In the past ten years, the number of Chinese language learners worldwide has increased dramatically. Compared to the English learning environment, there are many systems to assist in English learning, and the auxiliary tools for Chinese learners are relatively scarce, especially systems that can automatically detect and correct typos and grammatical errors. We focus on developing Chinese error diagnosis systems using language analysis and machine learning techniques to detect and correct writing errors of Chinese learners. When multiple Chinese sentences are entered, such a diagnostic system can check possible typos and grammatical error types in each Chinese sentence, and then provide correction suggestions for learners' correct usage in that context. So far, the project has used rules and convolutional neural networks to connect long-term and short-term memory models to detect whether Chinese sentences written by Chinese learners are grammatically wrong. However, this research is a difficult task. Currently, The results achieved are not yet satisfactory. The project also builds and completes a TOCFL learner corpus, collects 2,837 essays from Chinese learners from 46 different mother tongues, proposes a hierarchical error marker set, and completes 33,835 grammatical error markers. This project organized the 8th International Natural Language Processing Symposium, and organized the 4th Natural Language Processing Technology and Education Application Workshop. This project also published a special book chapter. We present the corpus we have built and used in the research process, the international competitions and workshops, and the technologies and methods used by the participating teams, which is published by Springer in Chapter 12 of the book "Computational and Corpus Approaches to Chinese Language Learning". Finally, this project participated in the "2018 Chinese Language Teaching Application Software Competition" of the National Institute of Education, leading students Lin Yu-Chi and Wu Wen-Hsiuan to win the 2nd place with "NTNU-NCU Chinese Typo Detection and Correction System".
|Effective start/end date||2017/08/01 → 2019/10/31|
- computer-assisted language learning
- Chinese language learning
- grammatical error detection
- learner corpora
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.