09-22, 13:35–14:10 (US/Eastern), Main Room
This talk will overview how NLP is being used in research and industry to preserve at-risk languages, power technologies to solve pressing problems (like employment matches), and create culturally-attuned NLP tools (like sentiment analysis). Current challenges include data ownership and local population rights to their data. Examples come from partnerships in Sub-Saharan Africa, but they apply to other regions of the world, as well.
Examples in the talk will include building customized sentiment dictionaries and BERT embeddings for languages not currently supported by open-source and commercial NLP platforms through cooperation with local dialect speakers and leveraging NLP to power job application platforms.
Previous knowledge expected
Colleen Farrelly is a lead data scientist whose expertise spans many industries and machine learning applications, including topological data analysis, NLP, and time series methods. She volunteers her time as a mentor on projects related to her expertise in Sub-Saharan Africa.