The author presents a novel approach, Knowledge Injection and Guidance (KIG), to enhance Court View Generation using Pretrained Language Models (PLMs) by incorporating domain knowledge efficiently. The KIG method integrates claim-related knowledge through prompt tuning and a generating navigator to guide the generation process.
プレトレーニング言語モデルを使用して、法廷視点生成タスクを効果的に改善する方法を提案します。