Times are displayed in your local timezone. This is a preliminary schedule and subject to change. Search by author and title is available on the accepted paper listing. See the virtual infrastructure blog post for more information about the formats of the presentations.

Show times in:
Nov 16th
15 minutes

Opening Remarks

Nov 16th
60 minutes

Keynote I: Claire Cardie

Nov 16th
60 minutes

Zoom Q&A Session 1A: Sentiment Analysis, Stylistic Analysis, and Argument Mining

Chair: Chenghua Lin (Sheffield)
  • Detecting Attackable Sentences in Arguments
  • Extracting Implicitly Asserted Propositions in Argumentation
  • Quantitative Argument Summarization and Beyond: Cross-Domain Key Point Analysis
  • Unsupervised Stance Detection for Arguments from Consequences

Zoom Q&A Session 1B: Machine Translation and Multilinguality

Chair: Colin Cherry (Google)
  • BLEU might be Guilty but References are not Innocent
  • Statistical Power and Translationese in Machine Translation Evaluation
  • Simulated Multiple Reference Training Improves Low-Resource Machine Translation
  • Automatic Machine Translation Evaluation in Many Languages via Zero-Shot Paraphrasing
  • Unsupervised Quality Estimation for Neural Machine Translation

Zoom Q&A Session 1C: Question Answering

Chair: Avi Sil (IBM Research)
  • PRover: Proof Generation for Interpretable Reasoning over Rules
  • Learning to Explain: Datasets and Models for Identifying Valid Reasoning Chains in Multihop Question-Answering
  • Self-Supervised Knowledge Triplet Learning for Zero-Shot Question Answering
  • More Bang for Your Buck: Natural Perturbation for Robust Question Answering
  • What Does My QA Model Know? Devising Controlled Probes using Expert

Zoom Q&A Session 1D: Interpretability and Analysis of Models for NLP

Chair: Ryan Cotterell (ETH Zürich)
  • A Matter of Framing: The Impact of Linguistic Formalism on Probing Results
  • Information-Theoretic Probing with Minimum Description Length
  • Intrinsic Probing through Dimension Selection
  • Learning Which Features Matter: RoBERTa Acquires a Preference for Linguistic Generalizations (Eventually)
Nov 16th
60 minutes

Zoom Q&A Session 2A: Machine Learning for NLP

Chair: Dani Yogatama
  • KERMIT: Complementing Transformer Architectures with Encoders of Explicit Syntactic Interpretations
  • ETC: Encoding Long and Structured Inputs in Transformers
  • Pre-Training Transformers as Energy-Based Cloze Models
  • Calibration of Pre-trained Transformers

Zoom Q&A Session 2B: NLP Applications

Chair: Maria Liakata
  • Data Weighted Training Strategies for Grammatical Error Correction
  • Near-imperceptible Neural Linguistic Steganography via Self-Adjusting Arithmetic Coding
  • Multi-Dimensional Gender Bias Classification
  • FIND: Human-in-the-Loop Debugging Deep Text Classifiers
  • Conversational Document Prediction to Assist Customer Care Agents

Zoom Q&A Session 2C: Dialog and Interactive Systems

Chair: Mark Hasegawa-Johnson (UIUC)
  • Incremental Processing in the Age of Non-Incremental Encoders: An Empirical Assessment of Bidirectional Models for Incremental NLU
  • Task-Oriented Dialogue as Dataflow Synthesis
  • Augmented Natural Language for Generative Sequence Labeling
  • Dialogue Response Ranking Training with Large-Scale Human Feedback Data

Zoom Q&A Session 2D: Semantics: Sentence-level Semantics, Textual Inference and Other areas

Chair: Eduardo Blanco (UNT)
  • Semantic Evaluation for Text-to-SQL with Distilled Test Suites
  • Cross-Thought for Sentence Encoder Pre-training
  • AutoQA: From Databases To QA Semantic Parsers With Only Synthetic Training Data
  • Sketch-Driven Regular Expression Generation from Natural Language and Examples
Nov 16th
60 minutes

Industry Panel: Fei Sha, Chin-Yew Lin, Kristina Toutanova, Daniel Marcu, Joel Tetreault, João Graça

Nov 17th
60 minutes

Zoom Q&A Session 3A: Summarization

Chair: Massimo Piccardi (UTS)
  • A Spectral Method for Unsupervised Multi-Document Summarization
  • What Have We Achieved on Text Summarization?
  • Q-learning with Language Model for Edit-based Unsupervised Summarization
  • Friendly Topic Assistant for Transformer Based Abstractive Summarization

Zoom Q&A Session 3B: Machine Learning for NLP

Chair: Wray Buntine (Monash)
  • Contrastive Distillation on Intermediate Representations for Language Model Compression
  • TernaryBERT: Distillation-aware Ultra-low Bit BERT
  • Repulsive Attention: Rethinking Multi-head Attention as Bayesian Inference
  • Self-Supervised Meta-Learning for Few-Shot Natural Language Classification Tasks
  • Efficient Meta Lifelong-Learning with Limited Memory

Zoom Q&A Session 3C: Machine Translation and Multilinguality

Chair: Lei Li (ByteDance)
  • Don't Use English Dev: On the Zero-Shot Cross-Lingual Evaluation of Contextual Embeddings
  • Multilingual Denoising Pre-training for Neural Machine Translation
  • A Supervised Word Alignment Method based on Cross-Language Span Prediction using Multilingual BERT
  • Accurate Word Alignment Induction from Neural Machine Translation
  • ChrEn: Cherokee-English Machine Translation for Endangered Language Revitalization

Zoom Q&A Session 3D: Computational Social Science and Social Media

Chair: Alice Oh
  • Unsupervised Discovery of Implicit Gender Bias
  • Condolence and Empathy in Online Communities
  • An Embedding Model for Estimating Legislative Preferences from the Frequency and Sentiment of Tweets
  • Measuring Information Propagation in Literary Social Networks
  • Social Chemistry 101: Learning to Reason about Social and Moral Norms
Nov 17th
60 minutes

Zoom Q&A Session 4A: Information Extraction

Chair: Eunsol Choi
  • Event Extraction by Answering (Almost) Natural Questions
  • Connecting the Dots: Event Graph Schema Induction with Path Language Modeling
  • Joint Constrained Learning for Event-Event Relation Extraction
  • Incremental Event Detection via Knowledge Consolidation Networks
  • Semi-supervised New Event Type Induction and Event Detection

Zoom Q&A Session 4B: Language Generation

Chair: Greg Durrett
  • Language Generation with Multi-Hop Reasoning on Commonsense Knowledge Graph
  • Reformulating Unsupervised Style Transfer as Paraphrase Generation
  • PAIR: Planning and Iterative Refinement in Pre-trained Transformers for Long Text Generation
  • Plan ahead: Self-Supervised Text Planning for Paragraph Completion Task
  • Back to the Future: Unsupervised Backprop-based Decoding for Counterfactual and Abductive Commonsense Reasoning

Zoom Q&A Session 4C: Language Grounding to Vision, Robotics and Beyond

Chair: Xin Eric Wang (UCSC)
  • Where Are You? Localization from Embodied Dialog
  • Learning to Represent Image and Text with Denotation Graph
  • Video2Commonsense: Generating Commonsense Descriptions to Enrich Video Captioning
  • Does my multimodal model learn cross-modal interactions? It’s harder to tell than you might think!
  • MUTANT: A Training Paradigm for Out-of-Distribution Generalization in Visual Question Answering

Zoom Q&A Session 4D: Dialog and Interactive Systems

Chair: Kevin Small (Amazon)
  • Mitigating Gender Bias for Neural Dialogue Generation with Adversarial Learning
  • Will I Sound Like Me? Improving Persona Consistency in Dialogues through Pragmatic Self-Consciousness
  • TOD-BERT: Pre-trained Natural Language Understanding for Task-Oriented Dialogue
  • RiSAWOZ: A Large-Scale Multi-Domain Wizard-of-Oz Dataset with Rich Semantic Annotations for Task-Oriented Dialogue Modeling
  • Filtering Noisy Dialogue Corpora by Connectivity and Content Relatedness
Nov 17th
120 minutes

Gather Session 1A: Machine Translation and Multilinguality

  • Shallow-to-Deep Training for Neural Machine Translation
  • Iterative Refinement in the Continuous Space for Non-Autoregressive Neural Machine Translation
  • Why Skip If You Can Combine: A Simple Knowledge Distillation Technique for Intermediate Layers
  • Multi-task Learning for Multilingual Neural Machine Translation
  • Token-level Adaptive Training for Neural Machine Translation
  • Multi-Unit Transformers for Neural Machine Translation
  • On the Sparsity of Neural Machine Translation Models
  • Incorporating a Local Translation Mechanism into Non-autoregressive Translation
  • Self-Paced Learning for Neural Machine Translation
  • Long-Short Term Masking Transformer: A Simple but Effective Baseline for Document-level Neural Machine Translation
  • Pre-tokenization of Multi-word Expressions in Cross-lingual Word Embeddings
  • Generating Diverse Translation from Model Distribution with Dropout
  • Non-Autoregressive Machine Translation with Latent Alignments

Gather Session 1B: Machine Learning for NLP

  • Local Additivity Based Data Augmentation for Semi-supervised NER
  • Grounded Compositional Outputs for Adaptive Language Modeling
  • SSMBA: Self-Supervised Manifold Based Data Augmentation for Improving Out-of-Domain Robustness
  • SetConv: A New Approach for Learning from Imbalanced Data
  • Scalable Multi-Hop Relational Reasoning for Knowledge-Aware Question Answering
  • Improving Bilingual Lexicon Induction for Low Frequency Words
  • BERT-of-Theseus: Compressing BERT by Progressive Module Replacing
  • Learning VAE-LDA Models with Rounded Reparameterization Trick
  • Calibrated Language Model Fine-Tuning for In- and Out-of-Distribution Data
  • Scaling Hidden Markov Language Models
  • Coding Textual Inputs Boosts the Accuracy of Neural Networks
  • Learning from Task Descriptions

Gather Session 1C: Linguistic Theories, Cognitive Modeling and Psycholinguistics; Semantics: Sentence-level Semantics, Textual Inference and Other areas

  • Latent Geographical Factors for Analyzing the Evolution of Dialects in Contact
  • Predicting Reference: What do Language Models Learn about Discourse Models?
  • Word class flexibility: A deep contextualized approach
  • Benchmarking Meaning Representations in Neural Semantic Parsing
  • Analogous Process Structure Induction for Sub-event Sequence Prediction
  • SLM: Learning a Discourse Language Representation with Sentence Unshuffling
  • Detecting Fine-Grained Cross-Lingual Semantic Divergences without Supervision by Learning to Rank
  • A Bilingual Generative Transformer for Semantic Sentence Embedding
  • Semantically Inspired AMR Alignment for the Portuguese Language
  • An Unsupervised Sentence Embedding Method by Mutual Information Maximization
  • Compositional Phrase Alignment and Beyond

Gather Session 1D: Information Extraction

  • Table Fact Verification with Structure-Aware Transformer
  • Double Graph Based Reasoning for Document-level Relation Extraction
  • Event Extraction as Machine Reading Comprehension
  • MAVEN: A Massive General Domain Event Detection Dataset
  • Knowledge Graph Alignment with Entity-Pair Embedding
  • Adaptive Attentional Network for Few-Shot Knowledge Graph Completion
  • Pre-training Entity Relation Encoder with Intra-span and Inter-span Information
  • Two are Better than One: Joint Entity and Relation Extraction with Table-Sequence Encoders

Gather Session 1E: Dialog and Interactive Systems

  • BiST: Bi-directional Spatio-Temporal Reasoning for Video-Grounded Dialogues
  • UniConv: A Unified Conversational Neural Architecture for Multi-domain Task-oriented Dialogues
  • GraphDialog: Integrating Graph Knowledge into End-to-End Task-Oriented Dialogue Systems
  • Structured Attention for Unsupervised Dialogue Structure Induction
  • Cross Copy Network for Dialogue Generation
  • Multi-turn Response Selection using Dialogue Dependency Relations
  • Parallel Interactive Networks for Multi-Domain Dialogue State Generation
  • SlotRefine: A Fast Non-Autoregressive Model for Joint Intent Detection and Slot Filling

Gather Session 1F: Computational Social Science and Social Media; NLP Applications

  • Hashtags, Emotions, and Comments: A Large-Scale Dataset to Understand Fine-Grained Social Emotions to Online Topics
  • Named Entity Recognition for Social Media Texts with Semantic Augmentation
  • Coupled Hierarchical Transformer for Stance-Aware Rumor Verification in Social Media Conversations
  • Social Media Attributions in the Context of Water Crisis
  • Towards Medical Machine Reading Comprehension with Structural Knowledge and Plain Text
  • Generating Radiology Reports via Memory-driven Transformer
  • Planning and Generating Natural and Diverse Disfluent Texts as Augmentation for Disfluency Detection
  • Predicting Clinical Trial Results by Implicit Evidence Integration
  • Explainable Clinical Decision Support from Text
  • Routing Enforced Generative Model for Recipe Generation
  • A Knowledge-driven Generative Model for Multi-implication Chinese Medical Procedure Entity Normalization
  • Combining Automatic Labelers and Expert Annotations for Accurate Radiology Report Labeling Using BERT

Gather Session 1G: Information Retrieval and Text Mining; Speech and Multimodality

  • Beyond [CLS] through Ranking by Generation
  • Tired of Topic Models? Clusters of Pretrained Word Embeddings Make for Fast and Good Topics too!
  • Multi-document Summarization with Maximal Marginal Relevance-guided Reinforcement Learning
  • Improving Neural Topic Models using Knowledge Distillation
  • Short Text Topic Modeling with Topic Distribution Quantization and Negative Sampling Decoder
  • Querying Across Genres for Medical Claims in News
  • Incorporating Multimodal Information in Open-Domain Web Keyphrase Extraction
  • CMU-MOSEAS: A Multimodal Language Dataset for Spanish, Portuguese, German and French
  • Combining Self-Training and Self-Supervised Learning for Unsupervised Disfluency Detection
  • Multimodal Routing: Improving Local and Global Interpretability of Multimodal Language Analysis
  • Multistage Fusion with Forget Gate for Multimodal Summarization in Open-Domain Videos

Gather Session 1H: Language Grounding to Vision, Robotics and Beyond; Question Answering

  • Visually Grounded Continual Learning of Compositional Phrases
  • MAF: Multimodal Alignment Framework for Weakly-Supervised Phrase Grounding
  • Domain-Specific Lexical Grounding in Noisy Visual-Textual Documents
  • HERO: Hierarchical Encoder for Video+Language Omni-representation Pre-training
  • Vokenization: Improving Language Understanding with Contextualized, Visual-Grounded Supervision
  • Detecting Cross-Modal Inconsistency to Defend Against Neural Fake News
  • Look at the First Sentence: Position Bias in Question Answering
  • ProtoQA: A Question Answering Dataset for Prototypical Common-Sense Reasoning
  • IIRC: A Dataset of Incomplete Information Reading Comprehension Questions
  • Unsupervised Adaptation of Question Answering Systems via Generative Self-training
  • TORQUE: A Reading Comprehension Dataset of Temporal Ordering Questions

Gather Session 1I: Interpretability and Analysis of Models for NLP; Language Generation

  • An Information Bottleneck Approach for Controlling Conciseness in Rationale Extraction
  • LOGAN: Local Group Bias Detection by Clustering
  • RNNs can generate bounded hierarchical languages with optimal memory
  • Detecting Independent Pronoun Bias with Partially-Synthetic Data Generation
  • ToTTo: A Controlled Table-To-Text Generation Dataset
  • ENT-DESC: Entity Description Generation by Exploring Knowledge Graph
  • Small but Mighty: New Benchmarks for Split and Rephrase
  • De-Biased Court’s View Generation with Causality
  • Online Back-Parsing for AMR-to-Text Generation
  • Reading Between the Lines: Exploring Infilling in Visual Narratives
  • Acrostic Poem Generation

Gather Session 1J: Demos

  • OpenUE: An Open Toolkit of Universal Extraction from Text
  • Wikipedia2Vec: An Efficient Toolkit for Learning and Visualizing the Embeddings of Words and Entities from Wikipedia
  • CoSaTa: A Constraint Satisfaction Solver and Interpreted Language for Semi-Structured Tables of Sentences
  • The Language Interpretability Tool: Extensible, Interactive Visualizations and Analysis for NLP Models
  • SIMULEVAL: An Evaluation Toolkit for Simultaneous Translation
  • WantWords: An Open-source Online Reverse Dictionary System

Gather Session 1K: Sponsor Booths

Nov 17th
60 minutes

Zoom Q&A Session 5A: Information Extraction

Chair: Kang Liu
  • Enhancing Aspect Term Extraction with Soft Prototypes
  • FedED: Federated Learning via Ensemble Distillation for Medical Relation Extraction
  • Multimodal Joint Attribute Prediction and Value Extraction for E-commerce Product
  • A Predicate-Function-Argument Annotation of Natural Language for Open-Domain Information eXpression

Zoom Q&A Session 5B: Language Generation

Chair: Lei Li (ByteDance)
  • Retrofitting Structure-aware Transformer Language Model for End Tasks
  • Lightweight, Dynamic Graph Convolutional Networks for AMR-to-Text Generation
  • Modeling Global and Local Node Contexts for Text Generation from Knowledge Graphs
  • If beam search is the answer, what was the question?
  • A* Beam Search

Zoom Q&A Session 5C: Machine Learning for NLP

Chair: Reza Haffari (Monash University)
  • Understanding the Mechanics of SPIGOT: Surrogate Gradients for Latent Structure Learning
  • Is the Best Better? Bayesian Statistical Model Comparison for Natural Language Processing
  • Exploring Logically Dependent Multi-task Learning with Causal Inference
  • Masking as an Efficient Alternative to Finetuning for Pretrained Language Models
  • Interactive Text Ranking with Bayesian Optimisation: A Case Study on Community QA and Summarisation

Zoom Q&A Session 5D: Machine Translation and Multilinguality

Chair: Barry Haddow (University of Edinburgh)
  • Dynamic Context Selection for Document-level Neural Machine Translation via Reinforcement Learning
  • Data Rejuvenation: Exploiting Inactive Training Examples for Neural Machine Translation
  • Pronoun-Targeted Fine-tuning for NMT with Hybrid Losses
  • Learning Adaptive Segmentation Policy for Simultaneous Translation
  • Learn to Cross-lingual Transfer with Meta Graph Learning Across Heterogeneous Languages
Nov 17th
60 minutes

