Improved knowledge graph vector representation method based on Node2vec
A technology of knowledge graph and vector representation, which is applied in the direction of neural learning methods, relational databases, database models, etc., can solve the problem of unbalanced semantic model complexity and model accuracy, so as to reduce time complexity and space complexity, Prediction results are accurate and the effect of ensuring accuracy
Active Publication Date: 2020-01-17
JILIN UNIV
View PDF6 Cites 6 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0007] The purpose of the present invention is to solve the problems existing in the current knowledge map vector representation method: the representation vector with sufficient semantics cannot be obtained through training, and the complexity of the model and the accuracy of the model cannot be balanced, and an improved Node2vec-based The knowledge map vector representation method of
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0047] see Figure 1 to Figure 7 Shown:
[0048] In the WN18 data set, the whole process of operation:
[0049] Step 1. Processing the data set. WN18 includes 18 kinds of relationships and 40,942 entity nodes. In the original data, it is stored in the form of triples of node-relation-node. Now change the nodes in the data set into entity nodes, and use the relationship attributes in the data set as nodes to form a new data set. For example, there is a relationship of 3 between node 1 and node 2. For the original knowledge map, it will be expressed as There is a connecting line between node 1 and node 2, and the connecting line has attribute value 3. For the changed knowledge graph, it can be expressed as entity node 1 is connected to relational node 3, and relational node 3 is also connected to entity node 2. In the process of reconstructing the knowledge map dataset, both the relationships and nodes in the dataset should be turned into nodes, and the original nodes should ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses an improved knowledge graph vector representation method based on Node2vec, which comprises the following steps: 1, reconstructing a knowledge graph G'= (E, R, L); 2, setting atransition probability; 3, setting a walking path; 4, setting a training model according to the types of the nodes and the positions of the nodes; 5, training parameter optimization; and 6, outputting vectorized representation of the nodes through multiple times of training. The method has the beneficial effects that repeated training is carried out in combination with a reconstructed knowledge graph structure, semantic information of nodes can be obtained through sufficient training by carrying out cross walk on a walk sequence according to an entity and relationship, and the property significance of the nodes can also be obtained while a network topology structure is obtained. A homogeneous structure network is changed into a heterogeneous structure network, the method is more suitablefor the characteristics of a knowledge graph network structure, a walk sequence is closer to a natural language structure, and a prediction result is more accurate.
Description
technical field [0001] The present invention relates to a knowledge map vector representation method, in particular to an improved Node2vec-based knowledge map vector representation method. Background technique [0002] The knowledge graph was proposed by Google in 2012. Compared with the conventional graph structure, the nodes in the knowledge graph are given the meaning of entities, while the edges between entities represent the relationship between entities. Therefore, the knowledge map can connect different types of information together, providing the ability to analyze problems from the perspective of "relationship". [0003] In order to better deal with multi-relational data, knowledge representation learning technology is introduced, that is, representation learning for the relationship between entities. The relevant algorithm of KG Embedding also came into being. The general steps of KG Embedding: 1) represent entities and relationships in the graph; 2) define scor...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06F16/36G06F16/28G06N3/04G06N3/08
CPCG06F16/367G06F16/288G06N3/08G06N3/045
Inventor 董立岩马心陶王越群王浩
Owner JILIN UNIV