Bootstrapping incremental dialogue systems from minimal data: the generalisation power of dialogue grammars
Bootstrapping incremental dialogue systems from minimal data: the generalisation power of dialogue grammars
复制标题
DOI：
10.18653/v1/d17-1236
复制
发表时间：
2017-09
期刊：
影响因子：
--
通讯作者：
Arash Eshghi;Igor Shalyminov;Oliver Lemon
中科院分区：
文献类型：
--
作者：
Arash Eshghi;Igor Shalyminov;Oliver Lemon
我们研究了一种端到端方法，用于从少量未注释的对话数据中自动引入基于任务的对话系统。它将增量语义语法 - 动态语法和记录类型理论 (DS-TTR) - 与强化学习 (RL) 相结合，其中语言生成和对话管理是一个联合决策问题。由此产生的系统是渐进式的：对话是逐字处理的，之前的研究表明这对于支持自然、自发的对话至关重要。我们假设语法中丰富的语言知识应该能够处理大量的组合对话变体，即使是在很少的对话上进行训练。我们的实验表明，即使仅使用 0.13% 的数据（5 个对话）进行训练，我们的模型也可以处理 74% 的 Facebook AI bAbI 数据集。它还可以处理 bAbI+ 的 65%，bAbI+ 是我们通过系统地向 bAbI 添加重启和自我更正等增量对话现象而创建的语料库。我们将我们的模型与最先进的检索模型 MEMN2N 进行比较。我们发现，就语义准确性而言，即使在完整的 bAbI 数据集上进行训练，MEMN2N 模型对 bAbI+ 转换的鲁棒性也非常差。
We investigate an end-to-end method for automatically inducing task-based dialogue systems from small amounts of unannotated dialogue data. It combines an incremental semantic grammar - Dynamic Syntax and Type Theory with Records (DS-TTR) - with Reinforcement Learning (RL), where language generation and dialogue management are a joint decision problem. The systems thus produced are incremental: dialogues are processed word-by-word, shown previously to be essential in supporting natural, spontaneous dialogue. We hypothesised that the rich linguistic knowledge within the grammar should enable a combinatorially large number of dialogue variations to be processed, even when trained on very few dialogues. Our experiments show that our model can process 74% of the Facebook AI bAbI dataset even when trained on only 0.13% of the data (5 dialogues). It can in addition process 65% of bAbI+, a corpus we created by systematically adding incremental dialogue phenomena such as restarts and self-corrections to bAbI. We compare our model with a state-of-the-art retrieval model, MEMN2N. We find that, in terms of semantic accuracy, the MEMN2N model shows very poor robustness to the bAbI+ transformations even when trained on the full bAbI dataset.