一种构建-竞争聚类法被用于事件探测,该方法是受神经网络研究中构建-竞争学习的思想启发的.另外,提出了一种用于事件追踪的基于K近邻特征线(KNNFL)的分类方法,这种基于最近邻特征线(NFL)的方法本质上可以看作是对K近邻(KNN)法的推广,将改进后的KNN融入到NFL中形成KNNFL是为了更适合新闻事件的分析.研究结果表明,本文所提出的方法与传统的增量k均值法、Single-Pass法、Rocchio法以及KNN法相比较,可以获得更好的效果.通过分析可以看到,KNNFL即使在正例样本非常稀少的情况下仍然具有鲁棒性的表现. A construction-competition clustering method is used for event detection, which is inspired by the idea of ​​constructing-competing learning in neural network research.In addition, a KNNFL (Feature-Based Approach) , This method based on the nearest neighbor characteristic line (NFL) can essentially be regarded as an extension of the K-nearest neighbor (KNN) method. The improved KNN into the NFL to form KNNFL is more suitable for news events The results show that the proposed method can get better results than the traditional incremental k-means method, Single-pass method, Rocchio method and KNN method. It can be seen from the analysis that KNNFL even in the positive case Samples are still robust with very few samples.