as a feature propagation medium, combined with a graph convolutional neural network, to achieve effective propagation and fusion of resource features and user preference features between different institutions. This feature propagation method not only considers the semantic level of the features, but also fuses the knowledge point features through the attention mechanism to obtain comprehensive knowledge point features.

[0058] The present application realizes efficient training of the model by constructing a cross-entropy loss function and using the backpropagation of the neural network to calculate the gradient to optimize the parameters. This training method not only has fast convergence speed, but also can ensure the stability and accuracy of the model. The local feature updating step in the scheme allows the local resource features to be updated according to the global knowledge point features, and the updated features are input into the graph neural network together with the local user features for feature updating. This flexible model updating mechanism helps to adapt to changing user needs and resource environment.

[0059] The present application uses a science and education resource recommendation method based on federated collaborative filtering. This scheme can input users and resources into the graph neural network for prediction, and recommend resources with higher scores to each user. This recommendation method not only considers the personalized needs of users, but also fully utilizes cross-institution science and education resources to achieve accurate recommendation services. Since this scheme uses a knowledge graph as a feature propagation medium and realizes feature aggregation at the semantic level, it can effectively resist the performance degradation problem of the recommendation model caused by user preference heterogeneity and resource heterogeneity. This helps to improve the accuracy and stability of the recommendation and improve the user experience.

[0060] The present application uses an open knowledge graph as a barrier to break through the science and education resources and user preference information preserved by different educational institutions, and realizes the learning of a distributed recommendation model. This design not only ensures the security and privacy of data, but also promotes cross-institution cooperation and resource sharing.

[0061] In summary, the technical scheme realizes cross-institutional science and education resource recommendation services through the steps of constructing a knowledge graph, extracting features, training a model, and optimizing recommendations. This scheme not only has the characteristics of efficiency, accuracy, and flexibility, but also fully considers privacy protection and security issues, providing strong technical support for the sharing and utilization of science and education resources. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 The flowchart of the present application.

[0063] Figure 2 System model diagram. DETAILED DESCRIPTION

[0064] The present application is further described below in conjunction with the accompanying drawings:

[0065] Example 1, please refer to Figure 1 The present application provides a multi-source software-defined science and education resource recommendation method based on a knowledge graph, including:

[0066] Based on artificial intelligence knowledge entities and the relationships between entities, an artificial intelligence knowledge graph is constructed. Science and education resource data of science and education institutions is obtained, and knowledge entities in the science and education resource data are extracted and connected to the artificial intelligence knowledge graph to obtain a science and education resource knowledge graph.

[0067] Based on a graph neural network, local resource features and user features are extracted to obtain resource features and user features of each local client;

[0068] Based on the science and education resource knowledge graph, a local knowledge graph is constructed, and the resource features of each local client are used as initial features for model training until the model converges.

[0069] The user and the resource are input into the graph neural network for prediction, and the resources with high scores are recommended to each user.

[0070] By constructing an artificial intelligence knowledge graph, this method effectively organizes the knowledge entities and relationships between entities in the field of artificial intelligence, providing a rich knowledge background for subsequent science and education resource recommendation.

[0071] By combining the science and education resource data of science and education institutions with the artificial intelligence knowledge graph, a science and education resource knowledge graph is constructed, realizing the deep integration of knowledge and resources. This integration method not only improves the availability of resources, but also promotes the dissemination and sharing of knowledge.

[0072] By using a graph neural network to extract local resource features and user features, this method can more accurately capture the complex relationships between resources and users, improving the accuracy and robustness of feature representation.

[0073] By training the model with the resource features and user features of each client as initial features, the method can learn more representative features, providing strong support for subsequent recommendations.

[0074] The method adopts a model training method based on the knowledge graph of science and education resources, and iteratively optimizes the model parameters until the model converges. This training method not only improves the accuracy of the model, but also enhances the generalization ability of the model.

[0075] By predicting through the graph neural network, the method can quickly respond to user needs and provide real-time science and education resource recommendation services.

[0076] By inputting users and resources into the graph neural network for prediction and recommending resources based on the score, the method can more accurately match user needs and resource characteristics, improving the accuracy and satisfaction of recommendations.

[0077] Since the method considers the integration and utilization of multi-source science and education resources, it can provide users with more comprehensive and comprehensive science and education resource recommendation results.

