A Noise-Filtering Approach for Spatio-temporal Event Detection in Social Media

Yuan Liang, James Caverlee, Cheng Cao

Abstract: We propose an iterative spatial-temporal mining algorithm for identifying and extracting events from social media. One of the key aspects of the proposed algorithm is a signal processing-inspired approach for viewing spatial-temporal term occurrences as signals, analyzing the noise contained in the signals, and applying noise filters to improve the quality of event extraction from these signals. The iterative event mining algorithm alternately clusters terms and then generates new filters based on the results of clustering. Through experiments on ten Twitter data sets, we find improved event retrieval compared to two baselines.