复杂中文文本的实体关系抽取研究

Entity Relation Extraction for Complex Chinese Text

实体关系抽取是信息抽取研究领域中的重要研究课题之一。针对已有方法在处理复杂文本上的不足,提出了复杂中文文本的实体关系抽取方法。结合中文文本的语法特征,提出了7条抽取关系特征序列的启发式规则,并采用语义序列核和KNN机器学习算法结合的方法来分类和标注关系的类型。通过对ACE评测定义下的两个子类的实体关系抽取,关系抽取的平均F值达到了76%,明显高于传统的基于特征向量和最短依存路径核的方法。

Entity Relation Extraction is one of the important research fields in Information Extraction. Aiming at the problem of inefficiency of existing approaches dealing with entity relation extraction, this paper presented a novel approach. This new approach proposes seven heuristic rules to extract relation feature sequence through combining with grammar feature of Chinese text, and applies the semantic sequence kernel function with KNN learning algorithm to fulfill the entity relation extraction task. Experiments are carried out on two kinds of relation types defined in the ACE guidelines, results show that the new approach achieves an average F-score up to 76 %, significantly higher than the traditional feature-based approaches and traditional shortest path for dependency kernel approaches.