CN112529750A - Learning event recommendation method and system based on graph neural network model - Google Patents
Learning event recommendation method and system based on graph neural network model Download PDFInfo
- Publication number
- CN112529750A CN112529750A CN202011537470.3A CN202011537470A CN112529750A CN 112529750 A CN112529750 A CN 112529750A CN 202011537470 A CN202011537470 A CN 202011537470A CN 112529750 A CN112529750 A CN 112529750A
- Authority
- CN
- China
- Prior art keywords
- node
- learning
- nodes
- learning event
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003062 neural network model Methods 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 title claims abstract description 24
- 239000013598 vector Substances 0.000 claims description 89
- 238000004364 calculation method Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 6
- 230000002776 aggregation Effects 0.000 claims description 5
- 238000004220 aggregation Methods 0.000 claims description 5
- 235000013399 edible fruits Nutrition 0.000 claims description 5
- 230000004931 aggregating effect Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims 4
- 238000013527 convolutional neural network Methods 0.000 claims 1
- 238000013135 deep learning Methods 0.000 abstract description 3
- 239000011159 matrix material Substances 0.000 description 21
- 230000006870 function Effects 0.000 description 8
- 230000000694 effects Effects 0.000 description 7
- 238000013136 deep learning model Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000007596 consolidation process Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
- G06Q50/205—Education administration or guidance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Educational Technology (AREA)
- Educational Administration (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a learning event recommendation method based on a graph neural network model, which relates to the technical field of deep learning. In addition, the embodiment of the invention also provides a learning event recommendation system based on the graph neural network model.
Description
Technical Field
The invention relates to the technical field of deep learning, in particular to a learning event recommendation method and system based on a graph neural network model.
Background
Learning event recommendation in the existing online education learning system is knowledge point recommendation, learning events under the same knowledge point are carried out according to a preset learning process, one conventional process is watching a knowledge point teaching video, pushing exercise questions, displaying question answers and analyzing feedback after correction, however, different users can obtain different learning results from different learning events, and a solidified or fully independent user selection mode cannot realize comprehensive consideration of user characteristics, learning event characteristics and the relation between the users and the learning events, and recommend learning events which accord with user preferences and help the users to improve the maximum learning effect.
At present, the recommendation of learning events is mainly carried out based on a deep learning model, and the scheme has the following defects:
(1) although the recommendation scheme of the deep learning model is mature, the scheme based on the deep learning model is difficult to effectively adapt to the graph data, because the graph data is a relatively complex class of data;
(2) because the deep learning model processes the structure information and the attribute information of the graph data independently, the structure information and the attribute information of the graph cannot be naturally fused for learning;
(3) the deep learning model can make the low-frequency learning events or newly-added learning events difficult to recommend, because the learning events which are clicked by no user or few users are difficult to appear in the training data, the recommended learning events are inaccurate;
(4) the user is not provided with interpretable information, and the user is difficult to know the reason of algorithm recommendation, so that the learning effect is poor.
Disclosure of Invention
In order to overcome the defects in the prior art, the embodiment of the invention provides a learning event recommendation method and system based on a graph neural network model.
In a first aspect, a learning event recommendation method based on a graph neural network model provided in an embodiment of the present invention includes the following steps:
respectively acquiring each node in a pre-constructed graph, wherein each node comprises a plurality of user nodes, a plurality of learning event nodes and a plurality of achievement nodes, a first type of connecting edge is established between each user node and the corresponding learning event node, and a second type of connecting edge is established between each user node and the corresponding achievement node;
inputting the atlas into a trained atlas neural network model to obtain a characteristic vector set corresponding to each node;
respectively calculating the similarity between the feature vector corresponding to each user node in the feature vector set and the feature vector corresponding to each learning event node;
and selecting learning event nodes meeting set conditions from the feature vector set according to the similarity, and pushing the learning event nodes to a client.
