Issue |
ITM Web Conf.
Volume 70, 2025
2024 2nd International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2024)
|
|
---|---|---|
Article Number | 02008 | |
Number of page(s) | 8 | |
Section | Machine Learning in Healthcare and Finance | |
DOI | https://doi.org/10.1051/itmconf/20257002008 | |
Published online | 23 January 2025 |
Research on Sarcastic Emotion Recognition Based on Multiple Feature Fusion
Cyberspace Security Major, Northwestern Polytechnical University, 710129 Xi’an, China
Corresponding author: 130102200401151816@mail.nwpu.edu.cn
Sarcasm detection significantly enhances the performance of various natural language processing applications, such as sentiment analysis, opinion mining, and stance detection. Despite considerable advancements in this field, research results remain fragmented across diverse datasets and studies. This paper offers a critical review of two predominant models in sarcasm detection. The first model utilizes BERT within an intermediate task transfer learning framework, leveraging the connection between sarcasm and underlying negative emotions and sentiments. This model enhances the sarcasm detection capability through a strategic knowledge infusion into the transfer learning process. The second model reviewed deploys a multi-head attention-based bidirectional LSTM architecture. This approach incorporates pre-trained word embeddings, multi-head attention mechanisms, and custom-crafted features to proficiently identify sarcasm in social media datasets. Comparative assessments on standard datasets reveal that both models achieve superior performance over many existing approaches in the field. At last, this paper explores the direction for future improvement based on the conclusions.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.