Crisis MT: Developing A Cookbook for MT in Crisis Situations

Proceedings of the 6th Workshop on Statistical Machine Translation, pages 501–511, Edinburgh, Scotland, UK, July 30–31, 2011. c⃝2011 Association for Computational Linguistics Crisis MT: Developing A Cookbook for MT in Crisis Situations William D. Lewis Microsoft Research Redmond, WA 98052 wilewis@microsoft.com Robert Munro Stanford University Stanford, CA 94305 rmunro@stanford.edu Stephan Vogel Carnegie Mellon University Pittsburgh, PA 15213 stephan.vogel@cmu.edu Abstract In this paper, we propose that MT is an im- portant technology in crisis events, something that can and should be an integral part of a rapid-response infrastructure. By integrating MT services directly into a messaging infras- tructure (whatever the type of messages being serviced, e.g., text messages, Twitter feeds, blog postings, etc.), MT can be used to pro- vide first pass translations into a majority lan- guage, which can be more effectively triaged and then routed to the appropriate aid agen- cies. If done right, MT can dramatically in- crease the speed by which relief can be pro- vided. To ensure that MT is a standard tool in the arsenal of tools needed in crisis events, we propose a preliminary Crisis Cookbook, the contents of which could be translated into the relevant language(s) by volunteers imme- diately after a crisis event occurs. The result- ing data could then be made available to relief groups on the ground, as well as to providers of MT services. We also note that there are significant contributions that our commu- nity can make to relief efforts through con- tinued work on our research, especially that research which makes MT more viable for under-resourced languages. 1 Introduction The connected world contains approximately 5000 languages – at least that is how many languages you could find at the other end of your phone right now. However, the majority of these languages are under- resourced, and they have few or no digital resources. In the event of a sudden onset crisis, people will immediately begin using their communication tech- nologies – and their languages – to report their situ- ations, request help, and seek out loved ones. Yet, in the event that such a crisis occurs in a region of the world where an under-resourced language is spoken, delivery of support or aid could be affected due to the inability to communicate. This was felt most strongly in the wake of the January 12, 2010 earthquake in Haiti. Local emergency response ser- vices were inoperable, but 70-80% of cell-towers were quickly restored. With 83% of men and 67% of women possessing cellphones, the nation remained largely connected. People within Haiti were texting, calling, and interacting with social media, primarily in Haitian Krey`ol (Munro, 2011). Yet, most of the aid that was being delivered to the country – initially, soley by the American Military – was being deliv- ered by groups that did not communicate in Krey`ol. It was the first time that the world has seen a large- scale sudden onset crisis in a region with productive digital communications in an under-resourced lan- guage, but it certainly …

Content truncated to 500 words