CMU Multilingual NLP - The LORELEI Project

Explore the LORELEI (Low Resource Languages for Emergent Incidents) project in this lecture by Alan Black, focusing on rapidly developing low-resource information extraction for new languages. Delve into the US government's investment in language technologies and the project's scenario for evaluating systems in emergency situations. Examine the CMU Ariel system, various techniques employed, and performance comparisons between systems and human annotators. Learn about experiments conducted on English core data and the lessons gleaned from the project. Discover how LORELEI's principles were applied in a real disaster scenario and discuss the project's lasting impact on the field of multilingual natural language processing.

11-737 Multilingual NLP
Overview
Government Investment in Languages
US Government LT Investment
The Scenario
Lorelei Incident
Lorelei Evaluation Exercises
Lorelei Performers
CMU System: Ariel
Techniques
Lorelei Questions
System vs Annotator Performance
Experiments on English Core Data
Lessons Learned
Let's Try It in a Real Disaster
Lorelei's Legacy
IR: Discussion Point