[Retracted] Text Sentiment Analysis of German Multilevel Features Based on Self-Attention Mechanism

Table 3

Comparison of results for various sentence-level features.

Features	Kendall	Pearson	Spearman
Nothing	—	—	—	62.11	63.28
Bert vector	—	—	—	63.85	65.18
English translation Vader affective value	0.3836	0.4727	0.4683	63.16	64.77
Number of forward emoticons	0.3493	0.3472	0.3729	62.98	64.55
End bracket direction	0.3207	0.3410	0.3411	62.90	64.53
Lentil 40 emotional value	0.2628	0.2974	0.2931	62.58	64.16
Number of negative emoticons	−0.0651	−0.1739	−0.1749	62.35	63.94
Total exclamation marks	0.0523	−0.0260	−0.0260	62.24	63.78
Total number of question marks	−0.0246	—	—	62.24	63.51