|
Rao et al. (2014)
|
Sentiment pruning strategies to improve word-level sentiment lexicons. |
Sina Dataset, SemEval-2007 Dataset |
1,854,718, 109,129 |
|
Han et al. (2018)
|
The unlabeled reviews are used to generate domain-specific sentiment lexicon |
Movie Review Dataset, Amazon Dataset |
50,000, 1,140,496 |
|
Liang, Liu & Zhang (2020)
|
After regularizing the dataset and using the Doc2vec+SVM method |
Chinese Opinion Analysis Evaluation (COAE) 2013, COAE2015 |
22,890, 30,049 |
|
Elgeldawi et al. (2021)
|
Five hyperparameter methods for hyperparameter tuning of machine learning algorithms |
Hotel Review Dataset |
7,000 |
|
Al-Hadhrami, Al-Fassam & Benhidour (2019)
|
Sentiment weights are calculated using TF-IDF and classified using machine learning. |
Sentiment140 Dataset |
1,578,612 |
|
Kim (2014)
|
CNN for extracting word character features |
Movie Review (MR) Dataset, Stanford Sentiment Treebank (SST)-1 Dataset, SST-2 Dataset, Subjectivity Dataset, TREC Question Dataset, Customer reviews Dataset, Multi-Perspective Question Answering (MPQA) Dataset |
10,662, 11,855, 9,613, 10,000, 5,952, 3,773, 10,604 |
|
Zaremba, Sutskever & Vinyals (2014)
|
LSTM extracts semantic information |
Penn Tree Bank (PTB) Dataset |
900,000 |
|
Wang (2022)
|
Improved LSTM framework for tourism theme feature extraction |
Rural Tourism Dataset |
18,566 |
|
Li & Ning (2020)
|
CNN and LSTM hybrid models are used to derive both local and global features |
THUCNews Dataset |
740,000 |
|
Wang & He (2022)
|
BiGRU is used to extract global information, while the attention mechanism performs feature weighting to emphasize important features |
Weibo Sentiment 100k Binary Dataset |
119,988 |
|
Zeng, Yang & Zhou (2022)
|
The topics are classified by LDA, and the text features are extracted by BiLSTM-ATT |
COVID-19 Review Dataset |
30,525 |