Zoom Q&A Session 6A: Syntax: Tagging, Chunking, and Parsing

Chair: Mark Johnson
  • Syntactic Structure Distillation Pretraining for Bidirectional Encoders
  • UDapter: Language Adaptation for Truly Universal Dependency Parsing
  • Uncertainty-Aware Label Refinement for Sequence Labeling
  • Adversarial Attack and Defense of Structured Prediction Models
  • Position-Aware Tagging for Aspect Sentiment Triplet Extraction

Zoom Q&A Session 6B: Machine Translation and Multilinguality

Chair: Ekaterina Vylomova (University of Melbourne)
  • Simultaneous Machine Translation with Visual Context
  • XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning
  • The Secret is in the Spectra: Predicting Cross-lingual Task Performance with Spectral Similarity Measures
  • Bridging Linguistic Typology and Multilingual Machine Translation with Multi-View Language Representations
  • Semantic Drift in Multilingual Representations

Zoom Q&A Session 6C: Question Answering

Chair: Wanxiang Che
  • AnswerFact: Fact Checking in Product Question Answering
  • Context-Aware Answer Extraction in Question Answering
  • What do Models Learn from Question Answering Datasets?
  • Discern: Discourse-Aware Entailment Reasoning Network for Conversational Machine Reading

Zoom Q&A Session 6D: Semantics: Sentence-level Semantics, Textual Inference and Other areas

Chair: Gabriel Stanovsky (Hebrew University)
  • A Method for Building a Commonsense Inference Dataset based on Basic Events
  • Neural Deepfake Detection with Factual Structure of Text
  • MultiCQA: Zero-Shot Transfer of Self-Supervised Text Matching Models on a Massive Scale
  • XL-AMR: Enabling Cross-Lingual AMR Parsing with Transfer Learning Techniques
  • Improving AMR Parsing with Sequence-to-Sequence Pre-training
Nov 17th
120 minutes

Gather Session 2A: Machine Learning for NLP

  • Exploring the Linear Subspace Hypothesis in Gender Bias Mitigation
  • Lifelong Language Knowledge Distillation
  • Sparse Parallel Training of Hierarchical Dirichlet Process Topic Models
  • Multi-label Few/Zero-shot Learning with Knowledge Aggregated from Multiple Label Graphs
  • Word Rotator's Distance
  • Disentangle-based Continual Graph Representation Learning
  • Semi-Supervised Bilingual Lexicon Induction with Two-way Interaction
  • Wasserstein Distance Regularized Sequence Representation for Text Matching in Asymmetrical Domains
  • A Simple Approach to Learning Unsupervised Multilingual Embeddings
  • Bootstrapped Q-learning with Context Relevant Observation Pruning to Generalize in Text-based Games
  • BERT-EMD: Many-to-Many Layer Mapping for BERT Compression with Earth Mover's Distance
  • Slot Attention with Value Normalization for Multi-Domain Dialogue State Tracking

Gather Session 2B: Dialog and Interactive Systems

  • Knowledge-Grounded Dialogue Generation with Pre-trained Language Models
  • MinTL: Minimalist Transfer Learning for Task-Oriented Dialogue Systems
  • Variational Hierarchical Dialog Autoencoder for Dialog State Tracking Data Augmentation
  • Bridging the Gap between Prior and Posterior Knowledge Selection for Knowledge-Grounded Dialogue Generation
  • Counterfactual Off-Policy Training for Neural Dialogue Generation
  • Dialogue Distillation: Open-Domain Dialogue Augmentation Using Unpaired Data
  • Task-Completion Dialogue Policy Learning via Monte Carlo Tree Search with Dueling Network
  • Learning a Simple and Effective Model for Multi-turn Response Generation with Auxiliary Tasks
  • AttnIO: Knowledge Graph Exploration with In-and-Out Attention Flow for Knowledge-Grounded Dialogue
  • Amalgamating Knowledge from Two Teachers for Task-oriented Dialogue System with Adversarial Training

Gather Session 2C: Information Extraction

  • Learning from Context or Names? An Empirical Study on Neural Relation Extraction
  • SelfORE: Self-supervised Relational Feature Learning for Open Relation Extraction
  • Denoising Relation Extraction from Document-level Distant Supervision
  • Let's Stop Incorrect Comparisons in End-to-end Relation Extraction!
  • Exposing Shallow Heuristics of Relation Extraction Models with Challenge Data
  • Global-to-Local Neural Networks for Document-Level Relation Extraction
  • Recurrent Interaction Network for Jointly Extracting Entities and Classifying Relations
  • Temporal Knowledge Base Completion: New Algorithms and Evaluation Protocols
  • OpenIE6: Iterative Grid Labeling and Coordination Analysis for Open Information Extraction

Gather Session 2D: NLP Applications; Semantics: Lexical Semantics

  • Public Sentiment Drift Analysis Based on Hierarchical Variational Auto-encoder
  • Point to the Expression: Solving Algebraic Word Problems using the Expression-Pointer Transformer Model
  • Deep Attentive Learning for Stock Movement Prediction From Social Media Text and Company Correlations
  • Semantically-Aligned Universal Tree-Structured Solver for Math Word Problems
  • Neural Topic Modeling by Incorporating Document Relationship Graph
  • Selection and Generation: Learning towards Multi-Product Advertisement Post Generation
  • Form2Seq : A Framework for Higher-Order Form Structure Extraction
  • Task-oriented Domain-specific Meta-Embedding for Text Classification
  • Don't Neglect the Obvious: On the Role of Unambiguous Words in Word Sense Disambiguation
  • Exploring Semantic Capacity of Terms
  • Within-Between Lexical Relation Classification
  • With More Contexts Comes Better Performance: Contextualized Sense Embeddings for All-Round Word Sense Disambiguation

Gather Session 2E: Machine Translation and Multilinguality; Phonology, Morphology and Word Segmentation

  • Translation Quality Estimation by Jointly Learning to Score and Rank
  • CSP:Code-Switching Pre-training for Neural Machine Translation
  • Pre-training Multilingual Neural Machine Translation by Leveraging Alignment Information
  • Towards Enhancing Faithfulness for Neural Machine Translation
  • COMET: A Neural Framework for MT Evaluation
  • LNMap: Departures from Isomorphic Assumption in Bilingual Lexicon Induction Through Non-Linear Mapping in Latent Space
  • Uncertainty-Aware Semantic Augmentation for Neural Machine Translation
  • Can Automatic Post-Editing Improve NMT?
  • Domain Adaptation of Thai Word Segmentation Models using Stacked Ensemble
  • DagoBERT: Generating Derivational Morphology with a Pretrained Language Model
  • Attention Is All You Need for Chinese Word Segmentation
  • A Joint Multiple Criteria Model in Transfer Learning for Cross-domain Chinese Word Segmentation

Gather Session 2F: Discourse and Pragmatics; Machine Translation and Multilinguality

  • TED-CDB: A Large-Scale Chinese Discourse Relation Dataset on TED Talks
  • QADiscourse - Discourse Relations as QA Pairs: Representation, Crowdsourcing and Baselines
  • Discourse Self-Attention for Discourse Element Identification in Argumentative Student Essays
  • Self-Induced Curriculum Learning in Self-Supervised Neural Machine Translation
  • Towards Reasonably-Sized Character-Level Transformer NMT by Finetuning Subword Systems
  • Transfer Learning and Distant Supervision for Multilingual Transformer Models: A Study on African Languages
  • Direct Segmentation Models for Streaming Speech Translation
  • Not Low-Resource Anymore: Aligner Ensembling, Batch Filtering, and New Datasets for Bengali-English Machine Translation
  • Type B Reflexivization as an Unambiguous Testbed for Multilingual Multi-Task Gender Bias
  • Losing Heads in the Lottery: Pruning Transformer Attention in Neural Machine Translation
  • Reusing a Pretrained Language Model on Languages with Limited Corpora for Unsupervised NMT

Gather Session 2G: Language Grounding to Vision, Robotics and Beyond

  • STL-CQA: Structure-based Transformers with Localization and Encoding for Chart Question Answering
  • Learning to Contrast the Counterfactual Samples for Robust Visual Question Answering
  • Learning Physical Common Sense as Knowledge Graph Completion via BERT Data Augmentation and Constrained Tucker Factorization
  • A Visually-grounded First-person Dialogue Dataset with Verbal and Non-verbal Responses
  • Cross-Media Keyphrase Prediction: A Unified Framework with Multi-Modality Multi-Head Attention and Image Wordings
  • VD-BERT: A Unified Vision and Dialog Transformer with BERT
  • The Grammar of Emergent Languages
  • Sub-Instruction Aware Vision-and-Language Navigation

Gather Session 2H: Computational Social Science and Social Media; Sentiment Analysis, Stylistic Analysis, and Argument Mining

