├── README.md └── rules.md /README.md: -------------------------------------------------------------------------------- 1 | # Data Augmentation Techniques for NLP 2 | 3 | 4 | If you'd like to add your paper, do not email us. Instead, read the protocol for [adding a new entry](https://github.com/styfeng/DataAug4NLP/blob/main/rules.md) and send a pull request. 5 | 6 | We group the papers by [text classification](#text-classification), [translation](#translation), [summarization](#summarization), [question-answering](#question-answering), [sequence tagging](#sequence-tagging), [parsing](#parsing), [grammatical-error-correction](#grammatical-error-correction), [generation](#generation), [dialogue](#dialogue), [multimodal](#multimodal), [mitigating bias](#mitigating-bias), [mitigating class imbalance](#mitigating-class-imbalance), [adversarial examples](#adversarial-examples), [compositionality](#compositionality), and [automated augmentation](#automated-augmentation). 7 | 8 | This repository is based on our paper, ["A survey of data augmentation approaches in NLP (Findings of ACL '21)"](https://aclanthology.org/2021.findings-acl.84/). You can cite it as follows: 9 | ``` 10 | @inproceedings{feng-etal-2021-survey, 11 | title = "A Survey of Data Augmentation Approaches for {NLP}", 12 | author = "Feng, Steven Y. and 13 | Gangal, Varun and 14 | Wei, Jason and 15 | Chandar, Sarath and 16 | Vosoughi, Soroush and 17 | Mitamura, Teruko and 18 | Hovy, Eduard", 19 | booktitle = "Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021", 20 | month = aug, 21 | year = "2021", 22 | address = "Online", 23 | publisher = "Association for Computational Linguistics", 24 | url = "https://aclanthology.org/2021.findings-acl.84", 25 | doi = "10.18653/v1/2021.findings-acl.84", 26 | pages = "968--988", 27 | } 28 | ``` 29 | Authors: Steven Y. Feng, 30 | Varun Gangal, 31 | Jason Wei, 32 | Sarath Chandar, 33 | Soroush Vosoughi, 34 | Teruko Mitamura, 35 | Eduard Hovy 36 | 37 | Special thanks to Ryan Shentu, Fiona Feng, Karen Liu, Emily Nie, Tanya Lu, and Bonnie Ma for helping out with this repo. 38 | Note: WIP. More papers will be added from our survey paper to this repo soon. 39 | Inquiries should be directed to stevenyfeng@gmail.com or by opening an issue here. 40 | 41 | Also, check out our **talk for Google Research** (Steven Feng and Varun Gangal) [here](https://www.youtube.com/watch?v=kNBVesKUZCk&ab_channel=StevenFeng), and our **podcast episode** (Steven Feng and Eduard Hovy) [here](https://www.youtube.com/watch?v=qmqyT_97Poc) and [here](https://thedataexchange.media/data-augmentation-in-natural-language-processing/). 42 | 43 | 44 | ### Text Classification 45 | | Paper | Datasets | 46 | | -- | --- | 47 | | Unsupervised Word Sense Disambiguation Rivaling Supervised Methods ([ACL '95](https://www.aclweb.org/anthology/P95-1026.pdf)) | Paper-Specific/Legacy Corpus | 48 | | Synonym Replacement (Character-Level Convolutional Networks for Text Classification, [NeurIPS '15](https://papers.nips.cc/paper/2015/file/250cf8b51c773f3f8dc8b4be867a9a02-Paper.pdf)) | AG’s News, DBPedia, Yelp, Yahoo Answers, Amazon | 49 | | That’s So Annoying!!!: A Lexical and Frame-Semantic Embedding Based Data Augmentation Approach to Automatic Categorization of Annoying Behaviors using #petpeeve Tweets [(EMNLP '15)](https://www.aclweb.org/anthology/D15-1306.pdf) | twitter| 50 | | Robust Training under Linguistic Adversity [(EACL '17)](https://www.aclweb.org/anthology/E17-2004/) [code](https://github.com/lrank/Linguistic_adversity) | Movie review, customer review, SUBJ, SST | 51 | | Contextual Augmentation: Data Augmentation by Words with Paradigmatic Relations [(NAACL '18)](https://www.aclweb.org/anthology/N18-2072.pdf) [code](https://github.com/pfnet-research/contextual_augmentation) | SST, SUBJ, MRQA, RT, TREC | 52 | | Variational Pretraining for Semi-supervised Text Classification [(ACL '19)](https://www.aclweb.org/anthology/P19-1590.pdf) [code](http://github.com/allenai/vampire) | IMDB, AG News, Yahoo, hatespeech | 53 | | EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks [(EMNLP '19)](http://dx.doi.org/10.18653/v1/D19-1670) [code](https://github.com/jasonwei20/eda_nlp) | SST, CR, SUBJ, TREC, PC | 54 | | A Closer Look At Feature Space Data Augmentation For Few-Shot Intent Classification [(DeepLo @ EMNLP '19)](https://arxiv.org/abs/1910.04176) | SNIPS | 55 | | Nonlinear Mixup: Out-Of-Manifold Data Augmentation for Text Classification [(AAAI '20)](https://doi.org/10.1609/aaai.v34i04.5822) | TREC, SST, Subj, MR | 56 | | MixText: Linguistically-Informed Interpolation of Hidden Space for Semi-Supervised Text Classification [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.194/) [code](https://github.com/GT-SALT/MixText) | AG News, DBpedia, Yahoo, IMDb | 57 | | Unsupervised Data Augmentation for Consistency Training [(NeurIPS '20)](https://papers.nips.cc/paper/2020/hash/44feb0096faa8326192570788b38c1d1-Abstract.html) [code](https://papers.nips.cc/paper/2020/hash/44feb0096faa8326192570788b38c1d1-Abstract.html) | Yelp, IMDb, amazon, DBpedia | 58 | | Not Enough Data? Deep Learning to the Rescue! [(AAAI '20)](https://arxiv.org/abs/1911.03118) | ATIS, TREC, WVA | 59 | | Data Augmentation using Pre-trained Transformer Models [LifeLongNLP @ AACL '20](https://arxiv.org/abs/2003.02245), [code](https://github.com/varunkumar-dev/TransformersDataAugmentation) |SNIPS, TREC, SST2 | 60 | | SSMBA: Self-Supervised Manifold Based Data Augmentation for Improving Out-of-Domain Robustness [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.97/) [code](https://github.com/nng555/ssmba) | IWSLT'14 | 61 | | Data Boost: Text Data Augmentation Through Reinforcement Learning Guided Conditional Generation [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.726/) | ICWSM 20’ Data Challenge, SemEval '17 sentiment analysis, SemEval '18 irony | 62 | | Textual Data Augmentation for Efficient Active Learning on Tiny Datasets [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.600/) | SST2, TREC | 63 | | Text Augmentation in a Multi-Task View [(EACL '21)](https://www.aclweb.org/anthology/2021.eacl-main.252/) | SST2, TREC, SUBJ | 64 | | GPT3Mix: Leveraging Large-scale Language Models for Text Augmentation [(arXiv '21)](https://arxiv.org/abs/2104.08826) | SST2, CR, TREC, SUBJ, MPQA, CoLA | 65 | | Few-Shot Text Classification with Triplet Loss, Data Augmentation, and Curriculum Learning [(NAACL '21)](https://arxiv.org/abs/2103.07552) [code](https://github.com/jasonwei20/triplet-loss) | HUFF, COV-Q, AMZN, FEWREL | 66 | | Text AutoAugment: Learning Compositional Augmentation Policy for Text Classification [(EMNLP '21)](https://arxiv.org/abs/2109.00523) [code](https://github.com/lancopku/text-autoaugment) | IMDB, SST2, SST5, TREC, YELP2, YELP5 | 67 | | AEDA: An Easier Data Augmentation Technique for Text Classification [(EMNLP '21)](https://arxiv.org/abs/2108.13230) [code](https://github.com/akkarimi/aeda_nlp) | SST, CR, SUBJ, TREC, PC | 68 | 69 | ### Translation 70 | 71 | | Paper | Datasets | 72 | | -- | --- | 73 | | Backtranslation (Improving Neural Machine Translation Models with Monolingual Data, [ACL '16](https://www.aclweb.org/anthology/P16-1009.pdf)) | WMT '15 en-de, IWSLT '15 en-tr | 74 | | Adapting Neural Machine Translation with Parallel Synthetic Data [(WMT '17)](https://www.aclweb.org/anthology/W17-4714/) | COMMON, 1 Billion Words, dev2013, XRCE, IT, E-Com| 75 | | Data Augmentation for Low-Resource Neural Machine Translation [(ACL '17)](https://www.aclweb.org/anthology/P17-2090/) [code](https://github.com/marziehf/DataAugmentationNMT) | WMT '14/'15/'16 en-de/de-en| 76 | | Synthetic Data for Neural Machine Translation of Spoken-Dialects [(arxiv '17)](https://arxiv.org/abs/1707.00079) | LDC2012T09, OpenSubtitles-2013| 77 | | Multi-Source Neural Machine Translation with Data Augmentation [(IWSLT '18)](https://arxiv.org/abs/1810.06826) | TED Talks| 78 | | SwitchOut: an Efficient Data Augmentation Algorithm for Neural Machine Translation [(EMNLP '18)](https://www.aclweb.org/anthology/D18-1100/) | IWSLT '15 en-vi, IWSLT '16 de-en, WMT '15 en-de | 79 | | Generalizing Back-Translation in Neural Machine Translation [(WMT '19)](https://www.aclweb.org/anthology/W19-5205/) | ed NewsCrawl2, WMT'18 de-en| 80 | | Neural Fuzzy Repair: Integrating Fuzzy Matches into Neural Machine Translation [(ACL '19)](https://www.aclweb.org/anthology/P19-1175/) | DGT-TM en-ml/en-hu| 81 | | Augmenting Neural Machine Translation with Knowledge Graphs [(arxiv '19)](https://arxiv.org/abs/1902.08816) | WMT '14 -'18| 82 | | Generalized Data Augmentation for Low-Resource Translation [(ACL '19)](https://www.aclweb.org/anthology/P19-1579/) [code](https://github.com/xiamengzhou/DataAugForLRL)| ENG-HRL-LRL, HRL-LRL | 83 | | Improving Robustness of Machine Translation with Synthetic Noise [(NAACL '19)](https://www.aclweb.org/anthology/N19-1190/) [code](https://github.com/MysteryVaibhav/robust_mtnt)| EP, TED, MTNT en-fr en-jpn| 84 | | Soft Contextual Data Augmentation for Neural Machine Translation [(ACL '19)](https://www.aclweb.org/anthology/P19-1555/) [code](https://github.com/teslacool/SCA) | IWSLT '14 de/es/he-en, WMT '14 en-de | 85 | | Data augmentation using back-translation for context-aware neural machine translation [(DiscoMT @ EMNLP '19)](https://www.aclweb.org/anthology/D19-6504/) [code](https://github.com/sugi-a/discomt2019) | IWSLT'17 en-ja/en-fr, BookCorpus, Europarl v7, National Diet of Japan | 86 | | Improving Neural Machine Translation Robustness via Data Augmentation: Beyond Back-Translation [(W-NUT @ EMNLP '19)](https://www.aclweb.org/anthology/D19-5543/) | WMT'15/'19 en/fr, MTNT, IWSLT'17, MuST-C | 87 | | Data augmentation for pipeline-based speech translation [(Baltic HLT '20)](https://hal.inria.fr/hal-02907053) | WMT '17 | 88 | | Lexical-Constraint-Aware Neural Machine Translation via Data Augmentation [(IJCAI '20)](https://www.ijcai.org/proceedings/2020/496) [code](https://github.com/ghchen18/leca) | WMT '16 de-en, NIST zh-en | 89 | | A Diverse Data Augmentation Strategy for Low-Resource Neural Machine Translation [(Information '20)](https://www.mdpi.com/2078-2489/11/5/255) | IWSLT '14 en-de | 90 | | Syntax-aware Data Augmentation for Neural Machine Translation [(arxiv '20)](https://arxiv.org/abs/2004.14200) | WMT '14 en-de, IWSLT '14 de-en | 91 | | SSMBA: Self-Supervised Manifold Based Data Augmentation for Improving Out-of-Domain Robustness [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.97/) [code](https://github.com/nng555/ssmba) | IWSLT'14 | 92 | | Data diversification: A simple strategy for neural machine translation [(NeurIPS '20)](https://proceedings.neurips.cc/paper/2020/file/7221e5c8ec6b08ef6d3f9ff3ce6eb1d1-Paper.pdf) [code](https://github.com/nxphi47/data_diversification) | WMT '14 en-de/en-fr, IWSLT '13/'14/'15 en-de/de-en/en-fr | 93 | | AdvAug: Robust Adversarial Augmentation for Neural Machine Translation [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.529/) | NIST zh-en, WMT '14 en-de| 94 | | Dictionary-based Data Augmentation for Cross-Domain Neural Machine Translation [(arxiv '20)](https://arxiv.org/abs/2004.02577) | WMT '14/'19 | 95 | | Sentence Boundary Augmentation For Neural Machine Translation Robustness [(arxiv '20)](https://arxiv.org/abs/2010.11132) | IWSLT '14/'15/'18 en-de, WMT '18 en-de | 96 | | Valar nmt : Vastly lacking resources neural machine translation [(Stanford CS224N)](https://web.stanford.edu/class/archive/cs/cs224n/cs224n.1194/reports/custom/15811193.pdf) | Bible, Misc, Europarl v8, Newstest '18 | 97 | 98 | 99 | ### Summarization 100 | 101 | | Paper | Datasets | 102 | | -- | --- | 103 | | Transforming Wikipedia into Augmented Data for Query-Focused Summarization [(arxiv '19)](https://arxiv.org/abs/1911.03324) | DUC | 104 | | Iterative Data Augmentation with Synthetic Data (Abstract Text Summarization: A Low Resource Challenge [(EMNLP '19)](https://www.aclweb.org/anthology/D19-1616/) | Swisstext, commoncrawl | 105 | | Improving Zero and Few-Shot Abstractive Summarization with Intermediate Fine-tuning and Data Augmentation [(NAACL '21)](https://arxiv.org/abs/2010.12836) | CNN-DailyMail | 106 | | Data Augmentation for Abstractive Query-Focused Multi-Document Summarization [(AAAI '21)](https://arxiv.org/abs/2103.01863) [code](https://github.com/ramakanth-pasunuru/QmdsCnnIr) | QMDSCNN, QMDSIR, WikiSum, DUC 2006, DUC 2007 | 107 | 108 | 109 | ### Question Answering 110 | 111 | | Paper | Datasets | 112 | | -- | --- | 113 | | QANet: Combining Local Convolution with Global Self-Attention for Reading Comprehension [(ICLR '18)](https://openreview.net/forum?id=B14TlG-RW) | SQuAD, TriviaQA | 114 | | An Exploration of Data Augmentation and Sampling Techniques for Domain-Agnostic Question Answering [(EMNLP '19 Workshop)](https://www.aclweb.org/anthology/D19-5829/) | MRQA | 115 | | Data Augmentation for BERT Fine-Tuning in Open-Domain Question Answering [(arxiv '19)](https://arxiv.org/abs/1904.06652) | SQuAD, Trivia-QA, CMRC, DRCD | 116 | | XLDA: Cross-Lingual Data Augmentation for Natural Language Inference and Question Answering [(arxiv '19)](https://openreview.net/forum?id=BJgAf6Etwr) | XNLI, SQuAD | 117 | | Synthetic Data Augmentation for Zero-Shot Cross-Lingual Question Answering [(arxiv '20)](https://arxiv.org/abs/2010.12643) | MLQA, XQuAD, SQuAD-it, PIAF | 118 | | Logic-Guided Data Augmentation and Regularization for Consistent Question Answering [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.499/) [code](https://github.com/AkariAsai/logic_guided_qa) | WIQA, QuaRel, HotpotQA | 119 | 120 | 121 | ### Sequence Tagging 122 | 123 | | Paper | Datasets | 124 | | -- | --- | 125 | | Data Augmentation via Dependency Tree Morphing for Low-Resource Languages [(EMNLP '18)](https://www.aclweb.org/anthology/D18-1545.pdf) [code](https://github.com/gozdesahin/crop-rotate-augment) | universal dependencies project | 126 | | DAGA: Data Augmentation with a Generation Approach for Low-resource Tagging Tasks [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.488/) [code](https://github.com/ntunlp/daga) | CoNLL2002/2003 | 127 | | An Analysis of Simple Data Augmentation for Named Entity Recognition [(COLING '20)](https://www.aclweb.org/anthology/2020.coling-main.343/) | MaSciP, i2b2- 2010 | 128 | | SeqMix: Augmenting Active Sequence Labeling via Sequence Mixup [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.691/) [code](https://github.com/rz-zhang/SeqMix) | CoNLL-03, ACE05, Webpage | 129 | 130 | 131 | ### Parsing 132 | | Paper | Datasets | 133 | | -- | --- | 134 | | Data Recombination for Neural Semantic Parsing [(ACL '16)](https://www.aclweb.org/anthology/P16-1002/) [code](https://github.com/dongpobeyond/Seq2Act) | GeoQuery, ATIS, Overnight | 135 | | A systematic comparison of methods for low-resource dependency parsing on genuinely low-resource languages [(EMNLP '19)](https://www.aclweb.org/anthology/D19-1102/) | Universal Dependencies treebanks version 2.2 | 136 | | Named Entity Recognition for Social Media Texts with Semantic Augmentation [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.107/)[code](https://github.com/cuhksz-nlp/SANER) | WNUT16, WNUT17, Weibo | 137 | | Good-Enough Compositional Data Augmentation [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.676/) [code](https://github.com/jacobandreas/geca) | SCAN | 138 | | GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing [(ICLR '21)](https://openreview.net/forum?id=kyaIeYj4zZ) | SPIDER, WIKISQL, WIKITABLEQUESTIONS | 139 | 140 | 141 | ### Grammatical Error Correction 142 | | Paper | Datasets | 143 | | -- | --- | 144 | | GenERRate: Generating Errors for Use in Grammatical Error Detection [(BEA '09)](https://www.aclweb.org/anthology/W09-2112/) | Ungram-BNC | 145 | | Mining Revision Log of Language Learning SNS for Automated Japanese Error Correction of Second Language Learners [(IJCNLP '11)](https://www.aclweb.org/anthology/I11-1017/) [code](https://github.com/google-research-datasets/clang8) | Lang-8 | 146 | | Artificial error generation for translation-based grammatical error correction [(University of Cambridge Technical Report '16)](https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-895.pdf) | Several Datasets | 147 | | Noising and Denoising Natural Language: Diverse Backtranslation for Grammar Correction. [(NAACL'18)](https://www.aclweb.org/anthology/N18-1057/) | Lang-8, CoNLL-2014, CoNLL-2013, JFLEG | 148 | | Using Wikipedia Edits in Low Resource Grammatical Error Correction. [(WNUT @ EMNLP '18)](https://doi.org/10.18653/v1/W18-6111) | Falko-MERLIN GEC Corpus | 149 | | Sequence-to-sequence Pre-training with Data Augmentation for Sentence Rewriting [(arxiv '19)](https://arxiv.org/abs/1909.06002) | CoNLL-2014 , JFLEG | 150 | | Controllable Data Synthesis Method for Grammatical Error Correction [(arxiv '19)](https://arxiv.org/abs/1909.13302) [code](https://github.com/marumalo/survey/issues/21) | NUCLE, Lang-8, One-Billion, CoNLL2013, CoNLL2014| 151 | | Neural Grammatical Error Correction Systems with Unsupervised Pre-training on Synthetic Data. [(BEA @ ACL '19)](https://doi.org/10.18653/v1/W19-4427) | FCE, NUCLE, W&I+LOCNESS, Lang-8 | 152 | | Corpora Generation for Grammatical Error Correction [(NAACL'19)](https://doi.org/10.18653/v1/N19-1333) | CoNLL-2014, JFLEG, Lang-8 | 153 | | Erroneous data generation for Grammatical Error Correction [(BEA @ ACL '19)](https://www.aclweb.org/anthology/W19-4415/) | Lang-8,n CoNLL, JFLEG, CoNLL-2014, ABCN, FCE | 154 | | Sequence-to-sequence Pre-training with Data Augmentation for Sentence Rewriting [(arxiv '19)](https://arxiv.org/abs/1909.06002) [code](https://github.com/marumalo/survey/issues/6) | GYAFC, WMT14, WMT18 | 155 | | A neural grammatical error correction system built on better pre-training and sequential transfer learning. [(BEA @ ACL '19)](https://doi.org/10.18653/v1/W19-4423) | FCE, NUCLE, W&I+LOCNESS, Lang-8, Gutenberg, Tatoeba, WikiText-103 | 156 | | Improving Grammatical Error Correction with Data Augmentation by Editing Latent Representation [(COLING'20)](https://doi.org/10.18653/v1/2020.coling-main.200) | FCE, NUCLE, W&I+LOCNESS, Lang-8 | 157 | | A Comparative Study of Synthetic Data Generation Methods for Grammatical Error Correction [(BEA @ ACL '20)](https://www.aclweb.org/anthology/2020.bea-1.21/) | W&I+LOCNESS, FCE, News Crawl 2, W&I+L train, FCE-train, NUCLE, Lang-8, W&I+L dev, FCE-test, Tatoeba, WikiText-103 | 158 | | A syntactic rule-based framework for parallel data synthesis in Japanese GEC [(MIT Thesis '20)](https://dspace.mit.edu/handle/1721.1/127416) | Lang-8 | 159 | 160 | 161 | ### Generation 162 | 163 | | Paper | Datasets | 164 | | -- | --- | 165 | | TNT-NLG, System 2: Data repetition and meaning representation manipulation to improve neural generation [(E2E NLG Challenge System Descriptions)](http://www.macs.hw.ac.uk/InteractionLab/E2E/final_papers/E2E-TNT_NLG2.pdf) | TODO | 166 | | Findings of the Third Workshop on Neural Generation and Translation [(WNGT @ EMNLP '19)](https://www.aclweb.org/anthology/D19-5601/) | RotoWire English-German | 167 | | A Good Sample is Hard to Find: Noise Injection Sampling and Self-Training for Neural Language Generation Models [(INLG '19)](https://www.aclweb.org/anthology/W19-8672/) [code](https://github.com/kedz/noiseylg) | E2E Challenge Dataset, Laptops, TVs | 168 | | GenAug: Data Augmentation for Finetuning Text Generators [(DeeLIO @ EMNLP '20)](https://www.aclweb.org/anthology/2020.deelio-1.4/) [code](https://github.com/styfeng/GenAug) | Yelp | 169 | | Denoising Pre-Training and Data Augmentation Strategies for Enhanced RDF Verbalization with Transformers [(WebNLG+ @ INLG '20)](https://www.aclweb.org/anthology/2020.webnlg-1.9/) | WebNLG | 170 | 171 | 172 | ### Dialogue 173 | | Paper | Datasets | 174 | | -- | --- | 175 | | Sequence-to-Sequence Data Augmentation for Dialogue Language Understanding [(COLING '18)](https://www.aclweb.org/anthology/C18-1105/) [code](https://github.com/AtmaHou/Seq2SeqDataAugmentationForLU) | ATIS, Dec94, Stanford dialogue | 176 | | Task-Oriented Dialog Systems that Consider Multiple Appropriate Responses under the Same Context [(arxiv '19)](https://arxiv.org/abs/1911.10484) [code](https://github.com/thu-spmi/damd-multiwoz) | MultiWOZ | 177 | | Data Augmentation by Data Noising for Open-vocabulary Slots in Spoken Language Understanding [(Student Research Workshop @ NAACL '19)](https://www.aclweb.org/anthology/N19-3014/) | ATIS, Snips, MR | 178 | | Data Augmentation with Atomic Templates for Spoken Language Understanding [(EMNLP '19)](https://www.aclweb.org/anthology/D19-1375/) [code](https://github.com/sz128/DAAT_SLU) | DSTC 2&3, DSTC2 | 179 | | Data Augmentation for Spoken Language Understanding via Joint Variational Generation [(AAAI '19)](https://ojs.aaai.org/index.php/AAAI/article/view/4729) | ATIS, Snips, MIT | 180 | | Effective Data Augmentation Approaches to End-to-End Task-Oriented Dialogue [(IALP '19)](https://ieeexplore.ieee.org/document/9037690) | CamRest676, KVRET | 181 | | Paraphrase Augmented Task-Oriented Dialog Generation [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.60/) [code](https://github.com/thu-spmi/PARG) | TCamRest676, MultiWOZ | 182 | | Dialog State Tracking with Reinforced Data Augmentation [(AAAI '20)](https://ojs.aaai.org/index.php/AAAI/article/view/6491) | WoZ, MultiWoZ | 183 | | Data Augmentation for Copy-Mechanism in Dialogue State Tracking [(arxiv '20)](https://arxiv.org/abs/2002.09634) | WoZ, DSTC2, Multi | 184 | | Simple is Better! Lightweight Data Augmentation for Low Resource Slot Filling and Intent Classification [(PACLIC '20)](https://www.aclweb.org/anthology/2020.paclic-1.20/) [code](https://github.com/slouvan/saug) | ATIS, SNIPS, FB | 185 | | Conversation Graph: Data Augmentation, Training, and Evaluation for Non-Deterministic Dialogue Management [(TACL '21)](https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00352/97777/Conversation-Graph-Data-Augmentation-Training-and) | M2M, MultiWOZ | 186 | | GOLD: Improving Out-of-Scope Detection in Dialogues using Data Augmentation [(EMNLP '21)](https://aclanthology.org/2021.emnlp-main.35/) [code](https://github.com/asappresearch/gold) | SMCalFlow, ROSTD | 187 | | Improving Automated Evaluation of Open Domain Dialog via Diverse Reference Augmentation [(ACL '21 Findings)](https://aclanthology.org/2021.findings-acl.357/) [code](https://github.com/harsh19/Diverse-Reference-Augmentation/)| DailyDialog | 188 | 189 | ### Multimodal 190 | | Paper | Datasets | 191 | | -- | --- | 192 | | Data Augmentation for Visual Question Answering [(INLG '17)](https://www.aclweb.org/anthology/W17-3529/) | COCO-VQA, COCO-QA | 193 | | Low Resource Multi-modal Data Augmentation for End-to-end ASR [(CoRR ’18)](https://deepai.org/publication/low-resource-multi-modal-data-augmentation-for-end-to-end-asr) | TODO | 194 | | Multi-Modal Data Augmentation for End-to-end ASR [(Interspeech '18)](https://www.isca-speech.org/archive/Interspeech_2018/abstracts/2456.html) | Voxforge, HUB4 | 195 | | Augmenting Image Question Answering Dataset by Exploiting Image Captions [(LREC '18)](https://www.aclweb.org/anthology/L18-1436/) | IQA | 196 | | Multimodal Continuous Emotion Recognition with Data Augmentation Using Recurrent Neural Networks [(AVEC '18)](https://dl.acm.org/doi/10.1145/3266302.3266304) | TODO | 197 | | Multimodal Dialogue State Tracking By QA Approach with Data Augmentation [(DSTC8 @ AAAI '20)](https://arxiv.org/abs/2007.09903) | DSTC7-AVSD | 198 | | Data augmentation techniques for the Video Question Answering task [(arxiv '20)](https://arxiv.org/abs/2008.09849) | TGIF-QA, MSVD-QA | 199 | | Data Augmentation for Training Dialog Models Robust to Speech Recognition Errors [(NLP for ConvAI @ ACL '20)](https://arxiv.org/abs/2006.05635) | DSTC2 | 200 | | Semantic Equivalent Adversarial Data Augmentation for Visual Question Answering [(ECCV '20)](https://link.springer.com/chapter/10.1007/978-3-030-58529-7_26) | TODO | 201 | | Text Augmentation Using BERT for Image Captioning [(Applied Sciences '20)](https://www.mdpi.com/2076-3417/10/17/5978) | MSCOCO | 202 | | MDA: Multimodal Data Augmentation Framework for Boosting Performance on Image-Text Sentiment/Emotion Classification Tasks [(IEEE Intelligent Systems '20)](https://ieeexplore.ieee.org/document/9206007) | TODO | 203 | 204 | ### Mitigating Bias 205 | | Paper | Datasets | 206 | | -- | --- | 207 | | Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods. [(NAACL '18)](https://www.aclweb.org/anthology/N18-2003/) [code](https://github.com/uclanlp/corefBias) | WinoBias, OntoNotes| 208 | | Counterfactual Data Augmentation for Mitigating Gender Stereotypes in Languages with Rich Morphology [(ACL '19)](https://www.aclweb.org/anthology/P19-1161/) [code](https://github.com/rycolab/biasCDA) | TODO | 209 | | CONAN - COunter NArratives through Nichesourcing: a Multilingual Dataset of Responses to Fight Online Hate Speech [(ACL '19)](https://aclanthology.org/P19-1271.pdf) [Dataset](https://github.com/marcoguerini/CONAN)| New Dataset Created| 210 | | It’s All in the Name: Mitigating Gender Bias with Name-Based Counterfactual Data Substitution [(EMNLP '19)](https://www.aclweb.org/anthology/D19-1530/) [code](https://github.com/rowanhm/counterfactual-data-substitution) | SSA, Stanford Large Movie Review, SimLex-999 | 211 | | Gender Bias in Neural Natural Language Processing. [(Springer '20)](https://link.springer.com/chapter/10.1007%2F978-3-030-62077-6_14 ) | Wikitext-2, CoNLL-2012 | 212 | | Improving Robustness by Augmenting Training Sentences with Predicate-Argument Structures [(arxiv '20)](https://arxiv.org/abs/2010.12510) | SWAG, CoNLL2009, MultiNLI, HANS| 213 | 214 | ### Mitigating Class Imbalance 215 | | Paper | Datasets | 216 | | -- | --- | 217 | | SMOTE: Synthetic Minority Over-sampling Technique [(Journal of Artificial Intelligence Research '02)](https://www.jair.org/index.php/jair/article/view/10302) | Pima, Phoneme, Adult, E-state, Satimage, Forest Cover, Oil, Mammography, Can | 218 | | Active Learning for Word Sense Disambiguation with Methods for Addressing the Class Imbalance Problem [(EMNLP '07)](https://www.aclweb.org/anthology/D07-1082/) | TODO | 219 | | MLSMOTE: Approaching imbalanced multilabel learning through synthetic instance generation [(Knowledge-Based Systems '15)](https://www.sciencedirect.com/science/article/abs/pii/S0950705115002737?via%3Dihub) | bibtex, cal500, corel5k, slashdot, tmc2007, mediamill, medical, scene, enron, emotions | 220 | | SMOTE for Learning from Imbalanced Data: Progress and Challenges, Marking the 15-year Anniversary [(Journal of Artificial Intelligence Research '18)](https://www.jair.org/index.php/jair/article/view/11192) | TODO | 221 | 222 | ### Adversarial examples 223 | 224 | | Paper | Datsets | 225 | | -- | --- | 226 | | Adversarial Example Generation with Syntactically Controlled Paraphrase Networks [(NAACL '18)](https://www.aclweb.org/anthology/N18-1170/) [code](https://github.com/miyyer/scpn)| SST, SICK | 227 | | AdvEntuRe: Adversarial Training for Textual Entailment with Knowledge-Guided Examples [(ACL '18)](https://www.aclweb.org/anthology/P18-1225/) [code](https://github.com/dykang/adventure)| WordNet, PPDB, SICK, SNLI, SciTail | 228 | | Breaking NLI Systems with Sentences that Require Simple Lexical Inferences [(ACL '18)](https://www.aclweb.org/anthology/P18-2103/) | SNLI, SciTail, MultiNLI | 229 | | Certified Robustness to Adversarial Word Substitutions [(EMNLP '19)](https://www.aclweb.org/anthology/D19-1423/) [code](https://github.com/robinjia/certified-word-sub)| IMDB, SNLI | 230 | | PAWS: Paraphrase Adversaries from Word Scrambling [(NAACL '19)](https://www.aclweb.org/anthology/N19-1131/) [code](https://github.com/google-research-datasets/paws)| PAWS (QQP + Wikipedia) | 231 | | Generating Natural Language Adversarial Examples through Probability Weighted Word Saliency [(ACL '19)](https://aclanthology.org/P19-1103/) [code](https://github.com/JHL-HUST/PWWS) | IMDB, AG’s News, Yahoo Answers | 232 | 233 | 234 | ### Compositionality 235 | 236 | | Paper | Datsets | 237 | | -- | --- | 238 | | Good-Enough Compositional Data Augmentation [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.676.pdf) [code](https://github.com/jacobandreas/geca) | SCAN | 239 | | Sequence-Level Mixed Sample Data Augmentation [(EMNLP '20)](https://www.aclweb.org/anthology/2020.emnlp-main.447) [code](https://github.com/dguo98/seqmix) | IWSLT ’14, WMT ’14 | 240 | 241 | ### Automated Augmentation 242 | 243 | | Paper | Datsets | 244 | | ------------------------------------------------------------ | --------------------------- | 245 | | Learning Data Manipulation for Augmentation and Weighting [(NeurIPS '19)](https://papers.nips.cc/paper/2019/file/671f0311e2754fcdd37f70a8550379bc-Paper.pdf) [code](https://github.com/tanyuqian/learning-data-manipulation) | SST, IMDB, TREC, CIFAR-10 | 246 | | Data Manipulation: Towards Effective Instance Learning for Neural Dialogue Generation via Learning to Augment and Reweight [(ACL '20)](https://www.aclweb.org/anthology/2020.acl-main.564.pdf) | DailyDialog, OpenSubtitles | 247 | | Text AutoAugment: Learning Compositional Augmentation Policy for Text Classification [(EMNLP '21)](https://arxiv.org/abs/2109.00523) [code](https://github.com/lancopku/text-autoaugment) | IMDB, SST2, SST5, TREC, YELP2, YELP5 | 248 | 249 | 250 | ### Popular Resources 251 | - [A visual survey of data augmentation in NLP](https://amitness.com/2020/05/data-augmentation-for-nlp/) 252 | - [nlpaug](https://github.com/makcedward/nlpaug) 253 | - [TextAttack](https://github.com/QData/TextAttack) 254 | - [AugLy](https://github.com/facebookresearch/AugLy) 255 | - [NL-Augmenter 🦎 → 🐍](https://github.com/GEM-benchmark/NL-Augmenter/) 256 | -------------------------------------------------------------------------------- /rules.md: -------------------------------------------------------------------------------- 1 | # Adding a new entry 2 | 3 | Format your paper as the following: 4 | 5 | | Paper | Dataset | 6 | | -- | -- | 7 | | Best Data Augmentation Technique [(ACL '27)](https://github.com/styfeng/DataAug4NLP/) [code](https://github.com/styfeng/DataAug4NLP/) | GLUE | 8 | 9 | Specifically: 10 | 11 | - Only include the title of the paper. 12 | - Capitalize all important words in the title. 13 | - Include the conference in parenthesis, with a link to the paper. Use ACL Anthology if possible. Abbreviate if it's a workshop paper. You could also use the arxiv link (e.g., for AAAI, this might be a good idea). If your paper is arxiv-only, just put 'arxiv'. 14 | - Use '27 instead of 2027. 15 | - Include the code if you have it. 16 | - Add the papers chronologically. 17 | 18 | The only exception to this format is if your paper title is not about data augmentation, but it is well-known for doing data augmentation. 19 | This should follow the first example under text classification called `Synonym Replacement`. 20 | --------------------------------------------------------------------------------