Table 1 Comparison of different approaches in sentiment analysis
From: Stress detection using natural language processing and machine learning over social interactions
Researcher | Technique | Performances |
|---|---|---|
Pak and Paroubek [5] | Naïve Bayes Sentiment classifier with multinomial features | High accuracy, Low decision value |
Alec et al. [23] | Naïve Bayes classifier, Mutual information measure for feature selection | Accuracy: 81% |
Balahur et al. [24] | WordNet-lexicon | Accuracy: 82% Improvement in the baseline 21% |
Jonathon et Al. [25] | SVM, Naïve Bayes | Accuracy: 70% |
Boiy et al. [26] | Integrated approach: ML, Information retrieval, NLP | Accuracy: 83% (English texts),70% accuracy (Dutch texts), 68% (French texts) |
Li et al. [27] | Dependency- Sentiment, LDA, Markov chain | Accuracy: 70.7% with on tenfold cross-validation test set: 800 reviews |