项目名称: 基于多重分形和文本数据流技术的网络金融信息动态挖掘研究
项目编号: No.71301041
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 管理科学
项目作者: 倪丽萍
作者单位: 合肥工业大学
项目金额: 20.5万元
中文摘要: 随着互联网技术的发展,网络金融信息成为金融信息的主要来源,对市场有着重要影响。网络金融信息的准确挖掘有利于获取更多有用知识,提高市场预测准确率。本课题拟针对网络金融信息高维、动态、非结构化特点,采用多重分形和文本数据流技术对其进行动态挖掘,并在此基础上构建网络金融信息与市场波动的关联模型。 具体研究内容为:(1)研究基于多重分形和文本数据流技术的网络金融信息动态挖掘架构。(2)研究基于多重分形的特征选择算法,解决网络金融信息的高维问题。(3)研究文本数据流分类算法,对网络金融信息情感态度进行分类,并构建信息情感态度与市场波动的关联模型。(4)研究文本数据流聚类算法,检测网络金融信息中包含的特定主题事件和突发事件,并分析事件对市场波动的影响作用。 本研究可望部分解决网络金融信息动态挖掘问题,提高分析的准确率和时效性,更好地从网络金融信息中提取有价值的内容,提高市场预测准确率。
中文关键词: 网络金融信息;文本数据流;多重分形属性选择;案例推理;关联分析模型
英文摘要: With the rapid development of Internet,Online financial information becomes the main source of financial information and has a major impact on financial market.The accurate mining result of online financial information will not only be good for people to acquire more useful knowledge but can improve the prediction accuracy.The multifractal and text data stream technology will be used in this project to solve high-dimensional,dynamic and unstructured online financial information. On that basis the association model between online financial information and financial market volatility will be constructed. The main contents of this project include the following aspects: (1)The online financial information dynamic architecture based on multifractal and text data stream technology is proposed. (2) The multifractal feature extraction method is researched to solve the high-dimensional problem of online financial information. (3) Text data stream classification method is studied to get the sentiment classification of online financial information and the association model between information sentiment and market volatility is discovered. (4) Text data stream clustering method is discussed to detect specific event and emergency from online financial information and the effect of events on market volatility is analyze
英文关键词: Online Financial Information;Text Data Stream;Multi-fractal Feature Selection;Case Based Reasoning;Correlation Analysis Model