基于关系冗余度的少样本实体关系自适应抽取  

Adaptive Extraction of Entity Relationship with Few Samples Based on Relationship Redundancy

作　　者：高峰 龚珊珊[1,2,3] GAO Feng;GONG Shanshan(School of Computer Science and Technology,Wuhan University of Science and Technology,Wuhan 430065;Big Data Science and Engineering Research Institute,Wuhan University of Science and Technology,Wuhan 430065;Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System,Wuhan University of Science and Technology,Wuhan 430065)

机构地区：[1]武汉科技大学计算机科学与技术学院,武汉430065 [2]武汉科技大学大数据科学与工程研究院,武汉430065 [3]武汉科技大学智能信息处理与实时工业系统湖北省重点实验室,武汉430065

出　　处：《计算机与数字工程》2024年第8期2323-2328,共6页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:U1836118)资助。

摘　　要：医学文本中丰富的医学知识可为构建医学知识图谱提供数据支撑,但医学文本中存在部分知识文本数量较少,导致了知识分布不平衡、循证类知识样本少等现实问题,且现有实体关系抽取模型对关系冗余、实体重叠等问题并没有很好的解决方案。论文针对上述问题,提出了一种基于关系冗余度的少样本实体关系自适应抽取模型,该模型弥补了现有抽取模型过度依赖大量标注语料、无法解决实体重叠等不足。使用医学相关文本展开实验,结果表明该模型较现有抽取模型F1性能提高了4.9%。The rich medical knowledge in medical texts can provide data support for the construction of medical knowledge map,but there are a few knowledge texts in medical texts,which leads to practical problems such as unbalanced knowledge distribution and few evidence-based knowledge samples,and the existing entity relationship extraction model has no good solution to the problems of relationship redundancy and entity overlap.Aiming at the above problems,this paper proposes an adaptive entity relationship extraction model with few samples based on relationship redundancy,which makes up for the shortcomings of the existing extraction models,such as over reliance on a large number of labeled corpus and inability to solve entity overlap.The experimental results show that the performance of this model is improved by 4.9%compared with the existing extraction model F1.

关 键 词：医学领域 知识图谱 少样本 信息抽取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]