Convolution–deconvolution word embedding: An end-to-end multi-prototype fusion embedding method for natural language processing
- Submitting institution
-
The University of West London
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11007
- Type
- D - Journal article
- DOI
-
10.1016/j.inffus.2019.06.009
- Title of journal
- Information Fusion
- Article number
- -
- First page
- 112
- Volume
- 53
- Issue
- -
- ISSN
- 1566-2535
- Open access status
- Access exception
- Month of publication
- -
- Year of publication
- 2019
- 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
-
3
- Research group(s)
-
-
- Citation count
- 9
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This research found that the existing single-prototype word embedding model is incapable of dealing with the phenomena of polysemy and task variance in NLP, leading to misinterpretation of semantic meaning of the words that will directly influence the text reasoning and classification results. This paper is the first to introduce convolution-deconvolution structure (normally found in computer vision) into word embedding, hence called CDWE. With the technique, a new language model CDWE-BLSTM is proposed, which was found to outperform the existing state-of-the-art methods for text classification. This work represents the culmination of research developed with Beijing University of Posts and Telecommunications.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -