MSBKA: A Multi-Strategy Improved Black-Winged Kite Algorithm for Feature Selection of Natural Disaster Tweets...

Area of Science:

• Natural Language Processing
• Artificial Intelligence
• Computational Social Science

Background:

• Social media platforms are crucial for disseminating crisis information during natural disasters.
• Effective identification of relevant disaster tweets is vital for efficient rescue operations.
• Handling large volumes of text data presents challenges in feature selection and classification accuracy.

Purpose of the Study:

• To propose an effective feature selection method for natural disaster tweets.
• To enhance the classification performance of natural disaster-related content on social media.
• To develop a hybrid model for improved crisis informatics.

Main Methods:

• A multi-strategy improved black-winged kite algorithm (MSBKA) was developed for feature selection.
• MSBKA incorporates enhanced Circle mapping, hierarchical reverse learning, and the Nelder-Mead method.
• A hybrid model combining MSBKA with Support Vector Machine (SVM) using an RBF kernel was implemented.

Main Results:

• The MSBKA-SVM model achieved an accuracy of 0.8822 on natural disaster tweet classification.
• The proposed model demonstrated superior performance compared to Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Sine Cosine Algorithm (SSA).
• Accuracy improvements of 4.34% (GA), 2.13% (PSO), 2.94% (SSA), and 6.35% (BKA) were observed.

Conclusions:

• The MSBKA-SVM model offers a robust solution for feature selection and classification of natural disaster tweets.
• This approach can significantly support disaster risk reduction efforts by improving the analysis of crisis-related social media data.
• The study highlights the potential of advanced metaheuristic algorithms in enhancing crisis informatics systems.