基于改进权重计算的话题跟踪

话题跟踪(Topic Tracking)任务是话题识别与跟踪(Topic Detection and Tracking,简称TDT)中的一个子任务,它的目的是监控新闻报道流识别出与预先给定的几个新闻报道所表述的话题相关的后继报道。特征项权重的计算方法是话题跟踪中的一个重要问题,计算方法的选择关系到话题跟踪的效果。提供了一种改进的权重计算方法,该方法的主要思想是:在计算特征项的权重时考虑了特征项的位置信息,将特征项的位置信息作为加权来计算特征项的权重。实验结果表明该方法有效,并提高了跟踪系统的性能。

Topic Tracking Based on an Improved Term Weighting Method

Topic tracking task has grown out of the Topic Detection and Tracking initiative sponsored by DARPA.Its aim is to monitor a stream of news stories for detecting additional stories on the same topic.Weighting of terms is important and crucial to the performance of a TDT system.The performance may vary dramatically for different weighting methods.In this paper,we propose a new weighting method by taking the position information of terms into account.Specifically,the weight of a term is partially determined by its position information.The experiment result shows that the proposed method is pretty effective and improves the performance of a tracking system.