网络论坛文章数序列的自相似性建模与预测

发布时间：2018-03-02 08:12

本文关键词： 网络论坛自相似 FARIMA λ-ARMA 预测　出处：《复旦大学》2010年硕士论文　论文类型：学位论文

【摘要】： 随着Web2.0概念的提出,高速变化的互联网正得到人们越来越多的关注。网络论坛、博客等沟通交流形式已经成为互联网应用的热点。一方面,在网络论坛研究中,话题热点分析是一个主要的研究方向；另一方面,由于各种商业活动决策(如产品市场调查或广告投放)的需要,宏观上了解论坛或博客用户参与度的变化正在成为一个新的研究热点,对论坛中文章数随时间变化规律的研究最近也得到了人们的关注。因此,网络论坛特性的精确测量和刻画,对分析、理解和仿真网络论坛的动态变化,对指导网络论坛控制方案的设计都具有基础性的意义。本文以网络论坛中的文章数为研究对象,分析研究文章数序列的特性,提出文章数序列的描述模型,并在此基础上提出网络论坛文章数序列的预测方法。首先,对实际数据进行分析,验证和证明文章数序列的自相似性特性。然后,用短相关特性和长相关特性的时间序列分析方法对自相似网络论坛文章数进行建模分析,给出了利用马尔可夫模型、FARIMA模型进行建模的方法和步骤。为了避免FARIMA模型在结构辨识和参数估计上的复杂性,通过分析ARMA模型的参数估计方法,提出了基于传统ARMA模型的改进模型λ-ARMA。通过对实际数据的实验,验证了FARIMA模型和λ-ARMA模型在文章数序列建模方面的适用性。建立模型的一个重要用途是进行预测,本文结合实际数据给出了预测的算法和步骤,并通过实验验证了不同模型在不同置信区间内的适用性。本文的研究结果表明从文章数的角度分析网络论坛是必要的而且可行的。本文的工作将为下一代网络,尤其是网络论坛、博客等的构建和管理提供参考。
[Abstract]:With the introduction of the concept of Web2.0, people are paying more and more attention to the high-speed change of the Internet. Communication and communication forms such as Internet forums and blogs have become the focus of Internet applications. On the one hand, in the research of Internet forums, Hot topic analysis is a major research area; on the other hand, because of the needs of various business decisions (such as product market research or advertising), Macroscopically understanding the change of participation of forum or blog users is becoming a new research hotspot. The research on the regularity of the number of articles in the forum over time has been paid more and more attention recently. The accurate measurement and characterization of the characteristics of the network forum are of fundamental significance for analyzing, understanding and simulating the dynamic changes of the network forum and for guiding the design of the control scheme of the network forum. This paper takes the article number in the network forum as the research object, analyzes the characteristics of the article number sequence, puts forward the description model of the article number sequence, and then puts forward the prediction method of the article number sequence in the network forum. The paper analyzes the actual data, verifies and proves the self-similarity characteristic of the article number sequence, then uses the time series analysis method of the short correlation characteristic and the long correlation characteristic to model and analyze the number of articles in the self-similar network forum. In order to avoid the complexity of structure identification and parameter estimation of FARIMA model, the method of parameter estimation of ARMA model is analyzed. An improved model 位 -ARMA based on the traditional ARMA model is proposed. The applicability of the FARIMA model and 位 -ARMA model in the article number sequence modeling is verified by the experiment of the actual data. One of the important uses of modeling is to predict. In this paper, the algorithms and steps of prediction are given based on the actual data, and the applicability of different models in different confidence intervals is verified by experiments. The research results show that it is necessary and feasible to analyze the network forum from the point of view of the number of articles. The work of this paper will provide a reference for the construction and management of the next generation network, especially the network forum, blog and so on.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：G353

【参考文献】