A GPT-4o pipeline generates about 8,300 synthetic, sentiment-balanced ABSA conversations across four domains, plus a three-LLM benchmark with distinct precision, recall, and latency tradeoffs.
SemEval-2016 task 5: Aspect based sentiment analysis,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Multi-Domain ABSA Conversation Dataset Generation via LLMs for Real-World Evaluation and Model Comparison
A GPT-4o pipeline generates about 8,300 synthetic, sentiment-balanced ABSA conversations across four domains, plus a three-LLM benchmark with distinct precision, recall, and latency tradeoffs.