Skip to main content

Table 3 The experiments of state-of-the-art methods on aspect term level ABSA on all Chinese datasets

From: Attention-based Sentiment Reasoner for aspect-based sentiment analysis

Model Camera Car Notebook Phone
Acc. F1 Acc. F1 Acc. F1 Acc. F1
LSTM 78.31 68.72 81.99 58.83 74.63 62.32 81.38 72.13
TD-LSTM 70.48 51.46 76.53 46.67 67.10 40.58 69.17 53.40
AT-LSTM 85.05 83.44 80.09 72.34 79.34 77.99 86.41 84.46
ATAE-LSTM 85.54 84.09 81.90 76.88 83.47 82.14 85.77 83.87
MemNN 70.59 55.13 75.55 51.01 69.10 53.51 70.29 55.93
ATAM-S 82.88 72.50 82.94 64.18 75.59 60.09 84.86 75.35
ATAM-F 88.30 82.94 77.52 88.46
AS-Reasoner 89.71 88.66 85.52 79.88 85.95 84.41 89.17 88.02
  1. Italic values indicate the significance of the best result
  2. The metric “Acc.” means the accuracy and “F1” denotes the macro-F1