Multi-scale and Hierarchical Embedding for Polarity Shift Sensitive Sentiment Classification
- Publisher:
- Springer International Publishing
- Publication Type:
- Conference Proceeding
- Citation:
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2019, 11633 LNCS, pp. 227-238
- Issue Date:
- 2019-01-01
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| Li2019_Chapter_Multi-scaleAndHierarchicalEmbe.pdf | Published version | 1.07 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
© 2019, Springer Nature Switzerland AG. Appropriate paragraph embedding is critical for sentiment classification. However, the embedding for paragraph with polarity shift is very challenging and insufficiently explored. In this paper, a MUlti-Scale and Hierarchical embedding method, MUSH, is proposed to learn a more accurate paragraph embedding for polarity shift sensitive sentiment classification. MUSH adopts CNN with multi-size filters to reveal multi-scale sentiment atoms and utilizes hierarchical multi-line CNN-RNN structures to simultaneously capture polarity shift in both sentence level and paragraph level. Extensive experiments on four large real-world data sets demonstrate that the MUSH-enabled sentiment classification significantly enhances the accuracy compared with three state-of-the-art and four baseline competitors.
Please use this identifier to cite or link to this item: