Sentiment Analysis Model
Sentiment analysis models aim to automatically determine the emotional tone expressed in text, with applications ranging from social science research to market trend prediction. Current research focuses on improving model accuracy and interpretability, exploring architectures like transformers (e.g., BERT, RoBERTa), recurrent neural networks (RNNs), and convolutional neural networks (CNNs), as well as incorporating techniques like quantum-inspired methods and ensemble learning. Addressing biases stemming from training data, particularly political and sociodemographic biases, is a crucial area of investigation, alongside enhancing robustness against adversarial attacks and adapting models for low-resource languages and informal online communication.
Papers
Uncovering Political Bias in Emotion Inference Models: Implications for sentiment analysis in social science research
Hubert Plisiecki, Paweł Lenartowicz, Maria Flakus, Artur Pokropek
Dynamic Sentiment Analysis with Local Large Language Models using Majority Voting: A Study on Factors Affecting Restaurant Evaluation
Junichiro Niimi