G2C generates draft quads then applies a single-shot sequence-level corrector trained on LLM-synthesized error patterns to outperform linear decoding baselines on Rest15 and Rest16.
Attention-based LSTM for aspect-level sentiment classification,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Generate Then Correct: Single Shot Global Correction for Aspect Sentiment Quad Prediction
G2C generates draft quads then applies a single-shot sequence-level corrector trained on LLM-synthesized error patterns to outperform linear decoding baselines on Rest15 and Rest16.