來自 | 知乎 作者 | 李家琦

鏈接｜https://zhuanlan.zhihu.com/p/410984053

本文已獲作者授權(quán)，未經(jīng)許可禁止二次轉(zhuǎn)載

Dialogue-based Machine Reading Comprehension任務(wù)是近兩年比較新的機(jī)器閱讀理解（MRC）任務(wù)，任務(wù)目標(biāo)是讓機(jī)器去理解人們之間的對(duì)話。本文簡(jiǎn)要整理了該任務(wù)現(xiàn)有數(shù)據(jù)集，并推薦幾篇相關(guān)論文。

一、數(shù)據(jù)集

該任務(wù)現(xiàn)有的數(shù)據(jù)集主要有如下這些：

1. Ma et, al, 2018, NAACL（數(shù)據(jù)集沒有命名）

任務(wù)類型：完形填空

論文：Challenging reading comprehension on daily conversation: Passage completion on multiparty dialog.

數(shù)據(jù)集：GitHub - emorynlp/reading-comprehension: Reading comprehension on multiparty dialog.

2.?DREAM, TACL 2019

任務(wù)類型：單選題

論文：Dream: A challenge data set and models for dialogue-based reading comprehension.

數(shù)據(jù)集：A Challenge Dataset and Models for Dialogue-Based Reading Comprehension

3.?FriendsQA, SIGDial 2019

任務(wù)類型：Span-base

論文：FriendsQA: Open-Domain Question Answering on TV Show Transcripts

數(shù)據(jù)集：GitHub - emorynlp/FriendsQA: Question answering on multiparty dialogue

4.?Molweni，COLING 2020

任務(wù)類型：Span-based

論文：Molweni: A Challenge Multiparty Dialogue-based Machine Reading Comprehension Dataset with Discourse Structure

數(shù)據(jù)集：GitHub - HIT-SCIR/Molweni

5.?QAConv, arXiv 2021

任務(wù)類型：Span-based

論文：QAConv: Question Answering on Informative Conversations

數(shù)據(jù)集：GitHub - salesforce/QAConv: This repository maintains the QAConv dataset, a question-answering dataset on informative conversations including business emails, panel discussions, and work channels.

目前此任務(wù)上使用比較多的數(shù)據(jù)集主要是DREAM、FriendsQA和Molweni。在QAConv數(shù)據(jù)集論文中，作者還將現(xiàn)有的幾個(gè)數(shù)據(jù)集進(jìn)行了對(duì)比。

數(shù)據(jù)集對(duì)比，來自QAConv論文

二、模型

這部分主要推薦DREAM、FriendsQA和Molweni這3個(gè)數(shù)據(jù)集上比較有代表性的模型論文。

1. DREAM數(shù)據(jù)集相關(guān)模型論文推薦

a.?DUMA: Reading Comprehension with Transposition Thinking. arXiv 2020.

b.?Multi-task Learning with Multi-head Attention for Multi-choice Reading Comprehension. arXiv 2020.

2.?FriendsQA數(shù)據(jù)集相關(guān)模型論文推薦

a.?Transformers to Learn Hierarchical Contexts in Multiparty Dialogue for Span-based Question Answering. ACL 2020.

b.?Graph-based knowledge integration for question answering over dialogue. COLING 2020.

3.?Molweni數(shù)據(jù)集相關(guān)模型論文推薦

a.?DADgraph: A Discourse-aware Dialogue Graph Neural Network for Multiparty Dialogue Machine Reading Comprehension. IJCNN 2021.

b.?Self-and Pseudo-self-supervised Prediction of Speaker and Key-utterance for Multi-party Dialogue Reading Comprehension. EMNLP 2021 Findings.

c.?Enhanced Speaker-aware Multi-party Multi-turn Dialogue Comprehension. arXiv 2021.

以上是我簡(jiǎn)單整理的Dialogue MRC任務(wù)數(shù)據(jù)集和推薦的幾篇相關(guān)論文，歡迎補(bǔ)充！

往期精彩回顧適合初學(xué)者入門人工智能的路線及資料下載機(jī)器學(xué)習(xí)及深度學(xué)習(xí)筆記等資料打印機(jī)器學(xué)習(xí)在線手冊(cè)深度學(xué)習(xí)筆記專輯《統(tǒng)計(jì)學(xué)習(xí)方法》的代碼復(fù)現(xiàn)專輯 AI基礎(chǔ)下載黃海廣老師《機(jī)器學(xué)習(xí)課程》視頻課黃海廣老師《機(jī)器學(xué)習(xí)課程》711頁完整版課件

本站qq群554839127，加入微信群請(qǐng)掃碼：

總結(jié)

以上是生活随笔為你收集整理的【NLP】面向对话的机器阅读理解任务（Dialogue MRC）相关论文整理的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。