Sentiment Analysis Parser Model Datasets

The distribution of sentiment among datasets used to train the given models

BINARY

1. Netflix model --> /examples/result

File size: 33.1 MB = 0.33 GB (this is due to the large size of each record)

# of records: 1,019

• Positive: 625
• Negative: 394






2. HT Provider Review model --> /reviews/review_bin_dataset

File size: 1.9 MB = 0.0019 GB

# of records: 1,057

• Positive (score >= 6.0): 943
• Negative (score <= 4.0): 114

CATEGORICAL

1. Stanford model --> /examples/categorical_dataset

File size: 12 MB = 0.012 GB

# of records: 239,232

• Angry: 11,432
• Sad: 43,952
• Neutral: 119,468
• Like: 57,135
• Love: 17,960






2. HT Provider Review model --> /reviews/review_categ_dataset

File size: 2.1 MB = 0.0021 GB

# of records: 1,164

• Angry: 49
• Sad: 65
• Neutral: 278
• Like: 639
• Love: 133