Vocabulary extraction method and apparatus, electronic device, and storage medium

By introducing entity information constraints during the encoding and decoding process through the Encoder-Decoder model, the problem of inaccurate entity and attribute extraction in complex sentences in existing technologies is solved, and more accurate word extraction results are achieved.

CN115481624BActive Publication Date: 2026-07-03CHINA MOBILE INFORMATION TECHNOLOGY CO LTD +1
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CHINA MOBILE INFORMATION TECHNOLOGY CO LTD
Filing Date
2021-06-16
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing supervised learning-based stepwise and joint extraction methods are ineffective in entity and attribute word extraction, especially in complex sentences where entity and attribute relationships are difficult to label, and the labeling system is complex and time-consuming.

Method used

The Encoder-Decoder model is adopted. The semantic vector of the sentence is obtained at the encoding end, and the attribute vocabulary is jointly extracted by the decoding end in combination with the entity vocabulary information. Entity constraints are directly introduced to characterize the relationship between entities and attributes.

Benefits of technology

It improves the accuracy of entity and attribute extraction in complex sentences, captures the relationships between multiple entity and attribute pairs, and achieves more accurate word extraction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115481624B_ABST
    Figure CN115481624B_ABST
Patent Text Reader

Abstract

The application provides a vocabulary extraction method and device, electronic equipment and storage medium. The method comprises: obtaining a first semantic vector of each minimum semantic unit in a target sentence through an encoding end in a vocabulary extraction model; determining an entity vocabulary in the target sentence based on the first semantic vector; obtaining a second semantic vector of each minimum semantic unit in the target sentence through a decoding end in the vocabulary extraction model based on the entity vocabulary in the target sentence and the first semantic vector; and determining an attribute vocabulary corresponding to the entity vocabulary in the target sentence based on the first semantic vector and the second semantic vector. The vocabulary extraction method and device, electronic equipment and storage medium provided by the application can more accurately capture the entity and attribute pair in a complex sentence by determining the entity vocabulary in the target sentence, jointly extracting the entity and attribute, directly introducing the entity constraint for the extraction of the attribute, and directly introducing different entity information to depict the relationship for different attributes.
Need to check novelty before this filing date? Find Prior Art