Deep Learning for English and Chinese Grammatical Error Detection
| Autor: | Bo-Lin Lin, 林柏霖 |
|---|---|
| Rok vydání: | 2017 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 105 Automatic grammatical error detection is essential for all language learners of the world. It can not only reduce the errors but also increase the quality when writing articles. Moreover, it's able to decrease the costs of human revising. This research combines deep learning and nature language processing in order to design and construct a system of grammatical error detection. Under this system, it can verify the efficiency of deep learning in different language by using English and mandarin data. The research includes three sections: word embedding, deep learning model construction and validation. First of all, word embedding applies to some skills, such as GloVe and Word2Vec, to convert language into dimension of vector space. Next, model construction is based on fundamental model of deep learning: CNN and LSTM. It designs and compares a series of model structure combination. Last, this research concludes the resent experiment results, and provides the issues which can refer in the future. In advance, the research implements an automatic grammatical error detection. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|