Machine Learning based Classification of Online News Data for Disaster Management
Lakshmi Gopal, Rekha Prabha, Divya Pullarkatt, Maneesha Vinodini Ramesh
Abstract
The exponential escalation of disaster loss in our country has led to the awareness that disaster risks are presumably increasing. As per statistics, India has confronted 371 natural hazards over the past few decades and severe casualties, infrastructural, agricultural and economic damages were recorded [1]. Credible and real time data such as news content are accessible liberally in legitimate websites and its analysis may provide assistance in administering hazard emergencies, preparedness and relief efficiently. On this grounds, a data scraping approach is proposed to gather hazard relevant news stories from the web by building a crawler software and incorporate machine learning approaches to filter out insightful information. The developed crawler software visits news reporting web pages and extracts news stories related to hazards. News illustrations are often unstructured as it includes less newsworthy content such as author’s opinions, interview responses and past studies. Hence, a supervised learning based text classification is performed to classify newsworthy content from news articles and approximately 70 percent accuracy was achieved.