Preferably, selecting a learning event node satisfying a set condition from the feature vector set and pushing the learning event node to a client includes:
respectively calculating the similarity between the feature vector corresponding to the learning event node and the feature vector corresponding to each fruit node by using a similarity calculation method;
and (4) taking the result nodes with the similarity meeting the set conditions as interpretability information of the corresponding learning events and pushing the result nodes to the client.
Preferably, the map construction process comprises:
acquiring personal attribute data, learning event data and learning result data of a plurality of users within a set time period to obtain a sample data set;
respectively taking personal attribute data, learning event data and learning result data of a plurality of users as each node, wherein a first-class connecting edge is established between the user node and a learning event node corresponding to a learning event of the user in a preset time period; a second type connecting edge is established between the user node and an achievement node corresponding to the achievement of the user in the preset time period;
preferably, inputting the atlas into the trained atlas neural network model to obtain a feature vector set corresponding to each node includes:
acquiring a feature vector corresponding to each node in the map according to the original attribute of the node;
aggregating the feature vectors of the target nodes and the feature vectors of the adjacent nodes of the target nodes to generate the feature vectors of the target nodes, performing graph representation based on a neighborhood aggregation mode, representing the whole graph into a low-dimensional, real-valued and dense vector form, and obtaining a feature vector set corresponding to each node.
Preferably, the graph neural network model is a graph convolution neural network model.
In a second aspect, a learning event recommendation system based on a graph neural network model provided by an embodiment of the present invention includes the following modules:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for respectively acquiring a pre-constructed map, and the map comprises a plurality of user nodes, a plurality of learning event nodes and a plurality of result nodes;
the first generation module is used for inputting the atlas into the trained atlas neural network model to obtain a characteristic vector set corresponding to each node;
the first calculation module is used for calculating the similarity between the feature vector corresponding to each user node in the feature vector set and the feature vector corresponding to each learning event node;
and the first pushing module is used for selecting the characteristic vector meeting the conditions according to the numerical value of the similarity and pushing the learning event node corresponding to the characteristic vector to the client.
Preferably, the system further comprises:
the second calculation module is used for calculating the similarity between the feature vector corresponding to the learning event node and the feature vector corresponding to each fruit node by using a similarity calculation method;
and the second pushing module is used for taking the result nodes with the similarity meeting the set conditions as interpretable information of the corresponding learning events and pushing the result nodes to the client.
Preferably, the system further comprises:
the acquisition module is used for acquiring personal attribute data, learning event data and learning result data of a plurality of users to obtain a sample data set;
the second acquisition module is used for respectively taking the personal attribute data, the learning event data and the learning result data of a plurality of users as each node and acquiring a target node corresponding to each node;
the second generation module is used for generating a map according to the connection edges between the nodes and the target nodes corresponding to the nodes, wherein a first type of connection edge is established between the user node and the learning event node corresponding to the learning event of the user in a preset time period; and a second type connecting edge is established between the user node and the achievement node corresponding to the achievement of the user in the preset time period.
The learning event recommendation method and the system based on the graph neural network model have the following beneficial effects:
(1) by utilizing the graph neural network model, the characteristics of the nodes can be used, and the characteristics of the target nodes can be aggregated by using the connection relation between the nodes to obtain rich characteristic vectors, so that a more accurate recommendation effect can be achieved by aggregating graph information;
(2) the node information and the structural information of the graph can be aggregated by drawing the graph with a neural network, and are compressed to a low-dimensional vector space, and then various downstream tasks can be performed; the neural network of the graph is taken as a conductive module and can be embedded into any system supporting end-to-end learning, and the characteristic enables the neural network to be organically combined with (or combined in a fine tuning learning mode) supervised learning tasks of all levels, so that data representation more suitable for the tasks is learned;
(3) in each learning stage, a proper learning result measuring index is needed, the ability value is improved by a user when the user learns newly, memory forgetting occurs after a period of time, the reinforcement degree can better embody the review effect in the review stage, and when the user sees the assistance of the learning event to the learning result from explanatory information, the user can be promoted to confirm and select the recommended learning event and achieve better learning effect;
(4) the graph neural network model can solve the difficulty of sparsity and cold start of a deep learning recommendation system, and has strong expandability.
