Selective Reading for Arabic Sentiment Analysis
This work introduces a novel deep learning method for Arabic sentiment analysis, arguing that reading the entire input sequence is not always necessary. Many texts can be accurately classified without processing all input tokens. The method employs a reinforcement learning agent that selects relevan...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10947753/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This work introduces a novel deep learning method for Arabic sentiment analysis, arguing that reading the entire input sequence is not always necessary. Many texts can be accurately classified without processing all input tokens. The method employs a reinforcement learning agent that selects relevant tokens using a selection policy network. Instead of predicting sentiment polarity from the entire input, the model focuses only on tokens chosen by the policy network. To empirically evaluate the proposed method, experiments were carried out on three Arabic sentiment analysis datasets: Large Arabic Book Reviews (LABR), Hotels Arabic Reviews Data (HARD), and Arabic Sentiment Tweets Dataset (ASTD). The results demonstrate a significant improvement in Arabic sentiment classification with the selective reading method, achieving state-of-the-art accuracy while using only a fraction of the tokens. However, the approach introduces additional computational cost due to the reinforcement learning component, and its scalability to larger datasets might require further optimization. |
|---|---|
| ISSN: | 2169-3536 |