[0078] By constructing a unified knowledge graph of science and education resources, the method promotes cooperation and resource sharing among different science and education institutions. This cooperation not only improves the efficiency of resource utilization, but also promotes knowledge exchange and dissemination. The method provides a common platform and interface for science and education institutions, facilitating data exchange and resource complementation between them, thereby promoting the collaborative development of science and education.

[0079] In summary, the multi-source software-defined science and education resource recommendation method based on knowledge graph realizes the efficient integration and personalized recommendation of science and education resources through the steps of constructing a knowledge graph, extracting features, training a model, and optimizing recommendations. Its technical effects not only lie in the improvement of recommendation accuracy, but also in the promotion of cross-institutional cooperation and resource sharing, providing strong technical support and driving force for the development of science and education.

[0080] In embodiment 2, the present application provides a multi-source software-defined science and education resource recommendation method based on a knowledge graph, which specifically includes:

[0081]

[0082] The system model is:

[0083] As shown in Figure 2 , the system includes different science and education institutions . For each institution , it includes users, science and education resources, and the preference relationship between users and science and education resources ; in addition, there is a knowledge graph in the field of artificial intelligence in the system, which contains a knowledge graph composed of knowledge points related to the field of artificial intelligence . The main purpose of the software-defined science and education resource multi-source collaborative recommendation algorithm based on the knowledge graph is to learn the characteristics of users and resources from the user preference relationship network of science and education resources maintained by different science and education institutions, and to fuse them with the help of artificial intelligence knowledge graph, so as to extract more comprehensive user and resource characteristics, and to realize more accurate resource recommendation.

[0084] 1. Software-defined science and education resource knowledge graph based on knowledge graph, mainly including the following steps:

[0085] Artificial intelligence knowledge graph construction: construct a knowledge corpus in the field of artificial intelligence, use entity recognition algorithm and relation extraction algorithm to extract knowledge entities and entity relationships in the field of artificial intelligence, and form a knowledge graph in the field of artificial intelligence ;

[0086] Local science and education resource linking: for each science and education institution , process the metadata, text, video, image, file and other types of data contained in the science and education resources it holds, extract the knowledge entities and associations contained therein, and align and connect them with the constructed artificial intelligence knowledge graph, to construct a software-defined science and education resource knowledge graph that integrates artificial intelligence knowledge points and science and education resources ;

[0087] Local resource feature and user feature extraction based on graph neural network: at each client, the user-resource preference relationship is graph modeled to construct a user preference network ; a cross-entropy loss function is constructed using the preference relationship to train the model, and when the model converges, the local resource features and user features of each client are obtained;

[0088] Cross-entity feature propagation based on knowledge graph: each entity uses the local science and education resource set and the shared artificial intelligence knowledge graph to construct a local knowledge graph , uses the science and education resource features learned in step (3) as initial features, inputs them into a convolutional neural network for feature learning, and learns local features for each entity , in each science and education institution ; then, all are shared between each institution; then, each client fuses the knowledge point features with the help of an attention mechanism to obtain comprehensive knowledge point features ;

[0089] Local feature update: Given Update local resource features, and update the updated resource features with Input the graph neural network, complete the local user feature and resource feature update;

[0090] Model training: steps (4) - (5) are repeatedly run until the model converges;

[0091] Based on the federal collaborative filtering of science and education resources recommendation: input the user and resource into the graph neural network for prediction, and recommend the resources with higher scores to each user.

[0092] 2. The knowledge graph construction of 1 mainly includes the following processes:

[0093] Corpus construction: collect and annotate text data, and divide it into training set, validation set and test set. Then, the text is segmented, stop words are removed and encoded, so that the model can understand.

[0094] Named entity recognition: on the given corpus, the BERT language processing model is used to extract text features, then the model is trained and the performance is evaluated on the validation set, and the parameters are adjusted to improve the effect. The trained model is applied to the test set for named entity recognition.

[0095] Relation extraction: extract relevant features, use recurrent neural network for training, and evaluate model performance on the validation set.

[0096] 3. The science and education resource linking of 1 mainly includes the following steps:

[0097] Multimodal science and education resource feature extraction: the multimodal entity recognition process first collects and annotates a dataset containing text and corresponding images, and performs word segmentation and word embedding processing on the text to generate vector representation, while using convolutional neural network to extract image features. In model design, an independent text processing network and image processing network are constructed, and Transformer or convolutional neural network is used for processing respectively, and the processed results are weighted and fused to generate a comprehensive feature representation, which is used to output the category of each entity.