  • Hate-Speech and Offensive Language Detection in Roman Urdu
  • Suicidal Risk Detection for Military Personnel
  • Comparative Evaluation of Label-Agnostic Selection Bias in Multilingual Hate Speech Datasets
  • HENIN: Learning Heterogeneous Neural Interaction Networks for Explainable Cyberbullying Detection on Social Media
  • Reactive Supervision: A New Method for Collecting Sarcasm Data
  • Convolution over Hierarchical Syntactic and Lexical Graphs for Aspect Level Sentiment Analysis
  • Multi-Instance Multi-Label Learning Networks for Aspect-Category Sentiment Analysis
  • Aspect Sentiment Classification with Aspect-Specific Opinion Spans
  • Emotion-Cause Pair Extraction as Sequence Labeling Based on A Novel Tagging Scheme
  • End-to-End Emotion-Cause Pair Extraction based on Sliding Window Multi-Label Learning
  • Multi-modal Multi-label Emotion Detection with Modality and Label Dependence
  • Tasty Burgers, Soggy Fries: Probing Aspect Robustness in Aspect-Based Sentiment Analysis

Gather Session 2I: Information Retrieval and Text Mining; Language Generation

  • Top-Rank-Focused Adaptive Vote Collection for the Evaluation of Domain-Specific Semantic Models
  • Meta Fine-Tuning Neural Language Models for Multi-Domain Text Mining
  • Incorporating Behavioral Hypotheses for Query Generation
  • Conditional Causal Relationships between Emotions and Causes in Texts
  • COMETA: A Corpus for Medical Entity Linking in the Social Media
  • MEGATRON-CNTRL: Controllable Story Generation with External Knowledge Using Large-Scale Language Models
  • Incomplete Utterance Rewriting as Semantic Segmentation
  • Improving Grammatical Error Correction Models with Purpose-Built Adversarial Examples
  • Homophonic Pun Generation with Lexically Constrained Rewriting
  • How to Make Neural Natural Language Generation as Reliable as Templates in Task-Oriented Dialogue
  • Multilingual AMR-to-Text Generation

Gather Session 2J: Question Answering; Syntax: Tagging, Chunking, and Parsing

  • Don't Read Too Much Into It: Adaptive Computation for Open-Domain Question Answering
  • Multi-Step Inference for Reasoning Over Paragraphs
  • Learning a Cost-Effective Annotation Policy for Question Answering
  • Scene Restoring for Narrative Machine Reading Comprehension
  • A Simple and Effective Model for Answering Multi-span Questions
  • Parsing Gapping Constructions Based on Grammatical and Semantic Roles
  • Span-based discontinuous constituency parsing: a family of exact chart-based algorithms with time complexities from O(n^6) down to O(n^3)
  • Some Languages Seem Easier to Parse Because Their Treebanks Leak
  • Discontinuous Constituent Parsing as Sequence Labeling
  • Modularized Syntactic Neural Networks for Sentence Classification

Gather Session 2K: Interpretability and Analysis of Models for NLP; Summarization

  • Pareto Probing: Trading Off Accuracy for Complexity
  • Interpretation of NLP models through input marginalization
  • Generating Label Cohesive and Well-Formed Adversarial Claims
  • Cold-Start and Interpretability: Turning Regular Expressions into Trainable Recurrent Neural Networks
  • A Diagnostic Study of Explainability Techniques for Text Classification
  • Modeling Content Importance for Summarization with Pre-trained Language Models
  • Unsupervised Reference-Free Summary Quality Evaluation via Contrastive Learning
  • Neural Extractive Summarization with Hierarchical Attentive Heterogeneous Graph Network
  • Coarse-to-Fine Query Focused Multi-Document Summarization
  • Pre-training for Abstractive Document Summarization by Reinstating Source Text

Gather Session 2L: Interpretability and Analysis of Models for NLP; Semantics: Sentence-level Semantics, Textual Inference and Other areas

  • Are All Good Word Vector Spaces Isomorphic?
  • When BERT Plays the Lottery, All Tickets Are Winning
  • On the weak link between importance and prunability of attention heads
  • Towards Interpreting BERT for Reading Comprehension Based QA
  • How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking
  • Alignment-free Cross-lingual Semantic Role Labeling
  • Leveraging Declarative Knowledge in Text and First-Order Logic for Fine-Grained Propaganda Detection
  • X-SRL: A Parallel Cross-Lingual Semantic Role Labeling Dataset
  • Graph Convolutions over Constituent Trees for Syntax-Aware Semantic Role Labeling
  • Fast semantic parsing with well-typedness guarantees

Gather Session 2M: Demos

  • BERTweet: A pre-trained language model for English Tweets
  • AdapterHub: A Framework for Adapting Transformers
  • SciSight: Combining faceted navigation and research group detection for COVID-19 exploratory scientific search
  • BENNERD: A Neural Named Entity Linking System for COVID-19
  • Langsmith: An Interactive Academic Text Revision System
  • IsOBS: An Information System for Oracle Bone Script

Gather Session 2N: Sponsor Booths

Nov 17th
60 minutes

Ethics Panel: Publishing in an era of Responsible AI: How can NLP be proactive? Considerations and Implications; Moderator: Mona Diab; Panelists: Emily Bender, Rosie Campbell, Allan Dafoe, Pascale Fung, Meg Mitchell, Saif Mohammad

Nov 17th
60 minutes

Zoom Q&A Session 7A: Dialog and Interactive Systems

Chair: Seokhwan Kim (Amazon Alexa AI)
  • Improving Out-of-Scope Detection in Intent Classification by Using Embeddings of the Word Graph Space of the Classes
  • Supervised Seeded Iterated Learning for Interactive Language Learning
  • Spot The Bot: A Robust and Efficient Framework for the Evaluation of Conversational Dialogue Systems
  • Human-centric dialog training via offline reinforcement learning

Zoom Q&A Session 7B: Linguistic Theories, Cognitive Modeling and Psycholinguistics

Chair: Roger Levy
  • Speakers Fill Lexical Semantic Gaps with Context
  • Investigating Cross-Linguistic Adjective Ordering Tendencies with a Latent-Variable Model
  • Surprisal Predicts Code-Switching in Chinese-English Bilingual Text
  • Word Frequency Does Not Predict Grammatical Knowledge in Language Models
  • BLiMP: The Benchmark of Linguistic Minimal Pairs for English

Zoom Q&A Session 7C: Semantics: Lexical Semantics

Chair: Michael Roth (UStuttgart)
  • Improving Word Sense Disambiguation with Translations
  • Towards Better Context-aware Lexical Semantics:Adjusting Contextualized Representations through Static Anchors
  • Sequential Modelling of the Evolution of Word Representations for Semantic Change Detection
  • Do "Undocumented Workers" == "Illegal Aliens"? Differentiating Denotation and Connotation in Vector Spaces

Zoom Q&A Session 7D: Summarization

Chair: Asma Ben Abacha (NLM/NIH)
  • Multi-View Sequence-to-Sequence Models with Conversational Structure for Abstractive Dialogue Summarization
  • Few-Shot Learning for Opinion Summarization
  • Learning to Fuse Sentences with Transformers for Summarization
  • Stepwise Extractive Summarization and Planning with Structured Transformers
Nov 17th
60 minutes

Zoom Q&A Session 8A: Information Retrieval and Text Mining

Chair: Matthias Petri
  • CLIRMatrix: A massively large collection of bilingual and multilingual datasets for Cross-Lingual Information Retrieval
  • SLEDGE-Z: A Zero-Shot Baseline for COVID-19 Literature Search
  • Modularized Transfomer-based Ranking Framework
  • Ad-hoc Document Retrieval using Weak-Supervision with BERT and GPT2

Zoom Q&A Session 8B: Interpretability and Analysis of Models for NLP

Chair: Kai-Wei Chang (UCLA)
  • Adversarial Semantic Collisions
  • Learning Explainable Linguistic Expressions with Neural Inductive Logic Programming for Sentence Classification
  • AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts
  • Learning Variational Word Masks to Improve the Interpretability of Neural Text Classifiers

Zoom Q&A Session 8C: Language Generation

Chair: Yannis Konstas (Heriot-Watt)
  • Sparse Text Generation
  • PlotMachines: Outline-Conditioned Generation with Dynamic Plot State Tracking
  • Do sequence-to-sequence VAEs learn global features of sentences?
  • Content Planning for Neural Story Generation with Aristotelian Rescoring
  • Generating Dialogue Responses from a Semantic Latent Space

Zoom Q&A Session 8D: Language Grounding to Vision, Robotics and Beyond

Chair: Florian Metze (Facebook AI)
  • Refer, Reuse, Reduce: Generating Subsequent References in Visual and Conversational Contexts
  • Visually Grounded Compound PCFGs
  • ALICE: Active Learning with Contrastive Natural Language Explanations
  • Room-Across-Room: Multilingual Vision-and-Language Navigation with Dense Spatiotemporal Grounding
  • SSCR: Iterative Language-Based Image Editing via Self-Supervised Counterfactual Reasoning
Nov 17th
120 minutes

Gather Session 3A: Machine Translation and Multilinguality

  • Identifying Elements Essential for BERT’s Multilinguality
  • On Negative Interference in Multilingual Models: Findings and A Meta-Learning Treatment
  • Monolingual Adapters for Zero-Shot Neural Machine Translation
  • Do Explicit Alignments Robustly Improve Multilingual Encoders?
  • From Zero to Hero: On the Limitations of Zero-Shot Language Transfer with Multilingual Transformers
  • Distilling Multiple Domains for Neural Machine Translation
  • Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation
  • A Streaming Approach For Efficient Batched Beam Search
  • Improving Multilingual Models with Language-Clustered Vocabularies
  • Zero-Shot Cross-Lingual Transfer with Meta Learning
  • The Multilingual Amazon Reviews Corpus

