<p>Document-Level Event Extraction (DEE) faces two persistent core challenges: long-range context comprehension and multi-argument role handling. To address these, we propose MSBPD, a novel model that jointly enhances contextual semantics and prompts information. MSBPD integrates three dedicated modules: (1) the Multi-scale Contextual Semantics Enhancement (MSCSE) module, which strengthens the modeling of cross-sentence dependencies by aggregating local, inter-sentential, global, and salient semantic features; (2) the Feature Fusion module, which dynamically aligns and integrates contextual representations with task-specific prompt information to enhance semantic interaction; and (3) the Bidirectional Parallel Prompt Decoding (BPPD) module, which introduces a bidirectional cross-attention mechanism and non-autoregressive matrix prediction to enable simultaneous extraction of all arguments, addressing the inefficiency of traditional sequential decoding. Extensive experiments on the RAMS and WikiEvents benchmarks demonstrate that MSBPD achieves competitive state-of-the-art performance. It outperforms the strong prompt-based baseline PAIE by 1.2% and 1.8% in Argument Classification F1 score on the two datasets, respectively. Ablation studies further validate the contribution of each module to the overall performance, confirming the effectiveness of our integrated approach for advancing Document-Level Event Extraction.</p>