[0098] Entity matching of science and education resources to knowledge graph: measure the cosine similarity between the learned features and the knowledge point features in the knowledge graph, and select the nearest result for linking, complete the knowledge linking of science and education resources to entity in the knowledge graph.

[0099] 4. The resource feature and user feature extraction based on graph neural network of 1 includes the following steps:

[0100] Model building: facing the science and education resource recommendation scene, based on graph convolutional neural network to build recommendation network;

[0101] Model training: on the basis of the constructed model, construct the loss function, and input the data into the model for loss calculation and update the model parameters through back propagation until the model converges;

[0102] 5. The model building as shown in 4, mainly includes the following modules:

[0103] Feature embedding module: in each layer, for each edge type An Dimension embedding , and at the same time using edge type embedding and node embedding to calculate attention score:

[0104]

[0105] Among them Indicates the type of edge between node And node , Is a learnable matrix used to convert type embedding;

[0106] Residual link module: pre-activation residual connection is added, which is used for cross-layer node representation, assuming Is the attention weight of edge , Is the activation function, when the feature dimension changes in the Layer, the scheme uses a learnable linear conversion To convert the features, and the converted features are aggregated by residual:

[0107]

[0108] Similarly, for the learning of edge features, a residual module is also introduced, as follows:

[0109]

[0110] Multi-head attention module: in order to further enhance the learning ability of the model, multi-head attention mechanism is introduced, with Independent attention modules respectively learn features, and the learned features are spliced to obtain the final feature representation, and the corresponding update rule is as follows:

[0111]

[0112]

[0113]

[0114] Final output layer The output result of the layer is:

[0115]

[0116] Preference information prediction: the output features are then input into a prediction module built based on a multilayer perception mechanism, specifically, for user features and resource features , the features are spliced and input into a multilayer perception mechanism to obtain a prediction score as follows:

[0117]

[0118] 6. The model training in 4 mainly includes the following steps:

[0119] Cross-entropy construction: for the output prediction label result set and the real label , the cross-entropy loss is constructed as follows:

[0120]

[0121] Gradient propagation: then, the gradient is calculated by using the back propagation of the neural network and the parameters are optimized, and it is evaluated whether the stopping condition is reached, if the model does not converge, the step in 5 is repeated until the model converges.

[0122] In another embodiment of the present application, a multi-source software-defined science and education resource recommendation system based on a knowledge graph is provided, which can be used to implement the multi-source software-defined science and education resource recommendation method based on the knowledge graph, and specifically, the system comprises:

[0123] A science and education resource knowledge graph acquisition module is configured to construct an artificial intelligence knowledge graph based on artificial intelligence knowledge entities and the relationships between the entities, acquire science and education resource data of a science and education institution, extract knowledge entities in the science and education resource data, connect the knowledge entities to the artificial intelligence knowledge graph, and obtain a science and education resource knowledge graph.

[0124] A feature extraction module is configured to extract local resource features and user features based on a graph neural network to obtain resource features and user features of each local client.

[0125] A model training module is configured to construct a local knowledge graph based on the science and education resource knowledge graph, and train the model based on the resource features and user features of each local client as initial features until the model converges.

[0126] A prediction output module is configured to input users and resources into a graph neural network for prediction and recommend resources with high scores to each user.

[0127] The division of the modules in the embodiments of the present application is illustrative, and is merely logical functional division. In actual implementation, another division manner can be used. In addition, each functional module in each embodiment of the present application can be integrated in one processor, or can be physically separated, or two or more modules can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module.

[0128] In still another embodiment of the present application, a computer device is provided, which includes a processor and a memory. The memory is configured to store a computer program, and the computer program includes program instructions. The processor is configured to execute the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), and can also be another general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or another programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and the like. The processor is a computing core and a control core of the terminal, and is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement a corresponding method flow or a corresponding function. The processor in the embodiments of the present application can be used for the operation of the method for recommending multi-source software-defined science and education resources based on a knowledge graph.

[0129] In still another embodiment of the present application, the present application also provides a storage medium, specifically a computer readable storage medium (Memory), which is a memory device in a computer device, used for storing programs and data. It can be understood that the computer readable storage medium here can include the built-in storage medium in the computer device, and of course can also include the extended storage medium supported by the computer device. The computer readable storage medium provides a storage space, which stores the operating system of the terminal. Moreover, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer readable storage medium here can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory. One or more instructions stored in the computer readable storage medium can be loaded and executed by the processor to realize the corresponding steps of the knowledge graph-based multi-source software-defined education resource recommendation method in the above embodiments.

