摘要
[目的 /意义]作者身份识别是语言文体学的重要研究方向,利用文本特征的身份识别也是文本挖掘的重要任务。在开放和虚拟网络环境下海量信息的作者身份或发布者的识别难题和传统作者身份识别方法在处理效率和成本等方面存在的问题有待解决。[方法 /过程]将复杂网络理论引入该研究领域,在利用传统文体学特征识别作者身份方法的基础上结合文本词共现网络模型及其指标特征改进相关算法,使用文本文体学特征和文本网络模型度量指标构建作者风格特征集合,通过计算文本间风格相似度进行作者识别。[结果 /结论]基于复杂网络模型的作者身份识别方法可以有效的利用作者风格特征,提高识别的精度,与其他算法的对比试验表明其识别结果的准确性更高。
[ Purpose/significance ] Authorship analysis by means of textual features is an important task in text mining and linguistic studies. Tosolve the problem of low efficiency and high costs in authorship attribution using traditional method, complex networks theory has been employedto tackle this disputed problem. [Method/process ] In this paper, some measurable quantities of word co - occurrence complex network of text has been for used for authorship characterization. Based on stylistics and the network features, the approach is defined for authorship identification bycomputing theauthors＇stylefeatures similarity. [Result/condusion]The authorship attribution algorithm based on complex network can use authors style featureseffeetively. The experimental results show high accuracy rate in authorship attribution and prove the validity of this method.
出处
《图书情报工作》
CSSCI
北大核心
2015年第18期102-107,共6页
Library and Information Service
基金
国家自然科学基金"基于复杂网络的中文文本语义相似度研究"(项目编号:71373200)研究成果之一
关键词
作者识别
文本分类
复杂网络
特征提取
词共现
文体学
authorship attribution text classification complex network feature selection word co-occurrencelinguistic
