使用深度正交范数分析提升对话系统的语篇层次理解

对话 agent 的演进 driven by the need for more contextually aware systems that can effectively manage dialogue over extended interactions。为 address existing models 在 capture and utilize long-term conversational history 的 limitations，我们提出了一种 novel framework， integrates Deep Canonical Correlation Analysis (DCCA) for discourse-level understanding。该 framework learns discourse tokens to capture relationships between utterances and their surrounding context， enabling a better understanding of long-term dependencies。Experiments on the Ubuntu Dialogue Corpus demonstrate significant enhancement in response selection， based on the improved automatic evaluation metric scores。结果表明，DCCA 在 improve dialogue systems by allowing them to filter out irrelevant context and retain critical discourse information for more accurate response retrieval 有潜力。