基于迁移学习的实体关系抽取技术综述  

Review of transfer learning technology for entity and relation extraction

作　　者：郎春雨 侯霞[1] LANG Chunyu;HOU Xia(Computer School,Beijing Information Science&Technology University,Beijing 100101,China)

机构地区：[1]北京信息科技大学计算机学院,北京100101

出　　处：《北京信息科技大学学报（自然科学版）》2022年第1期65-70,共6页Journal of Beijing Information Science and Technology University

基　　金：国家自然科学基金资助项目(61672105)。

摘　　要：实体关系抽取任务的一大挑战是缺乏有效的训练语料,迁移学习可在一定程度上缓解其语料不足的问题。概述了迁移学习4种基本方法的原理及适用场景,分析总结了迁移学习在实体和关系抽取两方面的研究进展,最后总结展望了迁移学习技术在实体关系抽取领域的发展趋势。A major challenge of entity and relation extraction tasks is the lack of effective training corpus. Transfer learning can alleviate the problem of insufficient corpus to a certain extent. The principles and applicable scenarios of the four basic methods of transfer learning were outlined, the current situation and research progress of transfer learning in entity and relation extraction were analyzed and summarized, and finally the development trend of transfer learning technology was summarized and prospected in the field of entity and relation extraction.

关 键 词：迁移学习 实体抽取 关系抽取 机器学习 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]