智能手机电子取证中文本分析的研究

发布时间：2018-02-13 08:40

本文关键词： 短文本分类手机取证特征权重特征扩展　出处：《武汉邮电科学研究院》2014年硕士论文　论文类型：学位论文

【摘要】：随着移动电话的迅速普及和手机容量的增大,处理手机大批量短信数据的挑战也与日俱增。如何从手机中快速提取执法人员感兴趣的证据成为手机取证技术面临的问题之一,短文本自动分类技术可以一定程度上解决该难题。虽然传统的文本分类技术已经相对成熟,但短文本的格式不规范且文本信息稀疏,导致适用于长文本的分类方法不能直接应用于短文本领域。基于这样的背景,本文对手机取证中短文本分类方法进行相应的探索和对比。本文首先概述了手机取证的取证源、取证原则和取证过程,并以手机中短信信息为例分析手机取证系统提取电子证据的基本过程。接着重点针对手机取证系统中的短信自动分类子系统进行探讨。讨论了长文本和短文本存在的基本区别,分析传统分类方法中几种常使用的特征权值算法,并提出一种针对短文本特点的权重值改进算法。同时为了弥补短文本信息量稀少的劣势,提出引入维基百科数据库作为知识库对短文本的特征进行扩展的方法。本文最后详细介绍了短信信息分类子系统中的各个模块设计过程和实现过程,并对本文中提到的两种改进方法先后用六组试验进行分类效果对比。六组实验结果表明,基于传统分类方法上的两种改进方法均能一定程度上提高中文短文本分类的性能。同时也发现,若在建立知识库之初噪声没有被清理干净,会影响基于特征扩展的分类方法的分类效果。
[Abstract]:With the rapid popularization of mobile phones and the increase of mobile phone capacity, the challenge of dealing with mass SMS data is also increasing. How to quickly extract evidence of interest to law enforcement officials from mobile phones has become one of the problems faced by mobile phone forensics technology. Although the traditional text classification technology is relatively mature, the format of short text is not standardized and the text information is sparse. As a result, the classification method suitable for long text can not be directly applied to the field of short text. Based on this background, this paper explores and compares the classification methods of short text book in mobile phone forensics. Firstly, this paper summarizes the evidential source of mobile phone forensics. Principles and procedures for obtaining evidence, Taking short message information in mobile phone as an example, this paper analyzes the basic process of extracting electronic evidence from mobile phone forensics system. Then, it focuses on the automatic classification subsystem of short message in mobile phone forensics system, and discusses the basic differences between long text and short text. This paper analyzes several feature weight algorithms that are often used in traditional classification methods, and proposes an improved algorithm of weight value for the characteristics of short text books, in order to make up for the shortage of short text information. This paper proposes a method to extend the features of short text by using Wikipedia database as a knowledge base. Finally, the design process and implementation process of each module in the short message classification subsystem are introduced in detail. The two improved methods mentioned in this paper are compared with each other by six groups of experiments. The results of the six groups of experiments show that, The two improved methods based on the traditional classification method can improve the performance of the Chinese text classification to some extent. It is also found that the noise is not cleaned up at the beginning of building the knowledge base. It will affect the classification effect of the classification method based on feature expansion.
【学位授予单位】：武汉邮电科学研究院
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.53;TP391.1

【相似文献】