Title:
WORD CODING DEVICE, ANALYSIS DEVICE, LANGUAGE MODEL LEARNING DEVICE, METHOD, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2020/040255
Kind Code:
A1
Abstract:
According to the present invention, a vector representing a word is generated in consideration of importance of the character string of the word. A character N-gram vector acquisition unit acquires a vector representing each of character N-grams included in the word. A weight calculation unit 40 calculates weights of the vectors representing the respective character N-grams on the basis of vectors representing the respective character N-grams included in the word. A character N-gram vector synthesis unit 42 generates a synthesized vector obtained by synthesizing the vectors representing the respective character N-grams on the basis of the vectors representing the respective character N-grams included in the word and the weights of the vectors representing the respective character N-grams.
Inventors:
TAKASE SHO (JP)
NAGATA MASAAKI (JP)
NAGATA MASAAKI (JP)
Application Number:
PCT/JP2019/032854
Publication Date:
February 27, 2020
Filing Date:
August 22, 2019
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G06F17/27
Domestic Patent References:
| WO2018081089A1 | 2018-05-03 |
Other References:
SHINDO, HIROYUKI ET AL.: "Analysis of multiword expression using convolutional neural networks", IPSJ TECHNICAL REPORT: NATURAL LANGUAGE PROCESSING (NL) 2015- NL - 223, 20 September 2015 (2015-09-20), pages 1 - 7
Attorney, Agent or Firm:
TAIYO, NAKAJIMA & KATO (JP)
Download PDF: