Transitivity of Transformation Matrices to Bridge Word Vector Spaces Over Time
Transitivity of Transformation Matrices to Bridge Word Vector Spaces Over Time
复制标题
变换矩阵随时间桥接词向量空间的传递性
DOI:
10.1109/bigcomp.2019.8679169
复制
发表时间:
2019
期刊:
影响因子:
--
通讯作者:
Hiroaki Ohshima
中科院分区:
文献类型:
--
作者:
Katsurou Takahashi;Hiroaki Ohshima
We proposed a search method with a query by example in a known domain for information in an unknown domain. It seems natural the relation that “Pikachu in 2000s” which is similar to “Dumbo in 1980s.” Pikachu and Dumbo are the popular animation characters in these decades. For example, given three entities, such as Pikachu, 2000s and 1980s, we search Dumbo by the relation “Pikachu in 2000s as Dumbo in 1980s.” The proposed method is based on transformation matrices to bridge word vector spaces. Especially, we focus on transitive relations such as the relation that “Dumbo in 1980s” is similar to “Garfield in 1990s” and to “Pikachu in 2000s” to solve the problem. Experimental results show that the proposed method has some advantage when there are a little common words between source and target terms.
登录
查看更多内容
DOI:
10.3115/v1/p15-1063
发表时间:
2015-07
期刊:
--
影响因子:
--
作者:
Yating Zhang;A. Jatowt;S. Bhowmick;Katsumi Tanaka
通讯作者:
Yating Zhang;A. Jatowt;S. Bhowmick;Katsumi Tanaka
DOI:
--
发表时间:
2012
期刊:
Annual Meeting of the Association for Computational Linguistics
影响因子:
--
作者:
Rada Mihalcea;Vivi Nastase
通讯作者:
Vivi Nastase
DOI:
--
发表时间:
2003
期刊:
影响因子:
--
作者:
M. Gordon
通讯作者:
M. Gordon
影响因子:
2.4
作者:
J. Aitchison
通讯作者:
J. Aitchison
影响因子:
2.1
作者:
W. Francis;G. Hughes
通讯作者:
G. Hughes