原标题：A Novel Chinese Entity RelationshipExtraction Method Based on the Bidirectional Maximum Entropy Markov Model
作者：Chengyao Lv , Deng Pan, Yaxiong Li ,etal.
关键词：Deep learning, MEMM, Natural languageprocessing, Entity Relationship Extraction
中文摘要：
为了识别文本中实体之间的关系，实体关系提取技术为知识图谱、智能信息检索和语义分析提供了基础支持，促进了知识库的构建，提高了搜索和语义分析的效率。传统的关系提取方法，无论是早期提出的，还是基于传统机器学习和深度学习的，都侧重于使关系和实体在各自内部固定的关系中: 也就是关系和实体的提取是在获得映射（确定关系）之前一步一步进行的。为解决这一问题，本文提出了一种新的中文关系抽取方法。首先，将三元组视为一个实体关系链，可以识别出关系之前的实体，并预测其对应的关系和关系之后的实体。其次，使用了基于LSTM和最大熵马尔可夫的模型(Bi-MEMM)来进行实体和关系的联合提取。实验结果表明，该模型的精度达到79.2%，大大高于传统模型。
英文摘要：
To identify relationships among entities innatural language texts, extraction of entity relationships technically providesa fundamental support for knowledge graph, intelligent information retrieval,and semantic analysis, promotes the construction of knowledge bases, andimproves efficiency of searching and semantic analysis. Traditional methods ofrelationship extraction, either those proposed at the earlier times or thosebased on traditional machine learning and deep learning, have focused onkeeping relationships and entities in their own silos: extracting relationshipsand entities are conducted in steps before obtaining the mappings. To addressthis problem, a novel Chinese relationship extraction method is proposed inthis paper. Firstly, the triple is treated as an entity relation chain and canidentify the entity before the relationship and predict its correspondingrelationship and the entity after the relationship. Secondly, the JointExtraction of Entity Mentions and Relations model is based on the BidirectionalLong Short-Term Memory and Maximum Entropy Markov Model (Bi-MEMM). Experimentalresults indicate that the proposed model can achieve a precision of 79.2% whichis much higher than that of traditional models.
原文链接：
https://kns.cnki.net/KCMS/detail/detail.aspx?dbcode=IPFD&filename=LRCM201905004010
文献总结：
