Handling polysemous triggers and arguments in event extraction: an adaptive semantics learning strategy with reward–penalty mechanism

• Addresses the critical challenge of polysemy and semantic imbalance in event extraction triggers and arguments. • Introduces a reward–penalty mechanism that balances semantic distribution by rewarding correct classifications and penalizing incorrect ones. • Adds a sentence-level event situation awareness mechanism to enhance target event semantics. • Demonstrates state-of-the-art performance on ACE2005 and ERE datasets, outperforming single-task and multi-task baselines.