W&M Featured Events
This calendar presented by
William & Mary
[PAST EVENT] Colloquium on Feb. 25 at 8AM in McGl 020
February 25, 2015
8am
Title: Automated Methods for Text Correction
Speaker: Alla Rozovskaya, Columbia University
Abstract: 200 billion emails are being written each day; 200 billion tweets and 1.8 million scientific articles are written a year. Most of these are written in English by non-native speakers. Nevertheless, existing word processors only mark simple out-of-vocabulary errors and a very limited number of context-sensitive mistakes. With the development and pervasiveness of machine-learning algorithms, text correction is becoming an important research area with a range of applications, from automatic writing assistance, to supporting second language learning, to a broad range of applications dealing with noisy natural language.
In this talk, I will present some of my research on developing computational models for correcting writing mistakes in text, with a focus on errors made by English as a Second Language (ESL) learners. I identify and address several key issues that are essential to making progress in this area, and describe a robust, state-of-the-art system that combines machine-learning methods and linguistic knowledge, and corrects some of the most common (context-sensitive) mistakes.
My work builds on the fact that mistakes made by ESL writers are, for the most part, systematic and often source-language specific, and develops machine-learning models that utilize knowledge about error regularities with minimal annotation costs. The techniques and systems described in the talk have been evaluated empirically in the context of several competitions in this area, where they have demonstrated superior performance.
Bio: Alla Rozovskaya is a Postdoctoral Research Scientist at Columbia University. She received her Ph.D. in Computational Linguistics from the University of Illinois at Urbana-Champaign under the direction of Professor Dan Roth. She also holds an M.S. degree in Computer Science from UIUC and an M.A. degree in French Studies from SUNY Albany. Her research focuses on applying machine learning and natural language processing techniques for developing data-driven intelligent learning systems. She is particularly interested in building educational technology and social media data analytics. She was an invited participant of the Rising Stars in EECS Workshop (2014). Her thesis focused on developing automated methods for text correction. The models that she developed won several text correction competitions.
Speaker: Alla Rozovskaya, Columbia University
Abstract: 200 billion emails are being written each day; 200 billion tweets and 1.8 million scientific articles are written a year. Most of these are written in English by non-native speakers. Nevertheless, existing word processors only mark simple out-of-vocabulary errors and a very limited number of context-sensitive mistakes. With the development and pervasiveness of machine-learning algorithms, text correction is becoming an important research area with a range of applications, from automatic writing assistance, to supporting second language learning, to a broad range of applications dealing with noisy natural language.
In this talk, I will present some of my research on developing computational models for correcting writing mistakes in text, with a focus on errors made by English as a Second Language (ESL) learners. I identify and address several key issues that are essential to making progress in this area, and describe a robust, state-of-the-art system that combines machine-learning methods and linguistic knowledge, and corrects some of the most common (context-sensitive) mistakes.
My work builds on the fact that mistakes made by ESL writers are, for the most part, systematic and often source-language specific, and develops machine-learning models that utilize knowledge about error regularities with minimal annotation costs. The techniques and systems described in the talk have been evaluated empirically in the context of several competitions in this area, where they have demonstrated superior performance.
Bio: Alla Rozovskaya is a Postdoctoral Research Scientist at Columbia University. She received her Ph.D. in Computational Linguistics from the University of Illinois at Urbana-Champaign under the direction of Professor Dan Roth. She also holds an M.S. degree in Computer Science from UIUC and an M.A. degree in French Studies from SUNY Albany. Her research focuses on applying machine learning and natural language processing techniques for developing data-driven intelligent learning systems. She is particularly interested in building educational technology and social media data analytics. She was an invited participant of the Rising Stars in EECS Workshop (2014). Her thesis focused on developing automated methods for text correction. The models that she developed won several text correction competitions.
Contact
[[cs|kemper, Peter Kemper]]