Gather Session 3B: NLP Applications

  • Optimus: Organizing Sentences via Pre-trained Modeling of a Latent Space
  • BioMegatron: Larger Biomedical Domain Language Model
  • Text Segmentation by Cross Segment Attention
  • RussianSuperGLUE: A Russian Language Understanding Evaluation Benchmark
  • An Empirical Study of Pre-trained Transformers for Arabic Information Extraction
  • TNT: Text Normalization based Pre-training of Transformers for Content Moderation
  • Methods for Numeracy-Preserving Word Embeddings
  • An Empirical Investigation of Contextualized Number Prediction
  • Modeling the Music Genre Perception across Language-Bound Cultures
  • Joint Estimation and Analysis of Risk Behavior Ratings in Movie Scripts

Gather Session 3C: Machine Learning for NLP

  • Be More with Less: Hypergraph Attention Networks for Inductive Text Classification
  • Entities as Experts: Sparse Memory Access with Entity Supervision
  • Does the Objective Matter? Comparing Training Objectives for Pronoun Resolution
  • On Losses for Modern Language Models
  • We Can Detect Your Bias: Predicting the Political Ideology of News Articles
  • Semantic Label Smoothing for Sequence to Sequence Problems
  • Training for Gibbs Sampling on Conditional Random Fields with Neural Scoring Factors
  • Multilevel Text Alignment with Cross-Document Attention

Gather Session 3D: Computational Social Science and Social Media; Language Generation

  • A Computational Approach to Understanding Empathy Expressed in Text-Based Mental Health Support
  • Modeling Protagonist Emotions for Emotion-Aware Storytelling
  • Help! Need Advice on Identifying Advice
  • Quantifying Intimacy in Language
  • Writing Strategies for Science Communication: Data and Computational Analysis
  • Zero-Shot Crosslingual Sentence Simplification
  • Facilitating the Communication of Politeness through Fine-Grained Paraphrasing
  • On the Reliability and Validity of Detecting Approval of Political Actors in Tweets
  • CAT-Gen: Improving Robustness in NLP Models via Controlled Adversarial Text Generation
  • Seq2Edits: Sequence Transduction Using Span-level Edit Operations
  • Controllable Meaning Representation to Text Generation: Linearization and Data Augmentation Strategies
  • Blank Language Models
  • COD3S: Diverse Generation with Discrete Semantic Signatures

Gather Session 3E: Information Extraction; Phonology, Morphology and Word Segmentation

  • Weakly Supervised Subevent Knowledge Acquisition
  • Biomedical Event Extraction as Sequence Labeling
  • Annotating Temporal Dependency Graphs via Crowdsourcing
  • Introducing a New Dataset for Event Detection in Cybersecurity Texts
  • CHARM: Inferring Personal Attributes from Conversations
  • Event Detection: Gate Diversity and Syntactic Importance Scores for Graph Convolution Neural Networks
  • Severing the Edge Between Before and After: Neural Architectures for Temporal Ordering of Events
  • Automatic Extraction of Rules Governing Morphological Agreement
  • Tackling the Low-resource Challenge for Canonical Segmentation
  • IGT2P: From Interlinear Glossed Texts to Paradigms

Gather Session 3F: Dialog and Interactive Systems; Linguistic Theories, Cognitive Modeling and Psycholinguistics

  • Conversational Semantic Parsing
  • Probing Task-Oriented Dialogue Representation from Language Models
  • End-to-End Slot Alignment and Recognition for Cross-Lingual NLU
  • Discriminative Nearest Neighbor Few-Shot Intent Detection by Transferring Natural Language Inference
  • Simple Data Augmentation with the Mask Token Improves Domain Adaptation for Dialog Act Tagging
  • Low-Resource Domain Adaptation for Compositional Task-Oriented Semantic Parsing
  • Sound Natural: Content Rephrasing in Dialog Systems
  • Structural Supervision Improves Few-Shot Learning and Syntactic Generalization in Neural Language Models
  • Investigating representations of verb bias in neural language models
  • Generating Image Descriptions via Sequential Cross-Modal Alignment Guided by Human Gaze

Gather Session 3G: Question Answering; Syntax: Tagging, Chunking, and Parsing

  • How Much Knowledge Can You Pack Into the Parameters of a Language Model?
  • EXAMS: A Multi-subject High School Examinations Dataset for Cross-lingual and Multilingual Question Answering
  • End-to-End Synthetic Data Generation for Domain Adaptation of Question Answering Systems
  • Multi-Stage Pre-training for Low-Resource Domain Adaptation
  • ISAAQ - Mastering Textbook Questions with Pre-trained Transformers and Bottom-Up and Top-Down Attention
  • SubjQA: A Dataset for Subjectivity and Review Comprehension
  • Keep it Surprisingly Simple: A Simple First Order Graph Based Parsing Model for Joint Morphosyntactic Parsing in Sanskrit
  • Unsupervised Parsing via Constituency Tests
  • Please Mind the Root: Decoding Arborescences for Dependency Parsing
  • Unsupervised Cross-Lingual Part-of-Speech Tagging for Truly Low-Resource Scenarios
  • Unsupervised Parsing with S-DIORA: Single Tree Encoding for Deep Inside-Outside Recursive Autoencoders

Gather Session 3H: Interpretability and Analysis of Models for NLP; Semantics: Sentence-level Semantics, Textual Inference and Other areas

  • Utility is in the Eye of the User: A Critique of NLP Leaderboards
  • An Empirical Investigation Towards Efficient Multi-Domain Language Model Pre-training
  • Analyzing Individual Neurons in Pre-trained Language Models
  • Dissecting Span Identification Tasks with Performance Prediction
  • Assessing Phrasal Representation and Composition in Transformers
  • Analyzing Redundancy in Pretrained Transformer Models
  • GLUCOSE: GeneraLized and COntextualized Story Explanations
  • Character-level Representations Improve DRS-based Semantic Parsing Even in the Age of BERT
  • Infusing Disease Knowledge into BERT for Health Question Answering, Medical Inference and Disease Name Recognition
  • CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models
  • "You are grounded!": Latent Name Artifacts in Pre-trained Language Models
  • Unsupervised Commonsense Question Answering with Self-Talk
  • Reasoning about Goals, Steps, and Temporal Ordering with WikiHow

Gather Session 3I: Demos

  • ARES: A Reading Comprehension Ensembling Service
  • Transformers: State-of-the-Art Natural Language Processing
  • HUMAN: Hierarchical Universal Modular ANnotator
  • DeezyMatch: A Flexible Deep Learning Approach to Fuzzy String Matching
  • InVeRo: Making Semantic Role Labeling Accessible with Intelligible Verbs and Roles
  • ENTYFI: A System for Fine-grained Entity Typing in Fictional Texts

Gather Session 3J: Sponsor Booths

Nov 17th
60 minutes

Keynote II: Rich Caruana

Nov 18th
60 minutes

Zoom Q&A Session 9A: Speech and Multimodality

  • Widget Captioning: Generating Natural Language Description for Mobile User Interface Elements
  • Unsupervised Natural Language Inference via Decoupled Multimodal Contrastive Learning
  • Digital Voicing of Silent Speech
  • Sparse Transcription

Zoom Q&A Session 9B: Machine Learning for NLP

Chair: Wenhu Chen (UCSB)
  • Imitation Attacks and Defenses for Black-box Machine Translation Systems
  • Sequence-Level Mixed Sample Data Augmentation
  • Consistency of a Recurrent Language Model With Respect to Incomplete Decoding
  • An Exploration of Arbitrary-Order Sequence Labeling via Energy-Based Inference Networks
  • Ensemble Distillation for Structured Prediction: Calibrated, Accurate, Fast - Choose Three

Zoom Q&A Session 9C: Sentiment Analysis, Stylistic Analysis, and Argument Mining

Chair: Zhongyu Wei (Fudan University)
  • Inducing Target-Specific Latent Structures for Aspect Sentiment Classification
  • Affective Event Classification with Discourse-enhanced Self-training
  • Deep Weighted MaxSAT for Aspect-based Opinion Extraction
  • Multi-view Story Characterization from Movie Plot Synopses and Reviews
Nov 18th
60 minutes

Zoom Q&A Session 10A: Phonology, Morphology and Word Segmentation

Chair: Ryan Cotterell (ETH Zürich)
  • Mind Your Inflections! Improving NLP for Non-Standard Englishes with Base-Inflection Encoding
  • Measuring the Similarity of Grammatical Gender Systems by Comparing Partitions
  • RethinkCWS: Is Chinese Word Segmentation a Solved Task?
  • Learning to Pronounce Chinese Without a Pronunciation Dictionary

Zoom Q&A Session 10B: Information Extraction

Chair: Jing Huang (JD AI Research)
  • Dynamic Anticipation and Completion for Multi-Hop Reasoning over Sparse Knowledge Graph
  • Knowledge Association with Hyperbolic Knowledge Graph Embeddings
  • Domain Knowledge Empowered Structured Neural Net for End-to-End Event Temporal Relation Extraction
  • TeMP: Temporal Message Passing for Temporal Knowledge Graph Completion

Zoom Q&A Session 10C: Machine Translation and Multilinguality

Chair: Veselin Stoyanov (Facebook AI)
  • Understanding the Difficulty of Training Transformers
  • An Empirical Study of Generation Order for Machine Translation
  • Inference Strategies for Machine Translation with Conditional Masking
  • Reproducible and Efficient Benchmarks for Hyperparameter Optimization of Neural Machine Translation Systems