Drawings
Fig. 1 is a schematic flow chart of a learning event recommendation method based on a graph neural network model according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating a map provided by an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a learning event recommendation system based on a graph neural network model according to an embodiment of the present invention.
Detailed Description
The present invention will be described in detail with reference to the following embodiments.
As shown in fig. 1, a learning event recommendation method based on a graph neural network model according to an embodiment of the present invention includes the following steps:
s101, respectively obtaining each node in a pre-constructed graph, wherein each node comprises a plurality of user nodes, a plurality of learning event nodes and a plurality of achievement nodes, a first type of connection edge is established between each user node and the corresponding learning event node, and a second type of connection edge is established between each user node and the corresponding achievement node.
The learning event nodes are man-machine interaction of users and a learning system, relevant attribute features can be recorded into a database for extraction and use when the users generate the learning events through a lesson monitoring system, the learning event nodes can comprise a knowledge point watching teaching video, a question making function, a step-by-step solution function, an answer checking and analysis function, an error book checking or practicing function, a note checking or recording function, a report page watching function, a question asking function for a teacher and the like, one learning event node represents one learning event, and the learning event node features can comprise types, duration, times, operation periods and the like.
The result nodes are multiple measuring indexes of the learning result of the user and can comprise a positive answer rate, a capability value, a mastery degree, a consolidation degree and the like, one result node represents one measuring index, the result node characteristics can comprise data counted according to the measuring indexes or grades classified according to the data, and the result nodes can be divided into multiple levels according to the mastery degree, or can be measured values obtained through a machine learning algorithm.
A first type connecting edge is established between the user node and a learning event node corresponding to a learning event of the user in a preset time period; and a second type of connecting edge is established between the user node and the achievement node corresponding to the achievement of the user in the preset time period to form the graph information provided for the graph neural network model.
Fig. 2 is a graph of An exemplary style in An embodiment of the present invention, where a user node U (U1, U2 … … Un), a learning event a (a1, a2 … … An), and a result node K (K1, K2 … … Kj). The obtained map can accurately represent the association relation between the user and the learning event and the learning result.
In a preferred embodiment, a user U1 watches the knowledge point teaching video a1 and the exercise questions a2, and the ability value reaches the scholars level K1, first type connecting edges are established between the knowledge point teaching video a1 and the user U1, and between the exercise questions a2 and the user U1, and second type connecting edges are established between the user U1 and the scholars level K1, so that map information provided for the map neural network model is generated.
In another preferred embodiment, the positive answer rate of the user U2 after making the question a2 results in the result K2, a first type of connection edge is established between the question a2 and the user U2, and a second type of connection edge is established between the U2 and the K2 node.
In another preferred embodiment, the mastery degree of the user U3 doing the question a2 and the answer and analysis A3 is result K3, a first type of connecting edge is established between the question a2 and the user U3, between the analysis A3 and the user U3, and a second type of connecting edge is established between the user U3 and the result K3.
As a specific embodiment, one user node represents one user, that is, the user node and the user are in a one-to-one correspondence relationship. When the number of users is large, the users may be classified into a plurality of categories, and users of one category may have the same region or textbook, in which case one user node represents one category of users, not a single user. The original attribute characteristics of the user comprise personal attribute information corresponding to a user node, and the personal attribute information can comprise personal information of user gender, grade, region and the like or related information of selected subject, course and the like; the learning event node is man-machine interaction between a user and a learning system, and can record related attribute characteristics into a database for extraction and use when the user generates the learning events through a lesson monitoring system, wherein the learning event node can watch a knowledge point teaching video, do questions, gradually solve the questions according to steps, check answers and analyze, check or practice wrong problem books, check or record notes, watch report pages, ask questions for teachers and the like, one learning event node represents one learning event, and the learning event node characteristics can comprise types, duration, times, operation periods and the like; the result nodes are multiple measuring indexes of the learning result of the user and can comprise a positive answer rate, a capability value, a mastery degree, a consolidation degree and the like, one result node represents one measuring index, the result node characteristics can comprise data counted according to the measuring indexes or grades classified according to the data, and the result nodes can be divided into multiple levels according to the mastery degree, or can be measured values obtained through a machine learning algorithm.
