基于TDMA数据链的文本分类系统研究与应用

发布时间：2019-02-13 03:04

【摘要】：无线设备由于其操作简单、易于携带、交换信息方便等特点在现代通信中有十分重要的地位,被广泛用于各个领域,是军方不可或缺的通信手段。所以,无论在军事战争还是灾难救援中,无线设备是一种常用的指挥和联络工具。为了更快速、可靠的在无线设备之间传输数据,我们构建了一种基于TDMA(Time Division Multiple Access,时分多址)数据链的无线网络系统。本文的研究内容主要包括该系统中业务平台和无线平台之间的协议设计和实现以及业务平台上中文文本分类系统的研究和实现,具体如下:1)在基于TDMA数据链的无线网络系统中,每种类型的终端均包含无线平台和业务平台,无线平台完成各终端信息的传输,业务平台负责处理文件、定位信息、音频和视频等业务的管理和传输,为了保证各个业务可靠的传输,我们设计了 SWIP(Service Wireless Interface Protocol)协议。2)为了对传输内容进行过滤和分类处理,在业务平台上需要安装中文文本分类系统。首先,介绍了中文文本分类系统,并详细阐述了系统中各个模块,如文本预处理、特征降维、文本表示、分类算法等。并提出了新的中文文本分类系统,在新的系统中采用LDA(Latent Dirichlet Allocation)主题模型来表示文本,采用支持向量机(Support Vector Machines,SVM)和 K-最邻近算法(K-nearest neighbor classification)的结合算法KSVM进行分类。该系统使用JAVA编程语言在Windows系统下实现。3)为了分析提出的中文文本分类系统的性能,需要通过实验计算评价标准例如准确率、召回率、F1-measure等。本文提出了两种对比实验:基于LDA主题模型不同分类算法的分类效果分析;基于不同文本模型的KSVM分类效果分析。通过分析实验数据可知:基于LDA主题模型和KSVM分类算法的中文文本分类系统可以获得更好的分类效果。
[Abstract]:Wireless equipment is widely used in various fields because of its simple operation, easy to carry, convenient exchange of information and so on. It is an indispensable means of communication for the military. Therefore, wireless equipment is a common command and liaison tool in both military war and disaster relief. In order to transmit data between wireless devices more quickly and reliably, we construct a wireless network system based on TDMA (Time Division Multiple Access, (time Division multiple access) data link. The research content of this paper mainly includes the protocol design and implementation between the service platform and wireless platform, and the research and implementation of the Chinese text classification system on the service platform. The details are as follows: 1) in the wireless network system based on TDMA data link, each type of terminal includes wireless platform and service platform. The wireless platform completes the transmission of terminal information, and the business platform is responsible for processing files and locating information. The management and transmission of audio and video services, in order to ensure the reliable transmission of each service, we designed SWIP (Service Wireless Interface Protocol) protocol. 2) in order to filter and classify the transmission content, The Chinese text classification system needs to be installed on the business platform. Firstly, the Chinese text classification system is introduced, and various modules in the system, such as text preprocessing, feature dimensionality reduction, text representation, classification algorithm and so on, are described in detail. A new Chinese text classification system is proposed, in which the LDA (Latent Dirichlet Allocation) topic model is used to represent the text and the support vector machine (Support Vector Machines, is used to represent the text. SVM) and K-nearest neighbor algorithm (K-nearest neighbor classification) combined with the algorithm KSVM to classify. In order to analyze the performance of the proposed Chinese text classification system, the evaluation criteria such as accuracy, recall rate, F1-measure and so on need to be calculated experimentally. This paper presents two comparative experiments: the analysis of classification effect based on different classification algorithms based on LDA subject model and the analysis of KSVM classification effect based on different text models. By analyzing the experimental data, we can see that the Chinese text classification system based on LDA topic model and KSVM classification algorithm can achieve better classification effect.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN919.2;TP391.1

【参考文献】