Learning to predict distributions of words across domains
- Submitting institution
-
University of Sussex
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 25069_53735
- Type
- E - Conference contribution
- DOI
-
10.3115/v1/P14-1058
- Title of conference / published proceedings
- Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- First page
- 613
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- June
- Year of publication
- 2014
- URL
-
http://dx.doi.org/10.3115/v1/P14-1058
- 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
- 7
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- "Published at ACL'14 as a long paper; this category had 26% acceptance rate. ACL is rated A* in CORE2018. The paper reports the first successful attempt to project a word embedding (a feature vector approximating the meaning of the word) from one text domain to another. Evaluations on cross-domain text classification and part-of-speech tagging rigorously demonstrate the value of the approach. Has inspired further well-cited work in inducing embeddings for rare or unseen words [1], joint learning of embeddings and classifiers [2], and guiding the projection of embeddings by adding limited, supervised data [3].
[1] https://www.aclweb.org/anthology/P18-1002
[2] https://www.aclweb.org/anthology/C18-1070
[3] https://dl.acm.org/doi/10.5555/2832415.2832427"
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -