电子书阅读地址：https://windmissing.github.io/NLP-important-papers/
1. 论文来源：
https://github.com/terryum/awesome-deep-learning-papers/blob/master/README.md
https://paperswithcode.com/
https://github.com/yizhen20133868/NLP-Conferences-Code
https://www.jianshu.com/p/d80b065bdcf0
https://www.aclweb.org/anthology/events/acl-2020/
2. 表格说明
- 20%
阅读：title, abstract, introduction
回答问题：要解决什么问题，用了什么方法，作者主要贡献，得到了什么结果
- 40%
阅读：related work, figure, conclusion
回答问题：已有的有哪些方法，baseline是什么，本文的创新点在哪里，得到了什么结果
- 60%
阅读：全文，跳过公式、技巧、术语、定义
- 80%
阅读：全文
- 100%
阅读：相关文献、代码
3. 监督学习
3.1. 命令体识别
| papter list | 20% | 40% | 60% | 80% | 100% |
|---|---|---|---|---|---|
| [2016 terryum] Neural Architectures for Named Entity Recognition, G. Lample et al. [pdf] |
3.2. 语言模型
| papter list | 20% | 40% | 60% | 80% | 100% |
|---|---|---|---|---|---|
| [2016 terryum] Exploring the limits of language modeling, R. Jozefowicz et al. [pdf] |
3.3. 问答系统QA和阅读理解MRC
| papter list | 20% | 40% | 60% | 80% | 100% |
|---|---|---|---|---|---|
| [2015 terryum] Teaching machines to read and comprehend, K. Hermann et al. [pdf] | |||||
| [2020 ACL] Probabilistic Assumptions Matter: Improved Models for Distantly-Supervised Document-Level Question Answering | |||||
| [2020 ACL] Contextualized Sparse Representations for Real-Time Open-Domain Question Answering | |||||
| [2020 ACL] Recurrent Chunking Mechanisms for Long-Text Machine Reading Comprehension | |||||
| [2020 ACL] Document Modeling with Graph Attention Networks for Multi-grained Machine Reading Comprehension | |||||
| [2020 ACL] Improving Multi-hop Question Answering over Knowledge Graphs using Knowledge Base Embeddings |
4. 非监督学习
4.1. 问答系统QA和阅读理解MRC
| papter list | 20% | 40% | 60% | 80% | 100% |
|---|---|---|---|---|---|
| [2020 ACL] Harvesting and Refining Question-Answer Pairs for Unsupervised QA pdf code | |||||
| [2020 ACL] Template-Based Question Generation from Retrieved Sentences for Improved Unsupervised Question Answering | |||||
| [2020 ACL] Unsupervised Alignment-based Iterative Evidence Retrieval for Multi-hop Question Answering |
4.2. 机器翻译
| papter list | 20% | 40% | 60% | 80% | 100% |
|---|---|---|---|---|---|
| [2015 terryum] Effective approaches to attention-based neural machine translation, M. Luong et al. [pdf] |
5. 未分类papter list
| papter list | 20% | 40% | 60% | 80% | 100% |
|---|---|---|---|---|---|
| [2015 terryum] Conditional random fields as recurrent neural networks, S. Zheng and S. Jayasumana. [pdf] |