Sentiment-Driven Breakdown Detection Model Using Contextual Embedding ElMo
- Publisher:
- Springer
- Publication Type:
- Chapter
- Citation:
- Advanced Information Networking and Applications, 2021, 225, pp. 163-171
- Issue Date:
- 2021-01-01
Recently Added
Filename | Description | Size | |||
---|---|---|---|---|---|
Almansor-Hussain2021_Chapter_Sentiment-DrivenBreakdownDetec.pdf | Published version | 504.42 kB |
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is new to OPUS and is not currently available.
The recent rapid developments in neural networks have stimulated significant performance progress in detecting breakdown in dialogue. The existing research focuses on detecting breakdown from the data label using different features such as word similarity and topic transition; however, these features are insufficient to capture contextual features. Therefore, we used ELMo language model to extract the contextual feature to help to detect the breakdown which faces a challenge due to the basis of human opinion. Another crucial challenge facing breakdown detection is the lack annotated dataset. To take this challenge we used sentiment as the data label with contextualized ELMo embedding model to detect the breakdown in dialogue. The model was evaluated with state-of-the-art models and the results shows the proposed model outperforms other models.
Please use this identifier to cite or link to this item: