基于潜层关系增强的实体和关系联合抽取
Joint extraction of entities and relationships based on subtext relationship enhancement
王鹏 1刘小明 1杨关 1刘杰 2刘阳3
作者信息
- 1. 中原工学院计算机学院,河南郑州 450007;中原工学院河南省网络舆情监测与智能分析重点实验室,河南郑州 450007
- 2. 北方工业大学信息学院,北京 100144;国家语委中国语言智能研究中心,北京 102206
- 3. 西安电子科技大学通信工程学院,陕西西安 710071
- 折叠
摘要
为充分发掘文本序列中潜层语义关系信息,提出一种实体和关系联合抽取的潜层关系增强模型SREM(text sub-text relationship enhancement model).在潜层关系表示层利用结构化对齐的方式,获取并保持文本序列中的语义信息结构.在融合注意力机制的关系网络层中对数据进行建模,提高模型对文本词汇间关系信息的捕获能力.结合注意力机制获取细粒度语义信息,对上下文信息进行选择过滤.实验结果表明,在数据集NYT和WebNLG上取得的F1值分别为92.40％和92.52％,验证了模型的有效性.
Abstract
To fully explore the latent text semantic relations in text sequences,a latent relationship enhanced joint extraction model SREM(text subtext relationship enhancement model)based on the relationship attention mechanism was proposed.The structured alignment was used at subtext relationship presentation layer to maintain semantic information structure in the text se-quence.In the relationship network layer where the attention mechanism was integrated,the relationship network was used to model the data to improve the model's ability to capture the relationship information between text and vocabulary.The relation-ship attention mechanism was combined to obtain fine-grained semantic information,and the context information was selected and filtered to reduce the impact of useless information.Experimental results of the model on two public datasets NYT and WebN-LG,achieves F1 values of 92.40％and 92.52％,which verify that the model is effective.
关键词
联合抽取/语义关系/结构化知识/潜层表示/注意力机制/关系网路/信息过滤Key words
joint extraction/semantic relation/structured knowledge/submerged representation/attention/relationship net-work/information filtering引用本文复制引用
基金项目
国家科技创新重大基金(2030)(2020AAA0109700)
国家自然科学基金(62076167)
东北师范大学应用统计教育部重点实验室项目(135131007)
国家自然科学基金青年基金(61906141)
出版年
2024
NETL
NSTL
万方数据