And S102, inputting the map into the trained neural network model to obtain a characteristic vector set corresponding to each node.
The graph neural network model obtains the feature vector of each node according to the feature and the structure corresponding to each node in the graph, that is, the embedded vector of the target node is generated according to the adjacent node of the target node, and aggregation is performed to obtain the corresponding feature of each node in the graph.
The embedded vector of the target node is generated according to the adjacent nodes of the target node, the feature vector corresponding to each node is generated, and specifically, the original attribute features of the user node, the learning event node and the result node are subjected to graph embedding. The graph embedding belongs to the category of representation learning, and may also be called network embedding, graph representation learning, network representation learning, and the like, that is, the whole graph is represented in a low-dimensional, real-valued and dense vector form, and the graph neural network model may be applied to the graph embedding to obtain the vector representation of the graph or the graph nodes. Specifically, neighborhood aggregation is to generate embedded vectors of target nodes according to the neighboring nodes of the target nodes, each node has embedded vectors at each layer, the model can have any depth, and the embedded vectors of the nodes at the 0 th layer are the input features of the embedded vectors.
In a preferred embodiment, the graph neural network model is a graph convolution neural network model.
The graph convolution neural network model can be represented by the following formula:
where a is an adjacency matrix used to represent the connection between nodes. The number of the nodes is n, A is an n multiplied by n matrix, and the value of the matrix is 1 to indicate that edges exist among the nodes, namely, the nodes have a connection relation; a. the～I is an identity matrix, i.e. adding a self-join to each node in the graph to introduce the characteristics of the node itself, a + I～Is a symmetric matrix; d is a degree matrix, only the diagonal has a value, and the value is the degree of the corresponding node; w(l)Is the weight matrix of the l layer; h(l)The characteristic matrix of the l layer is used for representing the characteristics of the nodes; σ is a nonlinear activation function.
At initialization, the feature matrix of the first layer is the input matrix, denoted by the symbol H0X is the feature matrix of the node, the feature dimension is k, and X is an n × k matrix. Each layer of input includes an adjacent matrix A and a characteristic matrix H of a node, an inner product is directly made, then a learnable parameter matrix is multiplied, and nonlinear activation is carried out, so that the problem that the original distribution of characteristics is changed by multiplying the characteristic matrix, some unpredictable problems are caused, and the adjacent matrix A is subjected to normalization processing. Specifically, to make each row of A add to 1, the inverse of the degree matrix, D, may be multiplied-1Further converting D-1And (3) taking apart the product of the A and multiplying the product to obtain a symmetrical normalized matrix, and obtaining the characteristic matrix output fused with the characteristics and the structure of the neighbor node through iterative convergence of a plurality of layers, wherein generally speaking, the number of layers of the graph convolution neural network is not large, and the effect of 2-3 layers is good.
S103, respectively calculating the similarity between the feature vector corresponding to each user node in the feature vector set and the feature vector corresponding to each learning event node.
