LanguageTool (LT) currently uses neural networks to detect confusion between words. So far, it only considers 2 words of context in both directions (tri-grams). The aim of this project is to extend this to take the entire sentence as context to detect the confusion pairs. For this, the project aims to leverage the use of a sequence-to-sequence model to detect the word(s) that are used incorrectly and then replace them with the correct word(s). The project aims to give this functionality to English, French, and German.