Neural latent variable model for open domain event extraction, outperforming state-of-the-art on large-scale news datasets.

Xiao Liu, Heyan Huang, Yue Zhang

This study introduces a deep learning-based personalized persuasive dialogue system, achieving 59.6% F1 on a 300-sample annotated dataset, advancing social good applications.

Xuewei Wang, Weiyan Shi, Richard Kim et al.

This study analyzes BERT's attention heads, revealing their alignment with syntactic relations, and demonstrates dependency parsing with 77% accuracy using attention maps.

Kevin Clark, Urvashi Khandelwal, Omer Levy et al.

BIGPATENT dataset offers 1.3 million patent abstracts, advancing high-compression abstractive summarization.

Eva Sharma, Chen Li, Lu Wang

Proposes substantiated perspective discovery task; constructs PERSPECTRUM dataset with 1000 claims, 10,000 perspectives, 8,000 evidence paragraphs; baseline models outperform simple methods but lag behind humans.

Sihao Chen, Daniel Khashabi, Wenpeng Yin et al.

Proposes email subject line generation using a two-stage extractive-abstractive model, trained on AESLC, outperforming baselines with ROUGE-1 score of 25.41.

Rui Zhang, Joel Tetreault

Introduced Multi-News dataset and hierarchical abstractive model Hi-MAP, achieving ROUGE-1 35.78 and outperforming baselines.

Alexander R. Fabbri, Irene Li, Tianwei She et al.

Proposes Latent Retrieval with BERT and ICT pretraining, achieving 19-point EM gains over BM25 in open-domain QA.

Kenton Lee, Ming-Wei Chang, Kristina Toutanova

Introduces MathQA dataset with operation-based formalism and neural models for interpretable math problem solving, achieving 54.2% accuracy.

Aida Amini, Saadia Gabriel, Peter Lin et al.

This study reveals that many attention heads in multi-head attention are redundant; introduces gradient-based greedy pruning to improve efficiency.

Paul Michel, Omer Levy, Graham Neubig

LRP and Hard Concrete pruning show that retaining 10 of 48 encoder heads cuts WMT En–Ru BLEU by only 0.15.

Elena Voita, David Talbot, Fedor Moiseev et al.

HellaSwag dataset leverages adversarial filtering to challenge state-of-the-art models, exposing their limitations in commonsense reasoning with only ~48% accuracy versus 95% humans.

Rowan Zellers, Ari Holtzman, Yonatan Bisk et al.

Multi-hop reading comprehension via reasoning over heterogeneous graphs, achieving state-of-the-art on WIKIHOP dataset.

Ming Tu, Guangtao Wang, Jing Huang et al.

Using Graph Neural Networks to encode database schema structure, improving Text-to-SQL parsing accuracy to 39.4%.

Ben Bogin, Matt Gardner, Jonathan Berant

SuperGLUE introduces more challenging tasks and diverse formats, pushing NLP benchmarks toward deeper understanding.

Alex Wang, Yada Pruksachatkun, Nikita Nangia et al.

Proposes Nucleus Sampling, a dynamic probability truncation method, to enhance diversity and coherence in neural text generation.

Ari Holtzman, Jan Buys, Li Du et al.

PullNet uses iterative subgraph building with Graph CNNs, achieving state-of-the-art multi-hop QA performance, especially with incomplete KBs.

Haitian Sun, Tania Bedrax-Weiss, William W. Cohen

Introduced TextVQA dataset and LoRRA model, achieving 27.63% accuracy on text-based VQA, surpassing SOTA methods.

Amanpreet Singh, Vivek Natarajan, Meet Shah et al.

ClinicalBERT uses bidirectional transformers to improve 30-day readmission prediction accuracy.

Kexin Huang, Jaan Altosaar, Rajesh Ranganath

Proposes environmental Dropout combined with back-translation to improve unseen environment success rate from 63% to 68.9% in Room-to-Room.

Hao Tan, Licheng Yu, Mohit Bansal