After the characteristics of each node are obtained, the similarity between the characteristics corresponding to each user node and each learning event node is calculated, and the learning event with the similarity reaching the preset requirement is determined as a candidate list for pushing the learning event to the corresponding user. The characteristic majority output by the graph neural network model is a characteristic vector in the case of characteristic majority, and various algorithms for calculating the similarity between vectors are available, such as Euclidean distance algorithm, cosine similarity algorithm, Manhattan distance algorithm and the like. The learning event with the similarity reaching the preset requirement may be the learning event with the maximum similarity or the learning event with the similarity ranking several times, for example, the first 3 learning events with the greater similarity or the 1 learning event with the maximum similarity serve as a candidate list for pushing the learning event to the corresponding user. In an alternative embodiment, when the user activates "next" in the learning system, that is, determines the learning event to be pushed to the user, the user clicks "next" to enter the operation of the learning event, and in an actual situation, the learning event often needs to be supported by corresponding learning resources, so that when the corresponding resources of the learning event with the largest similarity are unavailable, recommendation is made from the candidate list.
And S104, selecting learning event nodes meeting set conditions from the feature vector set according to the similarity, and pushing the learning event nodes to the client.
In another embodiment, after obtaining the features, an access request of a user is waited, for example, an operation of clicking an entry of the learning system by the user is regarded as that the user initiates an access request to a designated learning system, the user is a target user, after the target user enters the learning system, a target feature of a user node corresponding to the target user is obtained, a similarity between the target feature and features of each learning event node is calculated, a learning event with the similarity reaching a preset requirement is determined as a target learning event, and the target learning event is pushed to the target user.
In addition, when the result node is designated as an output node, the characteristics corresponding to the result node are output through the graph neural network model, the similarity between the learning event node and the characteristics corresponding to the result node is calculated, the user result of the learning event node with the similarity meeting the preset requirement is used as the interpretability information of the corresponding learning event, specifically, the teaching video for watching the knowledge point represented by the learning event node A1 is determined as the learning event pushed to the user U1, calculating the similarity between the characteristics of the learning event node A1 and the characteristics corresponding to each result node, selecting the result node which is the most matched or more matched with the learning event node A1, when the explanatory information as the learning event node is the ability value K1, the user can see that the user is recommended to watch the knowledge point teaching video according to the learning result of the ability value when the user clicks the button triggering the learning event in the learning system.
Furthermore, the graph neural network model obtains characteristics through the information of the adjacent matrix, and for low-frequency knowledge points, even if a small number of users click, connecting edges between the graph neural network model and corresponding user nodes can be established, so that the graph carries the information of the low-frequency knowledge points; and connecting the newly added learning event to the corresponding class node through the class to which the newly added learning event belongs to the condition that the newly added learning event does not have any click so as to add the newly added learning event to the map. Specifically, a new learning event node used for representing a new learning event is added in the map; and determining the category of the newly added knowledge point, and connecting the newly added knowledge point node to the category node corresponding to the category to which the newly added knowledge point node belongs through the connecting edge. In another embodiment, the connection edge may be established by randomly pushing the newly added learning event to a plurality of users. Specifically, a newly added learning event is determined, the newly added learning event is randomly pushed to a plurality of users, if a user clicks the newly added learning event within a preset time period, a newly added learning event node for representing the newly added learning event is added in a map, and a connecting edge is established between a user node sending out the clicking operation and the newly added learning event node.
Optionally, selecting a feature vector meeting the set condition and pushing the learning event node corresponding to the feature vector to the client includes:
and respectively calculating the similarity between the feature vector corresponding to the learning event node and the feature vector corresponding to each achievement node by using a similarity calculation method, selecting the achievement node matched with the learning event node, and pushing the achievement node to the client.
Optionally, the map construction process includes:
collecting user data, learning event data and learning result data of a plurality of users within a set time period to obtain a sample data set;
respectively taking personal attribute data, learning event data and learning result data of a plurality of users as each node;
a first type of connecting edge is established between the user node and a learning event node corresponding to the learning event of the user in a preset time period, and a second type of connecting edge is established between the user node and an achievement node corresponding to the achievement of the user in the preset time period to generate a map.
