摘要
社会网络中的热点话题数据挖掘问题是目前的研究热点。针对现有方法应用到社会网络时挖掘结果不准确、不便于理解和时间复杂度高等不足,提出一种改进的挖掘算法。首先采用核密度估计法对话题进行分析,然后基于小世界理论和社会网络的链接特性对话题时间序列进行建模,最后提出基于学习的方法来对话题的走向和趋势进行预测,在不降低准确率的前提下,快速挖掘出下一时刻最有可能爆发的话题,达到准确预测话题传播趋势的目的。仿真实验结果表明,该方法是有效的,能够保证挖掘的整体效果,在挖掘准确率方面要优于已有的方法。
Mining the hot topic data in social network is the focus of research currently. Existing methods have the deficiencies when to be applied to the social networks such as inaccurate mining results, inconvenient in comprehension and high time complexity, etc. Aiming at these disadvantages, we propose an improved mining algorithm. First, the kernel density estimation method is used to analyse the topic; then the time sequence of the topic is modelled based on the small-world theory and the link characteristics of social network ; finally, the learning- based approach is presented to predict the direction and trends of the topic, and quickly mines the most possibly outbreak topics in the next time under the premise of not lowering the accuracy, achieves the goal of precisely predicting the propagation trend of the topic. Simulation experimental results show that our method is effective, it can ensure the overall effect of the mining, the accuracy of mining is better than the existing methods.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第6期24-28,共5页
Computer Applications and Software
基金
广西教育厅科研项目(201204LX350)
广西自然科学基金项目(2013GXNSFAA019078)
广西高等教育教学改革工程项目(2013JGA213)
玉林师范学院校级重点项目(2011YJZD16)
关键词
社会网络
热点话题挖掘
链接特性
时间序列
学习
准确率
Social network Hot topic mining Link characteristics Time sequence Learning Accuracy
作者简介
肖志军，讲师，主研领域：数据挖掘理论，智能决策。
