University of Twente Student Theses
Creating and evaluating Grammatical Error Correction models with arbitrary error correction profiles
Klampe, T. (2023) Creating and evaluating Grammatical Error Correction models with arbitrary error correction profiles.
|
PDF
458kB |
| Abstract: | Generating high-quality synthetic datasets with use case specific error frequencies can boost the performance of Grammatical Error Correction models substantially. In this paper, I propose a system in which datasets are created according to specific error frequencies with a tagged grammatical corruption model. The effect of these frequencies is then evaluated in error-specific accuracy testing. The system can be used to flexibly generate synthetic datasets and then train a grammatical error correction model. The accuracy of said model is analyzed and then can be iteratively improved by changing error frequencies in the dataset and comparing the effects on the accuracy. I will demonstrate the generation and evaluation of a grammatical error correction model that takes the expected error profile of a native English speaker into consideration. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/95997 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page