TSA-CNN-AOA: Twitter sentiment analysis using CNN optimized via arithmetic optimization algorithm

Area of Science:

• Computational Linguistics
• Social Media Analysis
• Public Health Informatics

Background:

• The COVID-19 pandemic has caused significant psychological distress globally.
• Social media platforms like Twitter are crucial for understanding public sentiment.
• Effective analysis of public opinion on social media is vital for mitigating negative psychological impacts.

Purpose of the Study:

• To develop and evaluate a novel Twitter Sentiment Analysis (TSA) approach for COVID-19 related tweets.
• To enhance TSA performance by optimizing a Convolutional Neural Network (CNN) with the Arithmetic Optimization Algorithm (AOA).
• To accurately classify public sentiment regarding COVID-19 from Twitter data.

Main Methods:

• Extracted 173,638 COVID-19 tweets from Twitter using a custom API.
• Utilized FastText Skip-gram for feature extraction from the tweet database.
• Employed a CNN as a feature extractor, optimized by AOA for feature selection.
• Classified tweets into positive, negative, and neutral categories using K-nearest neighbors (KNN), support vector machine, and decision tree algorithms.

Main Results:

• The proposed TSA-CNN-AOA approach, specifically with KNN classifier, achieved a high accuracy rate of 95.098% for tweet classification.
• The method demonstrated superior performance compared to various other TSA approaches.
• Experimental results confirm the effectiveness of the AOA optimization for the CNN model in sentiment analysis.

Conclusions:

• The TSA-CNN-AOA (KNN) method offers a highly accurate and effective solution for analyzing public sentiment on social media during health crises like COVID-19.
• This approach can be instrumental in understanding and addressing the psychological impact of pandemics.
• The study highlights the potential of optimized deep learning models for large-scale social media data analysis.