知识表示学习综述

A Review of Knowledge Representation Learning

近期,以深度学习为代表的表示学习不仅为机器学习算法提供更好的性能,同时也为知识表示提供了新思路。引入深度学习的知识表示学习实现了实体和关系的分布式表示,不仅提高了计算效率,还解决了传统三元组表示中会出现的数据稀疏问题。本文归纳和总结了知识表示学习的最新研究进展和应用方式,并且指出了知识表示学习面临的主要挑战以及未来可能实现的解决方案。

Recently,representation learning represented by deep learning not only provides better performance for machine learning algorithms,but also provides new ideas for knowledge representation.The introduction of deep learning knowledge indicates that learning achieves a distributed representation of entities and relationships,which not only improves computational efficiency,but also solves the problem of data sparsity that occurs in traditional triples representations.This paper summarizes the latest research progress and application methods of knowledge representation learning,and points out the main challenges of knowledge representation learning and possible solutions in the future.