音乐领域中文实体关系抽取研究

实体关系抽取的任务是从文本中抽取出两个或者多个实体之间预先定义

好的语义关系。本文将实体关系抽取定义为一个分类问题，主要研究内容是

中文音乐领域的实体关系抽取。针对这一问题，本文首先构建了中文音乐实

体关系语料库，然后分别采用了基于序列模式挖掘的无指导的方法和基于特

征提取的有指导的方法来解决这一问题。 -Dissertation for the Master Degree in Engineering urgently needed to deal with, efficient Web classification method is to extract the required letter from the online Sea Scout information, the interest rate on the key technology, feature selection is an important foundation for text classification mining, the tomb of generalized information theory as the theoretical basis, the tomb of selection methods at the secondary entropy of mutual trust Chi characteristics of each feature, independent assessment, then the feature set to analyze the relationship between the characteristics and categories, from high-dimensional feature space selected remove the lining of the characteristics of effective text classification, reducing the dimension of the text feature space, improve the nature of the text classification