References & Citations
Computer Science > Computation and Language
Title: DUMA: Reading Comprehension with Transposition Thinking
(Submitted on 26 Jan 2020 (v1), last revised 15 Sep 2020 (this version, v5))
Abstract: Multi-choice Machine Reading Comprehension (MRC) requires model to decide the correct answer from a set of answer options when given a passage and a question. Thus in addition to a powerful Pre-trained Language Model (PrLM) as encoder, multi-choice MRC especially relies on a matching network design which is supposed to effectively capture the relationships among the triplet of passage, question and answers. While the newer and more powerful PrLMs have shown their mightiness even without the support from a matching network, we propose a new DUal Multi-head Co-Attention (DUMA) model, which is inspired by human's transposition thinking process solving the multi-choice MRC problem: respectively considering each other's focus from the standpoint of passage and question. The proposed DUMA has been shown effective and is capable of generally promoting PrLMs. Our proposed method is evaluated on two benchmark multi-choice MRC tasks, DREAM and RACE, showing that in terms of powerful PrLMs, DUMA can still boost the model to reach new state-of-the-art performance.
Submission history
From: Pengfei Zhu [view email][v1] Sun, 26 Jan 2020 07:35:02 GMT (2335kb,D)
[v2] Wed, 5 Feb 2020 13:59:48 GMT (1990kb,D)
[v3] Sat, 8 Feb 2020 03:47:36 GMT (1990kb,D)
[v4] Wed, 18 Mar 2020 12:53:23 GMT (2459kb,D)
[v5] Tue, 15 Sep 2020 07:16:15 GMT (8748kb,D)
Link back to: arXiv, form interface, contact.