从复杂网络这一崭新视角对汉语进行系统的探索,无论是对.PDFVIP

摘 要 摘 要 从复杂网络这一崭新的视角对汉语进行系统的探索，无论是对汉语本体研 究还是对中文信息处理，都具有方法论意义上的创新性,并且涉及复杂系统、语 言学、自然语言处理、机器学习、统计学等多学科的交叉，因此具有十分重要 的科学意义，已成为当前自然语言处理的研究前沿与热点之一。 本文将主要介绍以下几方面的研究成果：(1) 利用目前可能得到的汉语资 源，构造了覆盖词法、句法、语义不同层次的各种类型大规模汉语语言网络， 他们包括汉语词同现网络、依存句法网络、标签语义网络和查询词语义网络， 并对上述语言网络的性质进行分析与对比；(2) 基于汉语语言网络，提出基于二 部图的资源分配算法进行查询词推荐，取得了较以往算法更好的效果。 本研究对汉语语言网络全面、深入的考察与研究, 在一定程度上丰富和深化 对汉语的科学认识，得出的一系列结果或结论对汉语语言学、语言认知、中文 信息处理等均具有一定参考价值。 关键词：复杂网络 语言网络 汉语 自然语言处理 I 摘 要 Abstract A systematic investigation of Chinese, no matter on Chinese ontology or Chinese information processing, from a brand -new perspective of complex network, is a great innovation at methodology level. It has referred to various disciplines, including complex system, linguistics, natural language processing, machine learning and st atistics. Therefore, it is of great scientific significance and has become the hot frontier issue of natural language processing. The thesis will introduce the following research results mainly: (1) I construct different types of large-scale Chinese language networks, including word co -occurrence network, syntactic network and blog tag co -occurrence network and query semantic network. Afterwards I analyze and compare the statistical, properties of these language networks. (2) I present a resource allocation method based on bipartite language network of query logs to recommend related queries, and achieve great improvement and significant features than previous methods. These research and investigation on Chinese