Skip to main content

Table 13 Mean rank table for datasets of social networks

From: SentiBench - a benchmark comparison of state-of-the-practice sentiment analysis methods

3-classes 2-classes
Pos Method Mean Rank Pos Method Mean Rank Coverage (%)
1 Umigon 2.57 1 SentiStrength 2.22 (2.57) 31.54 (32.18)
2 LIWC15 3.29 2 Sentiment140 3.00 46.98
3 VADER 4.57 (4.57) 3 Emoticons 5.11 18.04
4 AFINN 5.00 4 LIWC15 5.67 71.73
5 Opinion Lexicon 5.57 5 Semantria 5.89 61.98
6 Semantria 6.00 6 PANAS-t 6.33 5.87
7 Sentiment140 7.00 7 Opinion Lexicon 7.56 66.56
8 Pattern.en 7.57 8 Umigon 8.00 71.67
9 SO-CAL 9.00 9 AFINN 8.67 73.37
10 Emolex 12.29 10 SO-CAL 8.78 67.81
11 SentiStrength 12.43 (11.60) 11 VADER 8.78 (9.75) 83.29 (81.90)
12 Opinion Finder 13.00 12 Pattern.en 11.22 69.47
13 SentiWordNet 13.57 13 Sentiment140_L 14.00 94.61
14 SenticNet 14.14 14 Opinion Finder 14.33 39.58
15 SASA 14.86 15 Emolex 14.56 62.63
16 LIWC 15.43 16 USent 15.22 38.60
17 Sentiment140_L 15.43 17 SenticNet 17.22 75.46
18 USent 16.00 18 SentiWordNet 18.44 61.41
19 ANEW_SUB 19.14 19 NRC Hashtag 19.11 94.20
20 Emoticons 19.14 20 SASA 19.44 58.57
21 Stanford DM 19.43 21 LIWC 19.56 61.24
22 NRC Hashtag 20.00 22 ANEW_SUB 20.56 93.51
23 PANAS-t 20.86 23 Stanford DM 22.56 89.06
24 Emoticons DS 23.71 24 Emoticons DS 23.78 99.28