This project will focus on automatic correction of grammatical errors in non-native English texts. This is an important task with a multitude of applications in education and business. Current systems typically handle a small number of error types, and use local classifiers to identify and correct them. Departing from them, we first develop a new formulation for error correction as global prediction in a sequence labeling setting. Secondly, to further enhance the performance of our system, we plan to enrich the model with information particular to the authors’ native language, such as the expected distribution of mistake types for that language, as well as its lexical, syntactic and typological properties

I took a machine learning class. The knowledge covered was highly relevant to the project, and the open-ended project gave me the ability to explore new technologies. Also, as a non-native speaker, I always need a better grammar checking tool myself. As a result, I have a sense of how it should be like.