Optionally, inputting the graph spectrum into the trained graph neural network model, and obtaining the feature vector set corresponding to each node includes:
acquiring a feature vector corresponding to each node in the map according to the original attribute of the node;
aggregating the feature vectors of the target nodes and the feature vectors of the adjacent nodes of the target nodes to generate the feature vectors of the target nodes, performing graph representation based on a neighborhood aggregation mode, representing the whole graph into a low-dimensional, real-valued and dense vector form, and obtaining a feature vector set corresponding to each node.
Optionally, the graph neural network model is a graph convolution neural network model.
The present application is not limited to the above-mentioned graph nodes and structures, as long as the graph data of the nodes and structures in the graph, such as graph mapping and D2R conversion, can be generated by using the knowledge extraction technology and tool oriented to the structured data.
According to the learning event recommendation method based on the graph neural network model, provided by the embodiment of the invention, each node in a pre-constructed graph is obtained respectively, the graph is input into a trained graph neural network model to obtain a feature vector set corresponding to each node, the similarity between a feature vector corresponding to each user node in the feature vector set and a feature vector corresponding to each learning event node is calculated respectively, and the learning event node meeting set conditions is selected from the feature vector set according to the similarity and is pushed to a client, so that the recommendation accuracy and effect are improved.
The learning event recommendation system based on the graph neural network model provided by the embodiment of the invention comprises the following modules:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for respectively acquiring a pre-constructed map, and the map comprises a plurality of user nodes, a plurality of learning event nodes and a plurality of result nodes;
the first generation module is used for inputting the atlas into the trained atlas neural network model to obtain a characteristic vector set corresponding to each node;
the first calculation module is used for calculating the similarity between the feature vector corresponding to each user node in the feature vector set and the feature vector corresponding to each learning event node;
and the first pushing module is used for selecting the characteristic vector meeting the conditions according to the numerical value of the similarity and pushing the learning event node corresponding to the characteristic vector to the client.
Preferably, the system further comprises:
the second calculation module is used for calculating the similarity between the feature vector corresponding to the learning event node and the feature vector corresponding to each fruit node by using a similarity calculation method;
and the second pushing module is used for taking the result nodes with the similarity meeting the set conditions as interpretable information of the corresponding learning events and pushing the result nodes to the client.
In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
It will be appreciated that the relevant features of the method and apparatus described above are referred to one another.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (10)
1. A learning event recommendation method based on a graph neural network model is characterized by comprising the following steps:
respectively acquiring each node in a pre-constructed graph, wherein each node comprises a plurality of user nodes, a plurality of learning event nodes and a plurality of achievement nodes, a first type of connecting edge is established between each user node and the corresponding learning event node, and a second type of connecting edge is established between each user node and the corresponding achievement node;
inputting the atlas into a trained atlas neural network model to obtain a characteristic vector set corresponding to each node;
respectively calculating the similarity between the feature vector corresponding to each user node in the feature vector set and the feature vector corresponding to each learning event node;
and selecting learning event nodes meeting set conditions from the feature vector set according to the similarity, and pushing the learning event nodes to a client.
2. The learning event recommendation method based on the graph neural network model according to claim 1, wherein selecting learning event nodes satisfying a set condition from the feature vector set and pushing the learning event nodes to a client comprises:
respectively calculating the similarity between the feature vector corresponding to the learning event node and the feature vector corresponding to each fruit node by using a similarity calculation method;
and (4) taking the result nodes with the similarity meeting the set conditions as interpretability information of the corresponding learning events and pushing the result nodes to the client.
3. The method for recommending learning events based on the neural network model of claim 2, wherein the construction process of the map comprises:
acquiring personal attribute data, learning event data and learning result data of a plurality of users within a set time period to obtain a sample data set;
respectively taking personal attribute data, learning event data and learning result data of a plurality of users as each node and acquiring a target node corresponding to each node;
and generating a map according to the nodes and the connecting edges between the target nodes corresponding to the nodes.
