Repurposing Entailment for Multi-Hop Question Answering Tasks
Repurposing Entailment for Multi-Hop Question Answering Tasks
复制标题
DOI：
10.18653/v1/n19-1302
复制
发表时间：
2019-04
期刊：
影响因子：
--
通讯作者：
H. Trivedi;Heeyoung Kwon;Tushar Khot;Ashish Sabharwal;Niranjan Balasubramanian
中科院分区：
文献类型：
--
作者：
H. Trivedi;Heeyoung Kwon;Tushar Khot;Ashish Sabharwal;Niranjan Balasubramanian
问答自然地归结为蕴涵问题，即验证某个文本是否包含问题的答案。然而，对于需要用多个句子进行推理的多跳QA任务，如何最好地利用在SNLI等大规模数据集上预先训练的基于句子对的蕴涵模型仍不清楚。我们介绍了Multee，这是一个通用的体系结构，可以有效地将蕴涵模型用于多跳QA任务。Multee使用(I)帮助定位重要句子的本地模块，从而避免分散信息，以及(Ii)通过有效地结合重要性权重来聚合信息的全局模块。重要的是，我们证明了这两个模块都可以使用在大规模NLI数据集上预先训练的蕴涵函数。我们在两个多跳QA数据集MultiRC和OpenBookQA上进行了性能评估。当使用在NLI数据集上预先训练的蕴涵函数时，Multee的性能优于仅在目标QA数据集上训练的QA模型和OpenAI转换器模型。
Question Answering (QA) naturally reduces to an entailment problem, namely, verifying whether some text entails the answer to a question. However, for multi-hop QA tasks, which require reasoning with multiple sentences, it remains unclear how best to utilize entailment models pre-trained on large scale datasets such as SNLI, which are based on sentence pairs. We introduce Multee, a general architecture that can effectively use entailment models for multi-hop QA tasks. Multee uses (i) a local module that helps locate important sentences, thereby avoiding distracting information, and (ii) a global module that aggregates information by effectively incorporating importance weights. Importantly, we show that both modules can use entailment functions pre-trained on a large scale NLI datasets. We evaluate performance on MultiRC and OpenBookQA, two multihop QA datasets. When using an entailment function pre-trained on NLI datasets, Multee outperforms QA models trained only on the target QA datasets and the OpenAI transformer models.