【作者】 周峰；

【导师】 王柏；

【作者基本信息】 北京邮电大学 ， 计算机科学与技术， 2009， 硕士

【摘要】 复杂网络的研究,为我们提供了一个开展复杂性研究的新视角、新方法,对各种复杂网络进行比较、研究与综合概括已成为当今的科研热点之一。随着互联网的发展,非结构化和半结构化信息量增加,基于这些信息进行复杂网络分析成为必然趋势,信息抽取技术扮演了越来越重要的角色。将信息抽取技术融合到复杂网络中,能够有效的抽取网络的节点和边信息,为复杂网络的构建与展示提供数据准备,这将大大扩展复杂网络的应用范围。信息抽取与复杂网络的融合必将是一个新的研究和应用热点问题。本文以中文文本作为研究数据,针对信息抽取的研究热点及其相关技术进行了深入的分析和研究,并结合实际构建复杂网络的需要,取得了如下研究成果:1.实现了点信息的自动抽取。本文深入研究了实体识别领域常用的一些统计模型,并应用这些模型实现了点信息的核心抽取算法,包括应用隐马尔科夫模型实现中文分词和词性标注一体化,对于词性标注后的文本,应用最大熵的马尔科夫模型进行实体的识别,都取得了较好的效果。2.实现了边信息的自动抽取。本文通过阅读大量的国内外文献,总结了目前常用的一些实体关系抽取方法,然后针对我们边信息抽取的特殊需要,提出了一种新的基于语义的实体关系抽更多还原

【Abstract】 Complex Network provide us a new perspective of complexity research, to compare, research and summarize a variety of Complex Networks have become one of the scientific research hotspots. With the development of Internet, the amount of unstructured and semi-structured information increase, Complex Network analysis based on the information is an inevitable trend; the technology of Information Extraction plays a more and more important role. Integrating with Information Extraction and Complex Network, we can extract the information of vertexes and edges that can provide basic data for the construction of Complex Network, and greatly expand the Complex Network applications. Integration of Information Extraction and Complex Network will be a new hot issue of research and application.In the thesis, based on the data of Chinese text, we carried out in-depth analysis and research on Information Extraction and related technologies, combined with the practical needs of the construction of Complex Network, achieved the following results:1. Automatic extraction of vertexes. We carry out in-depth study to some commonly used statistical models of entity recognition, and apply the models to the core algorithm of the vertexes’ extraction, including using HMM to implement the Chinese word segmentation and part-of-speech tagging, using MEMM to implement the entity recognition, all of these have achieved good results.2. Automatic extraction of edges. Through extensive reading of domestic and foreign literatures, we sum up some of the most commonly used entity relationship extraction method, then according to the special needs, a new method of entity relationship extraction is proposed, which is accurate, flexible, good practicability.3. Automatic construction of complex network. Based on the implementation of the vertex and edge extraction algorithms, a prototype system of automatic construction of complex network is set up, which has friendly interface, good capacity of natural language processing, flexible software architecture and vivid display of results.Finally, the research and implementation on the construction of Complex network are summarized, and the prospects and the future directions are discussed.更多还原