Skip to main content

Table 10 Best method for each dataset - 2-class experiments

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

Dataset Method F 1-Pos F 1-Neg Macro- F 1 Coverage
Comments_BBC SentiStrength 70.59 96.61 83.60 32.85
Comments_Digg SentiStrength 84.96 94.64 89.80 27.49
Comments_NYT SentiStrength 70.11 86.52 78.32 17.63
Comments_TED Emoticons 85.71 94.12 89.92 1.65
Comments_YTB SentiStrength 96.94 89.62 93.28 38.24
Reviews_I SenticNet 97.39 93.66 95.52 69.41
Reviews_II SenticNet 94.15 93.87 94.01 94.25
Myspace SentiStrength 98.73 88.46 93.6 31.53
Amazon SentiStrength 93.85 79.38 86.62 19.58
Tweets_DBT Sentiment140 72.86 83.55 78.2 18.75
Tweets_RND_I SentiStrength 95.28 90.6 92.94 27.13
Tweets_RND_II VADER 99.31 98.45 98.88 94.4
Tweets_RND_III Sentiment140 97.57 95.9 96.73 50.77
Tweets_RND_IV Emoticons 94.74 86.76 88.6 58.27
Tweets_STF SentiStrength 95.76 94.81 95.29 41.78
Tweets_SAN SentiStrength 90.23 88.59 89.41 29.61
Tweets_Semeval SentiStrength 93.93 83.4 88.66 28.66
RW SentiStrength 90.04 75.79 82.92 23.12