4. The method of claim 1, wherein the inputting the atlas into the trained neural network model to obtain the feature vector set corresponding to each node comprises:
acquiring a feature vector corresponding to each node in the map according to the original attribute of the node;
aggregating the feature vectors of the target nodes and the feature vectors of the adjacent nodes of the target nodes to generate the feature vectors of the target nodes, performing graph representation based on a neighborhood aggregation mode, representing the whole graph into a low-dimensional, real-valued and dense vector form, and obtaining a feature vector set corresponding to each node.
5. The method of claim 1, wherein the neural network model is a convolutional neural network model.
6. A learning event recommendation system based on a graph neural network model is characterized by comprising the following modules:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for respectively acquiring a pre-constructed map, and the map comprises a plurality of user nodes, a plurality of learning event nodes and a plurality of result nodes;
the first generation module is used for inputting the atlas into the trained atlas neural network model to obtain a characteristic vector set corresponding to each node;
the first calculation module is used for calculating the similarity between the feature vector corresponding to each user node in the feature vector set and the feature vector corresponding to each learning event node;
and the first pushing module is used for selecting the characteristic vector meeting the conditions according to the numerical value of the similarity and pushing the learning event node corresponding to the characteristic vector to the client.
7. The graph neural network model-based learning event recommendation system of claim 6, further comprising:
the second calculation module is used for calculating the similarity between the feature vector corresponding to the learning event node and the feature vector corresponding to each fruit node by using a similarity calculation method;
and the second pushing module is used for taking the result nodes with the similarity meeting the set conditions as interpretable information of the corresponding learning events and pushing the result nodes to the client.
8. The graph neural network model-based learning event recommendation system of claim 7, further comprising:
the acquisition module is used for acquiring personal attribute data, learning event data and learning result data of a plurality of users to obtain a sample data set;
the second acquisition module is used for respectively taking the personal attribute data, the learning event data and the learning result data of a plurality of users as each node and acquiring a target node corresponding to each node;
and the second generation module is used for generating a map according to the nodes and the connecting edges between the target nodes corresponding to the nodes.
9. A computer program product, characterized in that the computer program product comprises a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1-4.
10. A non-transitory computer-readable storage medium storing computer instructions that cause a computer to perform the method of any one of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011537470.3A CN112529750A (en) | 2020-12-23 | 2020-12-23 | Learning event recommendation method and system based on graph neural network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011537470.3A CN112529750A (en) | 2020-12-23 | 2020-12-23 | Learning event recommendation method and system based on graph neural network model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112529750A true CN112529750A (en) | 2021-03-19 |
Family
ID=74976539
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011537470.3A Pending CN112529750A (en) | 2020-12-23 | 2020-12-23 | Learning event recommendation method and system based on graph neural network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112529750A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113704635A (en) * | 2021-07-30 | 2021-11-26 | 西安交通大学 | Social network event recommendation method and system |
| CN113887852A (en) * | 2021-05-21 | 2022-01-04 | 东北大学 | Personnel recommendation method based on heterogeneous graph nerves |
| CN114491082A (en) * | 2022-03-31 | 2022-05-13 | 南京众智维信息科技有限公司 | Plan matching method based on network security emergency response knowledge graph feature extraction |
| CN116824495A (en) * | 2023-06-26 | 2023-09-29 | 华东交通大学 | Dangerous behavior identification method, system, storage medium and computer equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111475637A (en) * | 2020-06-24 | 2020-07-31 | 支付宝(杭州)信息技术有限公司 | Data processing and training method and device for pushing knowledge points |
