arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1907.03590cs.CLcs.AI
收藏

知识驱动的主动人机对话Agent的多生成模型扩充

Multiple Generative Models Ensemble for Knowledge-Driven Proactive Human-Computer Dialogue Agent

Zelin Dai, Weitang Liu, Guanhua Zhan

摘要:

采用多序列到序列模型,借助数据增强技术和多种编解码器结构设计,建立了一个端到端的多轮主动对话生成代理.为了提高性能,开发了一种基于秩的集成方法。结果表明,平均而言,我们的单一模型在DuConv数据集上的F1评分和BLEU方面比基线有18.67%的明显改善。特别是,集成方法进一步显着优于基线35.85%。

英文摘要:

Multiple sequence to sequence models were used to establish an end-to-end multi-turns proactive dialogue generation agent, with the aid of data augmentation techniques and variant encoder-decoder structure designs. A rank-based ensemble approach was developed for boosting performance. Results indicate that our single model, in average, makes an obvious improvement in the terms of F1-score and BLEU over the baseline by 18.67% on the DuConv dataset. In particular, the ensemble methods further significantly outperform the baseline by 35.85%.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