上海大学学报(自然科学版) ›› 2020, Vol. 26 ›› Issue (5): 747-755.doi: 10.12066/j.issn.1007-2861.2068

• 研究论文 • 上一篇    下一篇

基于图的联合特征实体链接方法

周金1, 朱永华2(), 张铁男2, 邢毅雪1, 张克1   

  1. 1.上海大学 上海电影学院, 上海 200072
    2.上海大学 计算机工程与科学学院, 上海 200444
  • 收稿日期:2018-07-02 出版日期:2020-10-30 发布日期:2020-11-06
  • 通讯作者: 朱永华 E-mail:yhzhu@staff.shu.edu.cn
  • 基金资助:
    上海市科委基金资助项目(14590500500)

A graph-based method for multi-feature entity linking

ZHOU Jin1, ZHU Yonghua2(), ZHANG Tienan2, XING Yixue1, ZHANG Ke1   

  1. 1. Shanghai Film Academy, Shanghai University, Shanghai 200072, China
    2. School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China
  • Received:2018-07-02 Online:2020-10-30 Published:2020-11-06
  • Contact: ZHU Yonghua E-mail:yhzhu@staff.shu.edu.cn

摘要:

实体链接是指将文本中的实体指称映射到知识库实体的过程, 这一过程在知识图谱、知识融合领域都是关键的步骤之一. 提出了一种基于图的联合特征实体链接方法, 首先对知识库和文本进行预处理, 然后识别文本中的命名实体指称, 随后联合主题、上下文、元数据等多特征的语义相似度, 在经扩充的图模型中利用重启随机游走和联合消歧选出指称的链接实体. 实验结果表明, 基于图的联合特征实体链接方法有效提高了实体链接效果.

关键词: 实体链接, 实体消岐, 语义相似度, 重启随机游走, 自然语言处理

Abstract:

Entity linking refers to the process of linking entity mentioned in text with knowledge base entity, which is one of the key steps in knowledge graph and knowledge fusion. This paper proposes a graph-based method for multi-feature entity linking. This method first preprocesses the knowledge base and the text, then identifies the named entity references in the text, and then combines the semantic similarity of multiple features such as topics, context, metadata, etc. In the expanded graph model, the probability of restarting random walk is used, and the target candidate entity is selected by joint disambiguation. The results of experiment show that the joint feature-based entity linking method based on graphs effectively improves the effectiveness of entity linking.

Key words: entity linking, entity disambiguation, semantic relatedness, random walk with restart, natural language processing

中图分类号: