33 lines
1.3 KiB
Markdown
33 lines
1.3 KiB
Markdown
|
FCE - Grammatical error detection
|
||
|
|
===========================
|
||
|
|
|
||
|
|
Detect errors in English text.
|
||
|
|
|
||
|
|
This is a Gonito.net challenge based on data from https://ilexir.co.uk/datasets/index.html
|
||
|
|
The aim of the challenge is to predict which tokens are incorrect.
|
||
|
|
|
||
|
|
MultiLabel-F0.5 is used as the evaluation metric.
|
||
|
|
|
||
|
|
Dataset reference:
|
||
|
|
1) Compositional Sequence Labeling Models for Error Detection in Learner Writing
|
||
|
|
Marek Rei and Helen Yannakoudakis
|
||
|
|
In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (ACL-2016)
|
||
|
|
|
||
|
|
2) A New Dataset and Method for Automatically Grading ESOL Texts
|
||
|
|
Helen Yannakoudakis, Ted Briscoe and Ben Medlock
|
||
|
|
In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics (ACL-2011)
|
||
|
|
|
||
|
|
## Directory structure
|
||
|
|
|
||
|
|
* `README.md` — this file
|
||
|
|
* `config.txt` — configuration file
|
||
|
|
* `train/` — directory with training data
|
||
|
|
* `train/in.tsv` — Original input text for the train set
|
||
|
|
* `train/expected.tsv` — Incorrect token indexes. Indexes start from 1.
|
||
|
|
* `dev-0/` — directory with dev data
|
||
|
|
* `dev-0/in.tsv` — Original input text for the dev set
|
||
|
|
* `dev-0/expected.tsv` — Incorrect token indexes. Indexes start from 1.
|
||
|
|
* `test-A` — directory with test data
|
||
|
|
* `test-A/in.tsv` — Original input text for the test set
|
||
|
|
|