Disaster Named Entity Recognizer is a web application that performs named entity recognition in typhoon-related news articles. The application can identify the relevant information and classify into either of the six categories: Typhoon Name, Locations Affected, Affected, Casualties, Damage, and Donations. Articles from Rappler and Inquirer was used to build a corpus of typhoon-related news articles for the training data. Stanford NER served as the main tool of the development of this application. Articles about typhoons were gathered from news portals via an article scraper built using PHP, the scraper is designed only for Rappler and Inquirer websites. This is to build a corpus of typhoon-related news articles. The gathered data are composed of articles on the onset of predetermined typhoons. The corpus was composed of 104 articles from Rappler and 75 articles from Inquirer with a total of 179 articles, all written in the English language. The collection of articles didn’t particularly cover specific types of typhoons but mostly articles related to Typhoon Yolanda were included in the corpus since these articles contain loads of information that helps in training the model. This work is in partnership with Bicol University and National University under the PCARI-funded project entitled ‘E-Participation 2.0: Connective Diverse Philippine Populations for Disaster Risk Management with a Toolkit Integrating Text and Speech Analytics‘.
Alejandro Dela Rosa
Mary Joy Canon
Article Scraper – A web scraper built using PHP to collect news related articles. Designed only for Rappler and Inquirer.
Stanford NER CRF Classifier – NER system use to create the classifier model.
The Disaster Named Entity Recognizer is currently hosted locally at National University-Manila.