Title:
USER PORTRAIT REPRESENTATION LEARNING SYSTEM AND METHOD BASED ON DEEP NEURAL NETWORK
Document Type and Number:
WIPO Patent Application WO/2018/000281
Kind Code:
A1
Abstract:
A user portrait representation learning system based on a deep neural network, comprising: an intention recognition module (102) used for recognizing a use function of a user according to a received statement; a feature vector extraction module (103) used for modeling the context relation of a text or the relationship between entities by means of deep learning, and then extracting feature information of the user by means of text information input by the user; and a user portrait learning module (104) used for continually updating a user portrait by means of iterative training of the feature information and supervisory information. By means of learning a user portrait in a deep learning mode, the features of the user portrait can be abstractly extracted, the feature representation is more concise and accurate, and a deep level of implicit information can be extracted.
Inventors:
QIU NAN (CN)
YANG XINYU (CN)
WANG HAOFEN (CN)
YANG XINYU (CN)
WANG HAOFEN (CN)
Application Number:
PCT/CN2016/087773
Publication Date:
January 04, 2018
Filing Date:
June 29, 2016
Export Citation:
Assignee:
SHENZHEN GOWILD ROBOTICS CO LTD (CN)
International Classes:
G06F17/30
Foreign References:
CN105068661A | 2015-11-18 | |||
CN105096170A | 2015-11-25 | |||
CN105183848A | 2015-12-23 | |||
CN104933049A | 2015-09-23 | |||
US20150278688A1 | 2015-10-01 |
Attorney, Agent or Firm:
SHENZHEN HYVISION INTELLECTUAL PROPERTY ATTORNEY (CN)
Download PDF: