论文部分内容阅读
本文定义了指人成分的概念,分析了指人成分的构成和分布特征,并面向大型叙事作品,提出了一种基于邻字熵统计和规则发现相结合的指人成分识别方法。实验对小说《英雄出世》的生文本进行了多次抽样测试,取得了86.93%的正确率和91.83%的召回率。
This paper defines the concept of person composition, analyzes the composition and distribution characteristics of person composition, and proposes a method of finger identification based on the combination of orthographic entropy statistics and rule discovery. Experiments on the novel “hero” was born in the text of a number of sampling tests, achieved a correct rate of 86.93% and 91.83% of the recall rate.