EET: Efficient event tracking over emergency-oriented web data

In recent years, with the prevailing usage of web data repository over emergency field, extracting the web data plays an important role in tracking the evolution of event. With huge and still growing web data, an important task is to track emergency event evolution and analyze its latent events or topics over time. Unfortunately, emergency events are inherently with uncertainty because the complexity of evolution and timeliness in web data extraction. This is why the existing techniques in topic discovery and event tracking cannot effectively work, these approaches ignore the interplay between event, latent event and network information. In this paper, we propose a new emergency event tracking model EET (Emergency Event Tracking model), to effectively track emergency event and predict its latent events from time-stamped web data. Moreover, in order to estimate the finish time of EET over a specific event, we introduce a novel web data extraction algorithm WDEE (Web Data Extraction for emergency Event algorithm) which utilizes limit theory to determine the periodical convergence time. Experiment results demonstrate the capability of EET via WDEE to track the event evolution.