Zoom Q&A Session 10D: Question Answering

Chair: Danqi Chen (Princeton)
  • AmbigQA: Answering Ambiguous Open-domain Questions
  • Tell Me How to Ask Again: Question Data Augmentation with Controllable Rewriting in Continuous Space
  • Training Question Answering Models From Synthetic Data
  • Few-Shot Complex Knowledge Base Question Answering via Meta Reinforcement Learning
Nov 18th
120 minutes

Gather Session 4A: Machine Translation and Multilinguality

  • Iterative Domain-Repaired Back-Translation
  • Dynamic Data Selection and Weighting for Iterative Back-Translation
  • Revisiting Modularized Multilingual NMT to Meet Industrial Demands
  • LAReQA: Language-Agnostic Answer Retrieval from a Multilingual Pool
  • OCR Post Correction for Endangered Language Texts
  • X-FACTR: Multilingual Factual Knowledge Retrieval from Pretrained Language Models
  • CCAligned: A Massive Collection of Cross-Lingual Web-Document Pairs
  • Localizing Open-Ontology QA Semantic Parsers in a Day Using Machine Translation
  • Interactive Refinement of Cross-Lingual Word Embeddings
  • Exploiting Sentence Order in Document Alignment
  • XGLUE: A New Benchmark Datasetfor Cross-lingual Pre-training, Understanding and Generation

Gather Session 4B: Machine Learning for NLP

  • Structure Aware Negative Sampling in Knowledge Graphs
  • Neural Mask Generator: Learning to Generate Adaptive Word Maskings for Language Model Adaptation
  • Autoregressive Knowledge Distillation through Imitation Learning
  • Recall and Learn: Fine-tuning Deep Pretrained Language Models with Less Forgetting
  • T3: Tree-Autoencoder Constrained Adversarial Text Generation for Targeted Attack
  • Structured Pruning of Large Language Models
  • Effective Unsupervised Domain Adaptation with Adversarially Trained Language Models
  • BAE: BERT-based Adversarial Examples for Text Classification
  • Adversarial Self-Supervised Data-Free Distillation for Text Classification
  • BERT-ATTACK: Adversarial Attack Against BERT Using BERT
  • The Thieves on Sesame Street are Polyglots - Extracting Multilingual Models from Monolingual APIs

Gather Session 4C: Information Extraction

  • Coarse-to-Fine Pre-training for Named Entity Recognition
  • Exploring and Evaluating Attributes, Values, and Structures for Entity Alignment
  • Simple and Effective Few-Shot Named Entity Recognition with Structured Nearest Neighbor Learning
  • Learning Structured Representations of Entity Names using ActiveLearning and Weak Supervision
  • Entity Enhanced BERT Pre-training for Chinese NER
  • Scalable Zero-shot Entity Linking with Dense Entity Retrieval
  • A Dataset for Tracking Entities in Open Domain Procedural Text
  • Design Challenges in Low-resource Cross-lingual Entity Linking
  • Efficient One-Pass End-to-End Entity Linking for Questions
  • LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention

Gather Session 4D: Dialog and Interactive Systems

  • Towards Persona-Based Empathetic Conversational Models
  • Personal Information Leakage Detection in Conversations
  • Response Selection for Multi-Party Conversations with Dynamic Topic Tracking
  • Regularizing Dialogue Generation by Imitating Implicit Scenarios
  • MovieChats: Chat like Humans in a Closed Domain
  • Conundrums in Entity Coreference Resolution: Making Sense of the State of the Art
  • Semantic Role Labeling Guided Multi-turn Dialogue ReWriter
  • Continuity of Topic, Interaction, and Query: Learning to Quote in Online Conversations
  • Profile Consistency Identification for Open-domain Dialogue Agents

Gather Session 4E: Sentiment Analysis, Stylistic Analysis, and Argument Mining

  • A Multi-Task Incremental Learning Framework with Category Name Embedding for Aspect-Category Sentiment Analysis
  • Train No Evil: Selective Masking for Task-Guided Pre-Training
  • SentiLARE: Sentiment-Aware Language Representation Learning with Linguistic Knowledge
  • Weakly-Supervised Aspect-Based Sentiment Analysis via Joint Aspect-Sentiment Topic Embedding
  • APE: Argument Pair Extraction from Peer Review and Rebuttal via Multi-task Learning
  • Diversified Multiple Instance Learning for Document-Level Multi-Aspect Sentiment Classification
  • Identifying Exaggerated Language
  • Unified Feature and Instance Based Domain Adaptation for Aspect-Based Sentiment Analysis

Gather Session 4F: Computational Social Science and Social Media; Semantics: Sentence-level Semantics, Textual Inference and Other areas

  • Multilingual Offensive Language Identification with Cross-lingual Embeddings
  • Solving Historical Dictionary Codes with a Neural Language Model
  • Toward Micro-Dialect Identification in Diaglossic and Code-Switched Environments
  • Investigating African-American Vernacular English in Transformer-Based Text Generation
  • Grounded Adaptation for Zero-shot Executable Semantic Parsing
  • An Imitation Game for Learning Semantic Parsers from User Interaction
  • IGSQL: Database Schema Interaction Graph Based Neural Model for Context-Dependent Text-to-SQL Generation
  • "What Do You Mean by That?" A Parser-Independent Interactive Approach for Enhancing Text-to-SQL
  • DuSQL: A Large-Scale and Pragmatic Chinese Text-to-SQL Dataset
  • Mention Extraction and Linking for SQL Query Generation
  • Re-examining the Role of Schema Linking in Text-to-SQL

Gather Session 4G: NLP Applications; Semantics: Lexical Semantics

  • An Element-aware Multi-representation Model for Law Article Prediction
  • Recurrent Event Network: Autoregressive Structure Inferenceover Temporal Knowledge Graphs
  • Multi-resolution Annotations for Emoji Prediction
  • Less is More: Attention Supervision with Counterfactuals for Text Classification
  • MODE-LSTM: A Parameter-efficient Recurrent Network with Multi-Scale for Sentence Classification
  • Assessing the Helpfulness of Learning Materials with Inference-Based Learner-Like Agent
  • HSCNN: A Hybrid-Siamese Convolutional Neural Network for Extremely Imbalanced Multi-label Text Classification
  • Multi-Stage Pre-training for Automated Chinese Essay Scoring
  • When Hearst Is not Enough: Improving Hypernymy Detection from Corpus with Distributional Models
  • Interpreting Open-Domain Modifiers: Decomposition of Wikipedia Categories into Disambiguated Property-Value Pairs
  • A Synset Relation-enhanced Framework with a Try-again Mechanism for Word Sense Disambiguation

Gather Session 4H: Discourse and Pragmatics; Language Generation

  • BERT-enhanced Relational Sentence Ordering Network
  • Online Conversation Disentanglement with Pointer Networks
  • VCDM: Leveraging Variational Bi-encoding and Deep Contextualized Word Representations for Improved Definition Modeling
  • Generating similes effortlessly like a Pro: A Style Transfer Approach for Simile Generation
  • STORIUM: A Dataset and Evaluation Platform for Machine-in-the-Loop Story Generation
  • Substance over Style: Document-Level Targeted Content Transfer
  • Improving Low Compute Language Modeling with In-Domain Embedding Initialisation
  • Template Guided Text Generation for Task-Oriented Dialogue
  • MOCHA: A Dataset for Training and Evaluating Generative Reading Comprehension Metrics
  • Inquisitive Question Generation for High Level Text Comprehension

Gather Session 4I: Interpretability and Analysis of Models for NLP; Syntax: Tagging, Chunking, and Parsing

  • Asking without Telling: Exploring Latent Ontologies in Contextual Representations
  • Pretrained Language Model Embryology: The Birth of ALBERT
  • Learning Music Helps You Read: Using Transfer to Study Linguistic Structure in Language Models
  • What Do Position Embeddings Learn? An Empirical Study of Pre-Trained Language Model Positional Encoding
  • Birds have four legs?! NumerSense: Probing Numerical Commonsense Knowledge of Pre-Trained Language Models
  • AIN: Fast and Accurate Sequence Labeling with Approximate Inference Network
  • HIT: Nested Named Entity Recognition via Head-Tail Pair and Token Interaction
  • Supertagging Combinatory Categorial Grammar with Attentive Graph Convolutional Networks
  • DAGA: Data Augmentation with a Generation Approach for Low-resource Tagging Tasks
  • Interpretable Multi-dataset Evaluation for Named Entity Recognition
  • Adversarial Semantic Decoupling for Recognizing Open-Vocabulary Slots

Gather Session 4J: Question Answering; Summarization

  • Multi-hop Inference for Question-driven Summarization
  • Towards Interpretable Reasoning over Paragraph Effects in Situation
  • Question Directed Graph Attention Network for Numerical Reasoning over Text
  • Dense Passage Retrieval for Open-Domain Question Answering
  • Distilling Structured Knowledge for Text-Based Relational Reasoning
  • Diverse, Controllable, and Keyphrase-Aware: A Corpus and Method for News Multi-Headline Generation
  • Factual Error Correction for Abstractive Summarization Models
  • Compressive Summarization with Plausibility and Salience Modeling
  • Multi-XScience: A Large-scale Dataset for Extreme Multi-document Summarization of Scientific Articles
  • Understanding Neural Abstractive Summarization Models via Uncertainty
  • Better Highlighting: Creating Sub-Sentence Summary Highlights
  • Summarizing Text on Any Aspects: A Knowledge-Informed Weakly-Supervised Approach