[0130] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0131] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device implemented in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that implements the function specified in the flow or flows and / or block or blocks.

[0132] These computer program instructions can also be stored in a computer readable memory capable of directing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction apparatus, which implements the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocksFigure 1 the function specified in the one or more blocks.

[0133] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operation steps are performed on the computer or other programmable data processing devices to generate computer-implemented processing, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the flow Figure 1 the flow or flows and / or blocks Figure 1 the steps of the function specified in the one or more blocks.

[0134] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, but not to limit it, although the above embodiments of the present application have been described in detail, those skilled in the art should understand: the specific embodiments of the present application can be modified or replaced by the same, without departing from the spirit and scope of the present application, any modification or equivalent replacement, which should be covered within the scope of protection of the claims of the present application.

Claims

1. A multi-source software-defined science and education resource recommendation method based on a knowledge graph, characterized in that, Comprise: Based on artificial intelligence knowledge entity and the relationship between entities, build an artificial intelligence knowledge graph; Obtain the scientific and educational resources data of scientific and educational institutions, extract the knowledge entities in the scientific and educational resources data, and connect to the artificial intelligence knowledge graph to obtain the scientific and educational resources knowledge graph; Construct a user preference network model, extract local resource features and user features, and obtain local resource features and user features of each client; Based on the scientific and educational resources knowledge graph, build a local knowledge graph, and use the resource features of each client as initial features for model training until the model converges; Based on the federated collaborative filtering of scientific and educational resources, input the user and resource into the graph neural network for prediction, and recommend high-score resources to each user; Based on the software-defined scientific and educational resources knowledge graph of the knowledge graph, comprising the following steps: Artificial intelligence knowledge graph construction: construct a knowledge corpus in the field of artificial intelligence, use entity recognition algorithm and relation extraction algorithm to extract knowledge entities and entity relationships in the field of artificial intelligence, and form a knowledge graph in the field of artificial intelligence ; Local science and education resource link: for each science and education institution , the metadata, text, video, image, file and other types of data contained in the science and education resources held by the science and education institution are processed, the knowledge entities and the associations contained therein are extracted, and the knowledge entities and the associations are aligned and connected with the artificial intelligence knowledge graph constructed, to construct a software-defined science and education resource knowledge graph that integrates artificial intelligence knowledge points and science and education resources ; Local resource feature and user feature extraction based on graph neural network: at each client, by graph modeling the user-resource preference relationship, a user preference network is constructed ; Use the preference relationship to build a cross-entropy loss function to train the model, and obtain the resource features and user features of each client when the model converges; Cross-entity feature propagation based on knowledge graph: each entity Utilize local science and education resource set And shared artificial intelligence knowledge graph Construct local knowledge graph , use the learned science and education resource features as initial features, input them into the convolutional neural network for feature learning, and each for Each entity in , Learn local features ; then, all Between each institution; then, each client fuses the knowledge point features with the help of the attention mechanism to obtain comprehensive knowledge point features ; Local feature update: Given Update local resource features, and update the updated resource features with Input the graph neural network, complete the local user features and resource feature update.

2. The knowledge graph-based multi-source software-defined edutainment resource recommendation method according to claim 1, characterized in that, Based on artificial intelligence knowledge entities and the relationship between entities, build an artificial intelligence knowledge graph, comprising: Build a knowledge corpus in the field of artificial intelligence, use entity recognition algorithms and relationship extraction algorithms to extract knowledge entities and relationships in the field of artificial intelligence, and form a knowledge graph in the field of artificial intelligence; Specifically: Collect and label text data, divide it into training set, validation set and test set, perform word segmentation, remove stop words and encode the text; On the given corpus, use the BERT language processing model to extract text features, then use the trained model to evaluate the performance on the validation set, and apply the trained model to the test set for named entity recognition; Extract relevant features, use recurrent neural networks for training, and evaluate model performance on the validation set. 3.The knowledge graph based multi-source software-defined edutainment resource recommendation method according to claim 1, characterized in that, Obtain the scientific and educational resources data of scientific and educational institutions, extract the knowledge entities in the scientific and educational resources data, and connect to the artificial intelligence knowledge graph to obtain the scientific and educational resources knowledge graph, comprising: Process the metadata, text, video, image, and file data contained in each scientific and educational resource held by the scientific and educational institution, extract the knowledge entities and their associations, and align and connect them with the constructed artificial intelligence knowledge graph to build a software-defined scientific and educational resources knowledge graph that integrates artificial intelligence knowledge points and scientific and educational resources: The multi-modal entity recognition process first starts with collecting and labeling a dataset containing text and corresponding images, performs word segmentation and word embedding processing on the text to generate vector representations, and uses a convolutional neural network to extract image features; Build independent text processing network and image processing network, use Transformer or convolutional neural network for processing respectively and weighted fusion of the processed results to generate comprehensive feature representation, use the feature representation to output the category of each entity; Measure the cosine similarity between the learned features and the knowledge point features in the knowledge graph, and select the closest result for linking to complete the knowledge connection of scientific and educational resources to entities in the KG.

