Table 2.

Mantel correlations between the extracted feature sets.

Feature Set	VADERa	Empath	LIWCb	Universal Sentence	Emoji 64	DeepMoji	GPTc default	GPT fine-tuned
VADER	1.000	0.003	0.098	0.453	0.211	0.422	0.430	0.429
Empath	0.003	1.000	0.009	0.006	−0.005	−0.008	0.004	0.001
LIWC	0.098	0.009	1.000	0.148	0.403	0.507	0.267	0.253
Universal Sentence	0.453	0.006	0.148	1.000	0.193	0.509	0.823	0.823
Emoji 64	0.211	−0.005	0.403	0.193	1.000	0.523	0.302	0.335
DeepMoji	0.422	−0.008	0.507	0.509	0.523	1.000	0.632	0.631
GPT default	0.430	0.004	0.267	0.823	0.302	0.632	1.000	0.799
GPT fine-tuned	0.429	0.001	0.253	0.823	0.335	0.631	0.799	1.000

^aVADER: Valence Aware Dictionary and sEntiment Reasoner.

^bLIWC: Linguistic Inquiry and Word Count.

^cGPT: Generative Pretrained Transformer.