Gather Session 4K: Demos

  • NeuralQA: A Usable Library for Question Answering (Contextual Query Expansion + BERT) on Large Datasets
  • Youling: an AI-assisted Lyrics Creation System
  • TextAttack: A Framework for Adversarial Attacks, Data Augmentation, and Adversarial Training in NLP
  • Easy, Reproducible and Quality-Controlled Data Collection with CROWDAQ
  • NeuSpell: A Neural Spelling Correction Toolkit

Gather Session 4L: Sponsor Booths

Nov 18th
60 minutes

Zoom Q&A Session 11A: Interpretability and Analysis of Models for NLP

Chair: Yonatan Belinkov
  • Compositional and Lexical Semantics in RoBERTa, BERT and DistilBERT: A Case Study on CoQA
  • Attention is Not Only a Weight: Analyzing Transformers with Vector Norms
  • F1 is Not Enough! Models and Evaluation Towards User-Centered Explainable Question Answering
  • On the Ability and Limitations of Transformers to Recognize Formal Languages

Zoom Q&A Session 11B: NLP Applications

Chair: Shashi Narayan (Google)
  • An Unsupervised Joint System for Text Generation from Knowledge Graphs and Semantic Parsing
  • DGST: a Dual-Generator Network for Text Style Transfer
  • A Knowledge-Aware Sequence-to-Tree Network for Math Word Problem Solving
  • Generating Fact Checking Briefs
  • Improving the Efficiency of Grammatical Error Correction with Erroneous Span Detection and Correction

Zoom Q&A Session 11C: Question Answering

Chair: Alice Oh
  • Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehension
  • Coreferential Reasoning Learning for Language Representation
  • Is Graph Structure Necessary for Multi-hop Question Answering?
  • oLMpics - On what Language Model Pre-training Captures

Zoom Q&A Session 11D: Semantics: Lexical Semantics

Chair: Aline Villavicencio
  • XL-WiC: A Multilingual Benchmark for Evaluating Semantic Contextualization
  • Generationary or "How We Went beyond Word Sense Inventories and Learned to Gloss"
  • Probing Pretrained Language Models for Lexical Semantics
  • Nurse is Closer to Woman than Surgeon? Mitigating Gender-Biased Proximities in Word Embeddings
Nov 18th
60 minutes

Zoom Q&A Session 12A: Dialog and Interactive Systems

Chair: Sakriani Sakti (NAIST/RIKEN AIP)
  • Cross-lingual Spoken Language Understanding with Regularized Representation Alignment
  • SLURP: A Spoken Language Understanding Resource Package
  • Neural Conversational QA: Learning to Reason vs Exploiting Patterns
  • Improving Dialog Evaluation with a Multi-reference Adversarial Dataset and Large Scale Pretraining

Zoom Q&A Session 12B: Information Extraction

Chair: Aurélie Névéol (CNRS, LIMSI)
  • Counterfactual Generator: A Weakly-Supervised Method for Named Entity Recognition
  • Understanding Procedural Text using Interactive Entity Networks
  • A Rigorous Study on Named Entity Recognition: Can Fine-tuning Pretrained Model Lead to the Promised Land?
  • Nested Named Entity Recognition via Second-best Sequence Learning and Decoding

Zoom Q&A Session 12C: Machine Learning for NLP

Chair: Sebastian Ruder
  • DyERNIE: Dynamic Evolution of Riemannian Manifold Embeddings for Temporal Knowledge Graph Completion
  • Embedding Words in Non-Vector Space with Unsupervised Graph Learning
  • Debiasing knowledge graph embeddings
  • Message Passing for Hyper-Relational Knowledge Graphs
  • PERL: Pivot-based Domain Adaptation for Pre-trained Deep Contextualized Embedding Models

Zoom Q&A Session 12D: Sentiment Analysis, Stylistic Analysis, and Argument Mining

Chair: Rui Xia
  • Relation-aware Graph Attention Networks with Relational Position Encodings for Emotion Recognition in Conversations
  • BERT Knows Punta Cana is not just beautiful, it's gorgeous: Ranking Scalar Adjectives with Contextualised Representations
  • Feature Adaptation of Pre-Trained Language Models across Languages and Domains with Robust Self-Training
  • Textual Data Augmentation for Efficient Active Learning on Tiny Datasets
Nov 18th
60 minutes

Keynote III: Janet B. Pierrehumbert

Nov 18th
60 minutes

Zoom Q&A Session 13A: Discourse and Pragmatics

Chair: Vincent Ng
  • "I'd rather just go to bed": Understanding Indirect Answers
  • PowerTransformer: Unsupervised Controllable Revision for Biased Language Correction
  • MEGA RST Discourse Treebanks with Structure and Nuclearity from Scalable Distant Sentiment Supervision
  • Centering-based Neural Coherence Modeling with Hierarchical Discourse Segments
  • Keeping Up Appearances: Computational Modeling of Face Acts in Persuasion Oriented Discussions

Zoom Q&A Session 13B: NLP Applications

Chair: Thamar Solorio
  • To Schedule or not to Schedule: Extracting Task Specific Temporal Entities and Associated Negation Constraints
  • Predicting In-game Actions from Interviews of NBA Players
  • An Empirical Study on Large-Scale Multi-Label Text Classification Including Few and Zero-Shot Labels
  • Which *BERT? A Survey Organizing Contextualized Encoders
  • Fact or Fiction: Verifying Scientific Claims

Zoom Q&A Session 13C: Semantics: Sentence-level Semantics, Textual Inference and Other areas

Chair: Annemarie Friedrich (Bosch)
  • Semantic Role Labeling as Syntactic Dependency Parsing
  • PARADE: A New Dataset for Paraphrase Identification Requiring Computer Science Domain Knowledge
  • Causal Inference of Script Knowledge
  • Towards Debiasing NLU Models from Unknown Biases

Zoom Q&A Session 13D: Syntax: Tagging, Chunking, and Parsing

Chair: Ryan Cotterell (ETH Zürich)
  • Tractable Lexical-Functional Grammar
  • Efficient Outside Computation
  • Consistent Unsupervised Estimators for Anchored PCFGs
  • The Return of Lexical Dependencies: Neural Lexicalized PCFGs
  • On the Role of Supervision in Unsupervised Constituency Parsing
Nov 18th
60 minutes

Zoom Q&A Session 14A: Machine Translation and Multilinguality

Chair: Julia Kreutzer (Google)
  • Language Model Prior for Low-Resource Neural Machine Translation
  • Detecting Word Sense Disambiguation Biases in Machine Translation for Model-Agnostic Adversarial Attacks
  • MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfer
  • Translation Artifacts in Cross-lingual Transfer Learning
  • Consistent Transcription and Translation of Speech

Zoom Q&A Session 14B: Computational Social Science and Social Media

Chair: Dong Nguyen
  • A Time-Aware Transformer Based Model for Suicide Ideation Detection on Social Media
  • Weakly Supervised Learning of Nuanced Frames for Analyzing Polarization in News Media
  • Where Are the Facts? Searching for Fact-checked Information to Alleviate the Spread of Fake News
  • Fortifying Toxic Speech Detectors Against Veiled Toxicity
  • Explainable Automated Fact-Checking for Public Health Claims

Zoom Q&A Session 14C: Machine Learning for NLP

Chair: Yishu Miao (Imperial College London)
  • Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning
  • A Neural Generative Model for Joint Learning Topics and Topic-Specific Word Embeddings
  • Topic Modeling in Embedding Spaces
  • DORB: Dynamically Optimizing Multiple Rewards with Bandits

Zoom Q&A Session 14D: Information Extraction

Chair: Heng Ji (UIUC & Amazon)
  • MedFilter: Improving Extraction of Task-relevant Utterances through Integration of Discourse Structure and Ontological Knowledge
  • Hierarchical Evidence Set Modeling for Automated Fact Extraction and Verification
  • Program Enhanced Fact Verification with Verbalization and Graph Attention Network
  • Constrained Fact Verification for FEVER
  • Entity Linking in 100 Languages
Nov 18th
120 minutes

Gather Session 5A: Machine Learning for NLP

  • PatchBERT: Just-in-Time, Out-of-Vocabulary Patching
  • On the importance of pre-training data volume for compact language models
  • Plug and Play Autoencoders for Conditional Text Generation
  • Exploring and Predicting Transferability across NLP Tasks
  • To BERT or Not to BERT: Comparing Task-specific and Task-agnostic Semi-Supervised Approaches for Sequence Tagging
  • Cold-start Active Learning through Self-supervised Language Modeling
  • Active Learning for BERT: An Empirical Study
  • Transformer Based Multi-Source Domain Adaptation
  • Vector-Vector-Matrix Architecture: A Novel Hardware-Aware Framework for Low-Latency Inference in NLP Applications

Gather Session 5B: Semantics: Sentence-level Semantics, Textual Inference and Other areas

