关系信息抽取方法、设备及计算机可读存储介质
By using an end-to-end relational information extraction model, super-relational facts are directly extracted from text, solving the error propagation problem in the two-stage method, achieving more efficient super-relational fact extraction, and improving the model's recall and performance.
CN115587192BActive Publication Date: 2026-07-17ALIBABA (CHINA) CO LTD +1
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ALIBABA (CHINA) CO LTD
- Filing Date
- 2022-10-18
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing two-stage relation extraction methods are prone to error propagation, leading to a reduction in the number of super-relation facts extracted.
Method used
An end-to-end relational information extraction model is adopted. Sentence texts in the text corpus are input into the relational information extraction model, and the super-relational facts contained in the sentence text are extracted and output, including relational triples and limiting information. The extraction of super-relational facts is performed in an end-to-end manner.
Benefits of technology
This avoids the error propagation problem of the two-stage method, and can extract more effective hyperrelation facts, thus improving the model's recall and performance.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN115587192B_ABST
Abstract
本申请提供一种关系信息抽取方法、设备及计算机可读存储介质。该方法包括：获取当前应用领域的文本语料，文本语料包含至少一个句子文本；将句子文本输入关系信息抽取模型，通过关系信息抽取模型抽取句子文本包含的超关系事实，超关系事实包括关系三元组和关系三元组的限定信息。本申请的方法，利用端到端的关系信息抽取模型，将文本语料中的句子文本输入关系信息抽取模型，该关系信息抽取模型抽取并输出句子文本包含的超关系事实，实现了以端到端的方式执行超关系事实的抽取，避免了两阶段方法的错误传播问题，能够抽取到更多有效的超关系事实，提高了模型的召回率和性能。
Need to check novelty before this filing date? Find Prior Art