Knowledge map embedding method based on attention mechanism
A knowledge graph and attention technology, applied in the field of knowledge graph embedding, can solve the problem of not only focusing on several different attributes without different relationships, and achieve the effect of improving accuracy
Inactive Publication Date: 2018-11-06
ZHEJIANG UNIV
View PDF0 Cites 22 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] Although these works also mention that different relations may only focus on several different attributes of entities, and employ relation-related entity embeddings to make different relations compute transformation results in different spaces, they do not actually do so. Different relationships only focus on a few different attributes of entities
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
[0039] Specific embodiments of the present invention will be described below in conjunction with the accompanying drawings, so that those skilled in the art can better understand the present invention.
[0040] The validity of the method described in the present invention is verified by comparative experiments on WN18 and FB15k. These two publicly available knowledge graph datasets consist of information retrieved from WordNet and Freebase corpora, respectively. All these datasets consist of training set, validation set and test set, which are well organized. Table 1 lists the data statistics of the dataset. Each training entry is a relation triplet (h, r, t), denoting that h and t have relation r.
[0041] Table 1
[0042] data set
relationship number
Entity number
number of training items
Number of Validated Entries
Number of test entries
WN18
18
40493
141442
5000
5000
FB15k
1345
14951
483142
50000
...
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 present invention discloses a knowledge map embedding method based on an attention mechanism. The method comprises the following steps: (1) inputting a knowledge map data set that needs to be complemented, and initializing the knowledge map data set based on the attention mechanism; (2) based on the attention mechanism, updating the embedded representation to obtain an embedded representationresult and an attention mechanism parameter; and (3) complementing the knowledge map data set according to the embedded representation result and the attention mechanism parameter. By using the methoddisclosed by the present invention, complex relationships such as one-to-many, many-to-one and many-to-many relationships, which cannot be processed well by using the previous algorithms can be better processed, the obtained entity vector can reflect the category information well, and during relationship determination, the attention can be focused on some of the dimensions related to the relationships, so that the accuracy of determination can be improved.
Description
technical field [0001] The invention relates to the field of knowledge graph embedding, in particular to a knowledge graph embedding method based on an attention mechanism. Background technique [0002] A knowledge graph is a directed graph composed of entities as nodes and relationships as edges. Typically, a knowledge graph encodes structured information of millions of entities and billions of relational facts. But this is not complete enough, and the knowledge map needs to be completed. The purpose of knowledge graph completion is to predict whether there is a relationship between entities without edge connections based on the information in the existing knowledge graph, that is, connection prediction. [0003] A class of knowledge graph completion methods is based on embedding, which encodes each object in the knowledge graph into a vector in a continuous space. Recently, such methods have shown powerful effects on knowledge graph completion. So this method is becomi...
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 Patent Type & Authority Applications(China)
IPC IPC(8): G06F17/30
Inventor 钱炜蔡登付聪祝宇何晓飞
Owner ZHEJIANG UNIV