Improved gray wolf harris hawk algorithm based feature selection for sentiment analysis
Sentiment Analysis (SA) is a key area of data mining that focuses on analyzing emotions in social media documents. However, these documents often contain redundant and irrelevant features, leading to high-dimensional datasets that reduce SA performance. Efficient sentiment feature selection (FS) is...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Results in Control and Optimization |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2666720725000906 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Sentiment Analysis (SA) is a key area of data mining that focuses on analyzing emotions in social media documents. However, these documents often contain redundant and irrelevant features, leading to high-dimensional datasets that reduce SA performance. Efficient sentiment feature selection (FS) is crucial for reducing data dimensionality and isolating relevant features to improve results. This study aims to enhance FS performance by addressing the population diversity issues in the Adaptive β Binary Gray Wolf Optimization (Aβ-BGWO) algorithm, which struggles to escape local optima. To address this, a hybrid algorithm combining Aβ-BGWO with Harris Hawks Optimization (HHO) is proposed, resulting in the Aβ-BGWHHO approach for optimized FS in SA. The effectiveness of selected features is evaluated using the KNN classifier, and performance is assessed across 18 UCI datasets, comparing it with recent metaheuristic FS algorithms. Population convergence and diversity are measured using dimension-wise diversity to evaluate exploration and exploitation behavior. Furthermore, four Arabic benchmark datasets and six state-of-the-art optimization techniques are used for SA assessment. Experimental results show that the Aβ-BGWHHO algorithm outperforms other algorithms regarding accuracy, feature reduction, and fitness value. The hybrid approach enhances population diversity, allowing the algorithm to effectively balance exploration and exploitation, resulting in superior performance. The integration of HHO significantly improves the algorithm’s ability to escape local optima, making the binary hybrid Aβ-BGWHHO algorithm a more effective tool for SA, addressing Aβ-BGWO’s limitations and enhancing FS outcomes. |
|---|---|
| ISSN: | 2666-7207 |