Learning to Map Context-Dependent Sentences to Executable Formal Queries
Learning to Map Context-Dependent Sentences to Executable Formal Queries
复制标题
DOI：
10.18653/v1/n18-1203
复制
发表时间：
2018-04
期刊：
影响因子：
--
通讯作者：
Alane Suhr;Srini Iyer;Yoav Artzi
中科院分区：
文献类型：
--
作者：
Alane Suhr;Srini Iyer;Yoav Artzi
我们提出了一个与上下文相关的模型，以将互动中的话语映射到可执行的正式查询中。为了合并交互历史记录，该模型维护了一个交互级的编码器，该编码器在每回合后都会更新，并且可以在生成过程中复制先前预测的查询的子序列。我们的方法结合了话语之间的隐式和明确的引用建模。我们在ATIS的飞行计划互动上评估了模型，并展示了建模上下文和明确参考的好处。
We propose a context-dependent model to map utterances within an interaction to executable formal queries. To incorporate interaction history, the model maintains an interaction-level encoder that updates after each turn, and can copy sub-sequences of previously predicted queries during generation. Our approach combines implicit and explicit modeling of references between utterances. We evaluate our model on the ATIS flight planning interactions, and demonstrate the benefits of modeling context and explicit references.