基于Single-Pass的网络话题在线聚类方法研究

Study on Web Topic Online Clustering Approach Based on Single -Pass Algorithm

基于Single-Pass算法思想,研究网络话题的在线聚类方法,以期及时捕捉网络信息的动态变化在分析该方法聚类流程的基础上,重点研究网络动态信息流的文本特征抽取和权重计算方法,以及话题类表示和更新等关键问题,设计实验对比分析不同的标题中特征加权系数、特征权重计算和标准化方法以及话题类向量维度对话题聚类质量和时间效率的影响。

In order to get dynamics of Web information timely, an online Web topic clustering approach based on Single - Pass algorithm is researched. The clustering process of this approach is analyzed firstly, and the key problems including extracting and weight calculating of features as well as representation and modification of topic cluster are deliberated. Experiment is designed to compare the effects of different weight factor of features in title, weight calculating and normalizing methods of features and the vector dimension of topic cluster on cluster quality and time efficiency.