商务智能结课论文.docxVIP

题目: 文本挖掘方法研究 摘要如今互联网世界大数据日益发酵，各种数据一起奔涌而出，其中半结构化和非结构化数据所占比例明显增加，这是由于在信息传递的载体中文本发挥着越来越重要的作用。与之相对应，文本处理得到了众多学者和商业应用的关注。通过文本挖掘可以让用户在浩如烟海的文档中找到隐藏的信息，完成文本分类和聚类等各项功能，通过机器学习等自动化方法简化处理流程提高准确率和精确率。本文首先介绍了向量空间模型的文本表示和基于统计学及信息量的特征选择方法，然后介绍了基于本体的文本语义相似度度量方法，主要有基于信息量的概念相似度计算方法和基于本体结构的概念相似度计算方法，紧接着针对文本分类和聚类两个领域的机器学习方法进行分析，对其存在的问题提出了改进意见。最后介绍了这两类挖掘方法的评估标准，并提出了文本挖掘存在的问题及建议。关键词：文本挖掘，特征选择，本文相似度，文本分类，文本聚类AbstractNowadays big data is increasingly across the entire Internet. Various data surges up together of which the proportion of half structured and unstructured data has increased because texts play an more and more important role in the information transfer process. As a result, the text processing attracts the attention of many researchers and enterprises. With the help of text mining users can find voluminous information hidden in the document, finish the text classification and clustering and other functions. By the means of automation such as machine learning method can simplify process and improve the rate of accuracy and precision. First part of the paper introduces the vector space model of text representation and feature selection method based on statistics and information, secondly introduces the text semantic similarity measurement method based on ontology, which are mainly based on the concept of information similarity calculation method and based on the concept of ontology structure similarity calculation method. Then the paper analyze two methods of machine learning of text classification and clustering to find out the existing problems and put forward the improvement opinion. Finally introduces the criteria for the assessment of the two types of mining method, and puts forward the problems existing in the text mining and suggestion.Key words: text mining, feature selection, text similarity, text categorization, text clustering一、概述互联网的发展给人们提供了更快捷地了解世界，更高效地完成工作，更科学地进行组织的方式，用户通过文本、音频、视频等显性的表达形式完成与互联网的交互，尤其以文本为主。但由于文本的非结构化特征，有用信息需要进行深度挖掘与过滤