4. The knowledge graph-based multi-source software-defined edutainment resource recommendation method according to claim 1, characterized in that, A user preference network model is constructed, and local resource features and user features are extracted to obtain local resource features and user features of each client, including: At each client, a user preference network is constructed by graph modeling of the "user-resource" preference relationship; a cross-entropy loss function is constructed using the preference relationship to train the model, and when the model converges, local resource features and user features of each client are obtained.

5. The knowledge graph-based multi-source software-defined edutainment resource recommendation method according to claim 4, characterized in that, Based on the knowledge graph of scientific and educational resources, a local knowledge graph is constructed, and the local resource features of each client are used as initial features for model training until the model converges, including: For scientific and educational resource recommendation scenarios, a recommendation network is constructed based on graph convolutional neural networks. Based on the constructed model, a loss function is constructed, and data is input into the model for loss calculation and model parameter update through back propagation until the model converges.

6. The knowledge graph-based multi-source software-defined edutainment resource recommendation method according to claim 5, characterized in that, Model building, including: Feature embedding module: at each layer, for each edge type assign one dimensional embedding and simultaneously using edge type embeddings and node embeddings to compute attention scores: wherein represents a type of an edge between a node and a node , is a learnable matrix used to transform type embeddings; Residual connection module: add pre-activation residual connection for cross-layer node representation, assuming is the attention weight of edge , is the activation function, when the feature dimension of the first layer changes, the scheme uses a learnable linear transformation to convert the features, and the converted features are aggregated by residual Similarly, the residual module is also introduced for edge feature learning, as follows: Multi-head attention module: introduce multi-head attention mechanism, with the help of an independent attention module respectively learns the characteristics, and the learned characteristics are spliced to obtain the final feature representation. The corresponding update rule is as follows: Final output layer The output of the layer is: Preference information prediction: the output features are then input into a prediction module built based on a multi-layer perception mechanism. Specifically, for user features and resource features , the features are concatenated and input into a multi-layer perception mechanism to obtain a prediction score as follows: 。 7. The knowledge graph-based multi-source software-defined edutainment resource recommendation method according to claim 5, characterized in that, Model training, including: Constructing cross-entropy: for the set of predicted label results of the output and the true labels , the cross-entropy loss is constructed as follows: Gradient propagation: then use the back propagation of the neural network to calculate the gradient and optimize the parameters, and evaluate whether the stopping condition is reached, if the model has not converged, continue training until the model converges.

8. The knowledge graph-based multi-source software-defined science and education resource recommendation system capable of being used in the knowledge graph-based multi-source software-defined science and education resource recommendation method according to claim 1, characterized in that, Including: The scientific and educational resource knowledge graph acquisition module is used to construct an artificial intelligence knowledge graph based on artificial intelligence knowledge entities and the relationships between entities; Obtain scientific and educational resource data of scientific and educational institutions, extract knowledge entities in scientific and educational resource data, and connect to the artificial intelligence knowledge graph to obtain the scientific and educational resource knowledge graph; The feature extraction module is used to extract local resource features and user features based on graph neural networks to obtain local resource features and user features of each client; The model training module is used to construct a local knowledge graph based on the scientific and educational resource knowledge graph, and the local resource features of each client are used as initial features for model training until the model converges; The prediction output module is used to input users and resources into the graph neural network for prediction, and recommend resources with high scores to each user.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the multi-source software-defined scientific and educational resource recommendation method based on the knowledge graph according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1-9. The computer program is executed by the processor to implement the steps of the multi-source software-defined scientific and educational resource recommendation method based on the knowledge graph according to any one of claims 1 to 7.

Images