Title:
METHOD FOR IDENTIFYING COMMENT NATURE OF TEXT, AND METHOD AND APPARATUS FOR TRAINING MACHINE LEARNING MODEL
Document Type and Number:
WIPO Patent Application WO/2021/190275
Kind Code:
A1
Abstract:
Disclosed are a method for identifying the comment nature of text, and a method and apparatus for training a machine learning model. The identifying method comprises: obtaining constructed text, and obtaining a text vector matrix matching the constructed text; and inputting the text vector matrix into a machine learning model obtained by means of pre-training, and obtaining respective comment natures, for various added comment objects, of text to be identified which is output by the machine learning model obtained by means of pre-training. Since comment objects are added to the text to be identified, and the comment natures of both said text and the comment objects are jointly identified, adding the comment objects to the process of identification of the comment natures can effectively improve the identification accuracy of the comment nature of the text.
Inventors:
DAI ZEHUI (CN)
Application Number:
PCT/CN2021/079191
Publication Date:
September 30, 2021
Filing Date:
March 05, 2021
Export Citation:
Assignee:
BEIJING GRIDSUM TECHNOLOGY CO (CN)
International Classes:
G06F16/35
Foreign References:
| CN107862343A | 2018-03-30 | |||
| CN107291696A | 2017-10-24 | |||
| CN110517121A | 2019-11-29 | |||
| CN108932637A | 2018-12-04 | |||
| CN107491531A | 2017-12-19 |
Attorney, Agent or Firm:
UNITALEN ATTORNEYS AT LAW (CN)
Download PDF: