Title:
TEXT INFORMATION EXTRACTION METHOD AND APPARATUS
Document Type and Number:
WIPO Patent Application WO/2021/027283
Kind Code:
A1
Abstract:
A text information extraction method and apparatus, wherein the method may comprise: obtaining a text feature vector corresponding to text to be processed (S100); performing dimension transformation on the text feature vector according to the number of preset information features to obtain an information candidate vector matrix comprising said number of dimensions (S200); traversing the information candidate vector matrix to determine position information in the information candidate vector matrix of character feature vectors that match the preset information features (S300); and according to the determined position information, extracting a word from the text to be processed, and according to the matched preset information features, obtaining information requiring extraction (S400). In the present solution, an information candidate vector matrix can be constructed, and the information candidate vector matrix is traversed so as to determine position information in the information candidate vector matrix of a character feature vector that matches a feature of information requiring extraction. The present solution achieves the automatic and efficient extraction of information, and a large amount of information meeting a requirement can be conveniently acquired from text to be processed.
Inventors:
DAI WEI (CN)
Application Number:
PCT/CN2020/079695
Publication Date:
February 18, 2021
Filing Date:
March 17, 2020
Export Citation:
Assignee:
BEIJING GRIDSUM TECHNOLOGY CO (CN)
International Classes:
G06F16/36; G06N20/00
Foreign References:
| CN108536678A | 2018-09-14 | |||
| CN109446519A | 2019-03-08 | |||
| US20160041987A1 | 2016-02-11 | |||
| CN107545262A | 2018-01-05 |
Attorney, Agent or Firm:
UNITALEN ATTORNEYS AT LAW (CN)
Download PDF:
Previous Patent: MINIATURE SMART LOCK WITH ANTI-PRISING FUNCTION
Next Patent: METHOD AND DEVICE FOR DETERMINING COMMENT NATURE OF TEXT
Next Patent: METHOD AND DEVICE FOR DETERMINING COMMENT NATURE OF TEXT