【摘要】 为解决地质领域实体关系类型复杂且缺乏大量标注语料的问题,建立基于句法结构的开放式实体关系联合抽取模型CSSEM (Chinese syntactic structure extraction model)。给出一种基于模式的地质领域实体识别方法,解决地质领域缺乏实体标注语料的问题;基于少量标注语料自动学习关系抽取模式,使用基于句法结构的抽取模式从非结构化文本中开放式地抽取三元组。基于AUC (area under curve)评价标准,CSSEM与其它方法在地质领域和通用领域的数据集上进行对比分析。分析结果表明,该模型在上述两个领域取得了更好的准确率和召回率。更多还原

【Abstract】 To solve the problems of complex types of entities and relations as well as the lack of a large number of annotated corpora in the geological field,an open model CSSEM(Chinese syntactic structure extraction model)of entity relation extraction was established,in which relation was extracted based on syntactic structure.A pattern-based method was proposed for identifying entities in the geological field,which solved the problem of lack of entity annotation corpus.Relation extraction patterns were automatically learned based on a small amount of annotated corpora,and syntactic structure-based patterns were used to openly extract triples from unstructured text.Based on the AUC(area under curve)evaluation standard,CSSEM and other methods were compared and analyzed in the geological and general dataset.The results show that the model achieves better precision and recall in the field of geology and general.更多还原