784
LITMUS: Towards Multilingual Reporting of Landslides
Aibek Musaev, Qixuan Hou, Yang Yang and Calton Pu
University of Alabama, Georgia Institute of Technology, Georgia Institute of Technology, Georgia Institute of Technology

LITMUS is a real-time online and openly accessible service that collects high quality information on landslide events from social media. This service uses disaster related keywords, such as "landslide" and "mudslide", to analyze messages posted by English speaking users. However, comprehensive coverage of disasters must include multilingual support as there are events that are reported in languages other than English. We discuss and evaluate possible implementations of such support using "native" and "translated" approaches. "Native" approach involves a complete reimplementation of the existing infrastructure in another language whereas in the "translated" approach the existing infrastructure can be used without modification. As an illustration, we present a demo that extends LITMUS to implement a "native" approach for multilingual reporting of landslide events.