Constrained multi-task learning for automated essay scoring
- Submitting institution
-
University of Cambridge
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1859
- Type
- E - Conference contribution
- DOI
-
10.18653/v1/p16-1075
- Title of conference / published proceedings
- 54th Annual Meeting of the Association for Computational Linguistics, ACL 2016 - Long Papers
- First page
- 789
- Volume
- 2
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- January
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
-
- Citation count
- 6
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Scoring a free text essay using machine learning requires a dataset of manually scored essays to train a system. This paper demonstrates that it is possible to accurately score essays, given very limited training data scored according to the target rubric and scale, by utilising essays on a wider range of topics with differing scoring schemes in a more coarse-grained ranking model, which is then fitted to the target scoring system. The approach opened up the way to development of neural essay scoring models, which typically require considerable quantities of labelled training data, by our group and others.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -