On-line event and topic detection by using the compact sets clustering algorithm

摘要: In this paper we propose a new incremental clustering algorithm for Event Detection, which is based on the mathematical properties of compact sets. Additionally, makes use temporal references appearing in document texts to measure similarity between documents according events that they describe. order discover structure topics and composite events, hierarchically applied stream newspaper articles. Thus, first level, with high temporal-semantic are clustered together into events. next levels hierarchy, these successively more complex topics. The evaluation results demonstrate regarding improves quality system-generated clusters, overall performance proposed system compares favorably other on-line detection systems literature.