Unlock instant, AI-driven research and patent intelligence for your innovation.

Knowledge graph representation learning framework optimization method based on key embedding

A technology of knowledge graph and optimization method, applied in the fields of unstructured text data retrieval, climate sustainability, text database indexing, etc., can solve the problems of insufficient computing power of learning framework, high computational complexity, large scale of model parameters, etc. To achieve the effect of reducing communication overhead, improving communication efficiency and ensuring accuracy

Pending Publication Date: 2022-06-07
TIANJIN UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the increasing scale of knowledge graphs poses new challenges to the performance of existing representation learning models.
At present, most of the model parameters are large in scale and computationally complex, and the existing knowledge graph indicates that the learning framework has insufficient computing power

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Knowledge graph representation learning framework optimization method based on key embedding
  • Knowledge graph representation learning framework optimization method based on key embedding
  • Knowledge graph representation learning framework optimization method based on key embedding

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0043] figure 1 The overall architecture diagram of the key embedding-based knowledge graph representation learning framework optimization method proposed by the present invention is shown. Load balancing is achieved by launching multiple server processes on each machine that access locally stored entity and relation embeddings through shared memory. Before the training starts, the knowledge graph training data is divided into several shards and sent to each machine. The parameter server starts and initializes the model parameters. At the same time, the model parameter shards are stored in each machine, and the trainer starts the process to execute in parallel. training process. With the key embedding cache table, each worker node can avoid a lot of remote communication for key embeddings.

[0044] In each iteration, the key embedding-based training process includes:

[0045] 1) Sample a set of positive samples from the local subgraph, and randomly replace the entities in i...

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

No PUM Login to View More

Abstract

The invention discloses a knowledge graph representation learning framework optimization method based on key embedding, which comprises the following steps that: a key embedding cache table is initialized, the key embedding cache table is empty after initialization, the key embedding cache table comprises a first column and a second column, the first column identifies the ID (Identity) of an entity or a relationship in a sub-graph, and the second column is the embedding of the entity or the relationship; dynamically constructing a key embedded cache table in a training iteration process by adopting preloading and filtering; adopting a constant partial delay or dynamic partial delay mode, and utilizing a key embedding cache table to update embedding of entities and relationships in the working node storage sub-graph; and pushing the updated embedding back to the parameter server. The invention designs a representation learning framework optimization method, and dynamically stores and updates high-frequency hit entities and relationships as key embedding in a knowledge graph representation learning training process. And at the cost of increasing a small amount of calculation time, the communication cost for obtaining key embedding is reduced, so that the overall training efficiency is improved.

Description

technical field [0001] The present invention relates to the field of knowledge graph, in particular, to the field of knowledge graph representation learning. Background technique [0002] Knowledge graph is an important cornerstone for realizing a new generation of cognitive artificial intelligence. With the rapid development of artificial intelligence, knowledge graph has been widely used in many fields, and the scale of knowledge graph data is increasing rapidly, with a scale of millions of nodes (10 6 ) and hundreds of millions of edges (10 8 ) knowledge graphs are ubiquitous in various fields, with a scale of tens of millions of nodes (10 7 ) and a billion edges (10 9 ) large-scale knowledge graphs have also begun to appear. [0003] In the Semantic Web community, Resource Description Framework (RDF) is a set of technical specifications for markup languages ​​proposed by the World Wide Web Consortium (W3C). It is widely used to describe Web resources due to its simpli...

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
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/36G06F16/31
CPCG06F16/367G06F16/328Y02D10/00
Inventor 王鑫柳鹏凯董思聪
Owner TIANJIN UNIV