
Sentiment Classification and Visualization of Product Review Data 143
Feature Information degree regarding sentiment
(’return’, ’item’) = True neg : pos = 22.6 : 1.0
(’money’, ’back’) = True neg : pos = 22.2 : 1.0
(’sent’, ’back’) = True neg : pos = 21.6 : 1.0
(’high’, ’recommend’) = True pos : neg = 19.5 : 1.0
(’took’, ’phone’) = True neg : pos = 17.3 : 1.0
(’not’, ’return’) = True neg : pos = 17.3 : 1.0
crisp = True pos : neg = 17.3 : 1.0
(’worst’, ’phone’) = True neg : pos = 16.8 : 1.0
shot = True pos : neg = 16.6 : 1.0
(’not’, ’unlock’) = True neg : pos = 16.5 : 1.0
refund = True neg : pos = 15.6 : 1.0
TABLE 6.4: Example results for information degree ...