| CN111753098A (en) * | 2020-06-23 | 2020-10-09 | 陕西师范大学 | Teaching method and system based on cross-media dynamic knowledge graph |
| CN111859166A (en) * | 2020-07-28 | 2020-10-30 | 重庆邮电大学 | Article scoring prediction method based on improved graph convolution neural network |
-
2020
- 2020-12-23 CN CN202011537470.3A patent/CN112529750A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111753098A (en) * | 2020-06-23 | 2020-10-09 | 陕西师范大学 | Teaching method and system based on cross-media dynamic knowledge graph |
| CN111475637A (en) * | 2020-06-24 | 2020-07-31 | 支付宝(杭州)信息技术有限公司 | Data processing and training method and device for pushing knowledge points |
| CN111859166A (en) * | 2020-07-28 | 2020-10-30 | 重庆邮电大学 | Article scoring prediction method based on improved graph convolution neural network |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113887852A (en) * | 2021-05-21 | 2022-01-04 | 东北大学 | Personnel recommendation method based on heterogeneous graph nerves |
| CN113887852B (en) * | 2021-05-21 | 2024-04-16 | 东北大学 | Personnel recommendation method based on heterogeneous graph nerves |
| CN113704635A (en) * | 2021-07-30 | 2021-11-26 | 西安交通大学 | Social network event recommendation method and system |
| CN113704635B (en) * | 2021-07-30 | 2024-05-24 | 西安交通大学 | Social network event recommendation method and system |
| CN114491082A (en) * | 2022-03-31 | 2022-05-13 | 南京众智维信息科技有限公司 | Plan matching method based on network security emergency response knowledge graph feature extraction |
| CN116824495A (en) * | 2023-06-26 | 2023-09-29 | 华东交通大学 | Dangerous behavior identification method, system, storage medium and computer equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Nagy et al. | Predicting dropout in higher education based on secondary school performance | |
| CN109919810B (en) | Student modeling and personalized course recommendation method in online learning system | |
| Tunas Bangsa Pematangsiantar | Comparison of weighted sum model and multi attribute decision making weighted product methods in selecting the best elementary school in Indonesia | |
| CN112529750A (en) | Learning event recommendation method and system based on graph neural network model | |
| TW201923696A (en) | Method for recommending a teacher in a network teaching system | |
| Hien et al. | A decision support system for evaluating international student applications | |
| Asanbe et al. | Teachers’ performance evaluation in higher educational institution using data mining technique | |
| CN111950708B (en) | Neural network structure and method for finding daily life habits of college students | |
| CN111651676B (en) | Method, device, equipment and medium for performing occupation recommendation based on capability model | |
| CN114913729B (en) | Question selecting method, device, computer equipment and storage medium | |
| Lee et al. | Machine learning approaches for learning analytics: Collaborative filtering or regression with experts | |
| Malo et al. | Employment status mobility from a life-cycle perspective: A sequence analysis of work-histories in the BHPS | |
| CN114201684A (en) | Knowledge graph-based adaptive learning resource recommendation method and system | |
| Gutiérrez et al. | Using the Belbin method and models for predicting the academic performance of engineering students | |
| CN114429212A (en) | Intelligent learning knowledge ability tracking method, electronic device and storage medium | |
| CN115510286A (en) | Multi-relation cognitive diagnosis method based on graph convolution network | |
| CN111552796A (en) | Volume assembling method, electronic device and computer readable medium | |
| Mujtaba et al. | Artificial intelligence in computerized adaptive testing | |
| Fotheringham et al. | Multiscale Geographically Weighted Regression: Theory and Practice | |
| Hagedoorn et al. | Massive open online courses temporal profiling for dropout prediction | |
| Sarker et al. | Exploring student predictive model that relies on institutional databases and open data instead of traditional questionnaires | |
| CN113283488A (en) | Learning behavior-based cognitive diagnosis method and system | |
| CN111401525A (en) | Adaptive learning system and method based on deep learning | |
| Zhuang et al. | A Robust Computerized Adaptive Testing Approach in Educational Question Retrieval | |
| Kim et al. | Behavioral Testing of Deep Neural Network Knowledge Tracing Models. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210319 |
|
| RJ01 | Rejection of invention patent application after publication |