Over the last several decades, the number of electronic documents has increased dramatically. With the growing availability of computers, more and more people are using text editors. However, the development of automated methods for correcting mistakes in text has not progressed as far. Text editors usually employ basic spell checking techniques and address very few mistakes of other types.In this thesis, we propose two methods for correcting errors in grammar and usage. First, we propose a novel approach to the problem of training classifiers to detect and correct errors in text by selectively introducing mistakes into the training data and show that this method is superior to the traditional method of training using clean data. Second, we define high-level features and propose a method of correcting mistakes using these features. We combine the two methods and build a system for correcting mistakes in article usage made by non-native speakers of English.
展开▼
