LLMs for Fine-Grained Emotion Recognition in Reddit Posts and Comments on AI
- 5 February 2026
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Abstract
The present study is a computational analysis of Reddit posts and comments concerning users' opinions on AI technologies in terms of emotive words and their polarity, with regard to AI perceived advantages and disadvantages to society in different contexts of their occurrences. Thematic structures across the discussion layers were identified and a systematic analysis was performed of the relationship between agendasetting posts and their responses. Combining the complementary strengths of VADER and TextBlob with a confidence measure, we applied this multi-model approach for the successful portrayal of sentiment across social media discussions on AI. The strongest positive correlation between technology/job posts and work automation comments was identified with regard to labor displacement and other worries uncovering the users' tension between technological possibility and the human sense of insecurity. Further studies will elaborate on this finding and employ a more refined emotion panel.Keywords
This publication has 11 references indexed in Scilit:
- AI Trustworthiness and Persuasive Effects in Users’ Judgment2025
- KLLMs4Rec: Knowledge graph-enhanced LLMs sentiment extraction for personalized recommendationsExpert Systems with Applications, 2025
- Comparing large Language models and human annotators in latent content analysis of sentiment, political leaning, emotional intensity and sarcasmScientific Reports, 2025
- Opinion events and stance types: advances in LLM performance with ChatGPT and GeminiLodz Papers in Pragmatics, 2024
- Opinion Identification using a Conversational Large Language ModeThe International FLAIRS Conference Proceedings, 2024
- Opinion Events: Types and opinion markers in English social media discourseLodz Papers in Pragmatics, 2023
- GoEmotions: A Dataset of Fine-Grained Emotions2020
- VADER: A Parsimonious Rule-Based Model for Sentiment Analysis of Social Media TextProceedings of the International AAAI Conference on Web and Social Media, 2014
- A psychoevolutionary theory of emotionsSocial Science Information, 1982
- DIMENSIONALITY OF THE SEMANTIC SPACE FOR COMMUNICATION VIA FACIAL EXPRESSIONSScandinavian Journal of Psychology, 1966