一种基于动态进化模型的事件探测和追踪算法

A News Event Detection and Tracking Algorithm Based on Dynamic Evolution Model

在大量分析网络新闻特点的基础上 ,借鉴Single Pass聚类思想 ,并结合新闻要素给出了一种基于动态进化模型的事件探测和追踪算法 该动态模型是基于新闻事件的生存特点提出的 ,包括 :基于时间距离的相似度计算模型、事件模板进化策略以及动态阈值设置思想 该算法可以自动对新闻资料进行组织生成新闻专题 ,进而为用户提供个性化服务

Via analyzing news data on the Internet,an algorithm is presented for news event detection and tracking based on a dynamic evolution model,which borrows the idea of single-pass clustering and combines the specialties of news. The dynamic model is given based on the living characteristics of news event,including similarity computing model based on time distance between news story and news event, event model evolution algorithm,and dynamic threshold idea. This algorithm can automatically organize news data into news special topics,and furthermore provide personalized service for users. Finally,experimental results are used to indicate the validity of the algorithm.