  • Discriminatively-Tuned Generative Classifiers for Robust Natural Language Inference
  • New Protocols and Negative Results for Textual Entailment Data Collection
  • The Curse of Performance Instability in Analysis Datasets: Consequences, Source, and Suggestions
  • Universal Natural Language Processing with Limited Annotations: Try Few-shot Textual Entailment as a Start
  • ConjNLI: Natural Language Inference Over Conjunctive Sentences
  • Data and Representation for Turkish Natural Language Inference
  • Multitask Learning for Cross-Lingual Transfer of Broad-coverage Semantic Dependencies
  • Precise Task Formalization Matters in Winograd Schema Evaluations
  • Avoiding the Hypothesis-Only Bias in Natural Language Inference via Ensemble Adversarial Training

Gather Session 5C: NLP Applications

  • Chapter Captor: Text Segmentation in Novels
  • Authorship Attribution for Neural Text Generation
  • NwQM: A neural quality assessment framework for Wikipedia
  • Towards Modeling Revision Requirements in wikiHow Instructions
  • Natural Language Processing for Achieving Sustainable Development: the Case of Neural Labelling to Enhance Community Profiling
  • HABERTOR: An Efficient and Effective Deep Hatespeech Detector
  • Competence-Level Prediction and Resume & Job Description Matching Using Context-Aware Transformer Models
  • Grammatical Error Correction in Low Error Density Domains: A New Benchmark and Analyses

Gather Session 5D: Information Extraction

  • Learning Collaborative Agents with Rule Guidance for Knowledge Graph Reasoning
  • Exploring Contextualized Neural Language Models for Temporal Dependency Parsing
  • Systematic Comparison of Neural Architectures and Training Approaches for Open Information Extraction
  • SeqMix: Augmenting Active Sequence Labeling via Sequence Mixup
  • AxCell: Automatic Extraction of Results from Machine Learning Papers
  • Knowledge-guided Open Attribute Value Extraction with Reinforcement Learning
  • DualTKB: A Dual Learning Bridge between Text and Knowledge Base
  • Incremental Neural Coreference Resolution in Constant Memory

Gather Session 5E: Question Answering; Sentiment Analysis, Stylistic Analysis, and Argument Mining

  • Hierarchical Graph Network for Multi-hop Question Answering
  • A Simple Yet Strong Pipeline for HotpotQA
  • Is Multihop QA in DiRe Condition? Measuring and Reducing Disconnected Reasoning
  • Unsupervised Question Decomposition for Question Answering
  • SRLGRN: Semantic Role Labeling Graph Reasoning Network
  • CancerEmo: A Dataset for Fine-Grained Emotion Detection
  • Exploring the Role of Argument Structure in Online Debate Persuasion
  • Zero-Shot Stance Detection: A Dataset and Model using Generalized Topic Representations
  • Sentiment Analysis of Tweets using Heterogeneous Multi-layer Network Representation and Embedding
  • Introducing Syntactic Structures into Target Opinion Word Extraction with Deep Learning
  • EmoTag1200 👍: Understanding the Association between Emojis 😄 and Emotions 😻
  • MIME: MIMicking Emotions for Empathetic Response Generation

Gather Session 5F: Language Grounding to Vision, Robotics and Beyond; Speech and Multimodality

  • Experience Grounds Language
  • Keep CALM and Explore: Language Models for Action Generation in Text-based Games
  • CapWAP: Image Captioning with a Purpose
  • What is More Likely to Happen Next? Video-and-Language Future Event Prediction
  • X-LXMERT: Paint, Caption and Answer Questions with Multi-Modal Transformers
  • Towards Understanding Sample Variance in Visually Grounded Language Generation: Evaluations and Observations
  • Beyond Instructional Videos: Probing for More Diverse Visual-Textual Grounding on YouTube
  • The importance of fillers for text representations of speech transcripts
  • The role of context in neural pitch accent detection in English
  • VolTAGE: Volatility Forecasting via Text Audio Fusion with Graph Convolution Networks for Earnings Calls
  • Effectively pretraining a speech translation decoder with Machine Translation data

Gather Session 5G: Language Generation; Semantics: Lexical Semantics

  • KGPT: Knowledge-Grounded Pre-Training for Data-to-Text Generation
  • POINTER: Constrained Progressive Text Generation via Insertion-based Generative Pre-training
  • Unsupervised Text Style Transfer with Padded Masked Language Models
  • PALM: Pre-training an Autoencoding&Autoregressive Language Model for Context-conditioned Generation
  • Gradient-guided Unsupervised Lexically Constrained Text Generation
  • TeaForN: Teacher-Forcing with N-grams
  • Deconstructing word embedding algorithms
  • Compositional Demographic Word Embeddings
  • Sparsity Makes Sense: Word Sense Disambiguation Using Sparse Contextualized Word Representations

Gather Session 5H: Dialog and Interactive Systems; Discourse and Pragmatics

  • Iterative Feature Mining for Constraint-Based Data Collection to Increase Data Diversity and Model Robustness
  • Conversational Semantic Parsing for Dialog State Tracking
  • doc2dial: A Goal-Oriented Document-Grounded Dialogue Dataset
  • Interview: Large-scale Modeling of Media Dialog with Discourse Patterns and Knowledge Grounding
  • INSPIRED: Toward Sociable Recommendation Dialog Systems
  • Information Seeking in the Spirit of Learning: A Dataset for Conversational Curiosity
  • Queens are Powerful too: Mitigating Gender Bias in Dialogue Generation
  • Learning to Ignore: Long Document Coreference with Bounded Memory Neural Networks
  • Revealing the Myth of Higher-Order Inference in Coreference Resolution
  • Pre-training Mention Representations in Coreference Models

Gather Session 5I: Information Retrieval and Text Mining; Summarization

  • SynSetExpan: An Iterative Framework for Joint Entity Set Expansion and Synonym Discovery
  • Evaluating the Calibration of Knowledge Graph Embeddings for Trustworthy Link Prediction
  • Text Graph Transformer for Document Classification
  • CoDEx: A Comprehensive Knowledge Graph Completion Benchmark
  • META: Metadata-Empowered Weak Supervision for Text Classification
  • Towards More Accurate Uncertainty Estimation In Text Classification
  • A Preliminary Exploration of GANs for Keyphrase Generation
  • TESA: A Task in Entity Semantic Aggregation for Abstractive Summarization
  • MLSUM: The Multilingual Summarization Corpus
  • Intrinsic Evaluation of Summarization Datasets

Gather Session 5J: Demos

  • A Technical Question Answering System with Transfer Learning
  • Agent Assist through Conversation Analysis
  • LibKGE - A knowledge graph embedding library for reproducible research
  • RoFT: A Tool for Evaluating Human Detection of Machine-Generated Text
  • A Data-Centric Framework for Composable NLP Workflows
  • CoRefi: A Crowd Sourcing Suite for Coreference Annotation

Gather Session 5K: Sponsor Booths

Nov 18th
60 minutes

Zoom Q&A Session 15A: Information Retrieval and Text Mining

Chair: Lifu Huang (Virginia Tech)
  • Exploiting Structured Knowledge in Text via Graph-Guided Representation Learning
  • Named Entity Recognition Only from Word Embeddings
  • Text Classification Using Label Names Only: A Language Model Self-Training Approach
  • Neural Topic Modeling with Cycle-Consistent Adversarial Training

Zoom Q&A Session 15B: NLP Applications

  • Data Boost: Text Data Augmentation Through Reinforcement Learning Guided Conditional Generation
  • A State-independent and Time-evolving Network for Early Rumor Detection in Social Media
  • PyMT5: multi-mode translation of natural language and Python code with transformers
  • PathQG: Neural Question Generation from Facts
  • What time is it? Temporal Analysis of Novels

Zoom Q&A Session 15C: Semantics: Sentence-level Semantics, Textual Inference and Other areas

Chair: Siva Reddy (McGill/MILA)
  • COGS: A Compositional Generalization Challenge Based on Semantic Interpretation
  • An Analysis of Natural Language Inference Benchmarks through the Lens of Negation
  • On the Sentence Embeddings from Pre-trained Language Models
  • An Empirical Study on Robustness to Spurious Correlations using Pre-trained Language Models
  • What Can We Learn from Collective Human Opinions on Natural Language Inference Data?

Zoom Q&A Session 15D: Language Generation

  • Improving Text Generation with Student-Forcing Optimal Transport
  • UNION: An Unreferenced Metric for Evaluating Open-ended Story Generation
  • F^2-Softmax: Diversifying Neural Text Generation via Frequency Factorized Softmax
  • Partially-Aligned Data-to-Text Generation with Distant Supervision
  • How Can We Know What Language Models Know
Nov 19th
60 minutes

Zoom Q&A Session 16A: Dialog and Interactive Systems

Chair: Linfeng Song (Tencent AI Lab)
  • Like hiking? You probably enjoy nature: Persona-grounded Dialog with Commonsense Expansions
  • A Probabilistic End-To-End Task-Oriented Dialog Model with Latent Belief States towards Semi-Supervised Learning
  • The World is Not Binary: Learning to Rank with Grayscale Data for Dialogue Response Selection
  • GRADE: Automatic Graph-Enhanced Coherence Metric for Evaluating Open-Domain Dialogue Systems
  • MedDialog: Large-scale Medical Dialogue Datasets

Zoom Q&A Session 16B: Interpretability and Analysis of Models for NLP

Chair: Lei Li (ByteDance)
  • An information theoretic view on selecting linguistic probes
  • With Little Power Comes Great Responsibility
  • Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamics
  • Evaluating and Characterizing Human Rationales

Zoom Q&A Session 16C: Summarization

Chair: Logan Lebanoff (UCF)