Extracting 5W from Baidu Hot News Search Words for Societal Risk Events Analysis

Abstract

Nowadays risk events occur more frequently than ever in China during the critical periods of social and economic transformation and spread rapidly via a variety of social media, which have impacts on social stability. Online societal risk perception is acquired by mapping online community concerns into respective societal risks. What we concern is how to recognize and describe societal risk events in a formal and structured way. So to get a structured view of those risk events, we propose an event extraction framework on HNSW including 5 elements, namely where, who, when, why, and what (5W). The task for extracting 5W of risk events is converted into different machine learning tasks. Three methods are explored to tackle the extraction tasks. The framework of 5W extraction on the basis of online concerns can not only timely access to societal risk perception but also expose the events by a structured image, which is of great help for social management to monitor online public opinion timely and efficiently.