EACL 2021

683 papers
The 2021 edition of the EACL conference presented 313 papers, with a strong emphasis on model robustness, factual consistency, and multilingualism. Research directions included mitigating positional bias in long document processing and improving factual faithfulness in abstractive summarization. Multilingual studies focused on language-agnostic retrieval, internal English pivoting in large language models, and addressing linguistic inequality within the field. The proceedings also highlighted specialized applications in scientific NLP, including the development of the TDMSci corpus for entity tagging and the ParaSCI dataset for paraphrase generation. Methodological trends involved iterative reasoning for multi-hop tasks, syntactically controlled data augmentation, and the use of span-prediction for emotion classification. Efficiency and interpretability were addressed through techniques such as early exiting for BERT models and quantifying token influence via systematic perturbation. Prolific authors contributing to these themes included Ernie Chang, Vera Demberg, Timothy Baldwin, and Danushka Bollegala.

Top Research Directions

Research directions most represented by EACL 2021 papers.

Mitigating Data Scarcity in Low Resource Language Processing 43
Why Transformers Struggle with Code Mixed Social Media Text 40
Adapting Natural Language Processing to Dialectal Diversity 28
Computational Detection of Prosocial Online Discourse 21
Multilingual Transformers for Code Mixed Text 18
The Performance Advantage of Localized Language Pretraining 18
Shared Representation Learning via Simultaneous Multi-Task Training 17
Zero-Shot Cross-Lingual Transfer in Multilingual Language Models 16
Modeling Semantic Inversion in Sarcasm Detection 15
Statistical Term Weighting for Text Vectorization 14
Script Normalization for Multilingual and Code Mixed Text 12
Discriminating Between Closely Related Languages and Dialects 11
TF-IDF Feature Extraction with Traditional Machine Learning Classifiers 10
Offensive Language Detection in Code-Mixed Dravidian Languages 10
Addressing Class Imbalance in Text Classification 10

Top Authors

AuthorPapers
Anders Søgaard4
Hinrich Schuetze4
Timothy Baldwin4
William Yang Wang4
Asli Celikyilmaz3
Danushka Bollegala3
Gholamreza Haffari3
Hiroya Takamura3
Pontus Stenetorp3
Ryan Cotterell3
Trevor Cohn3
Vera Demberg3
Abhilasha Ravichander2
Akiko Aizawa2
Alexander Panchenko2
Andrew McCallum2
An Yan2
Artem Shelmanov2
Carolyn Rose2
Dan Roth2

Impactful Papers

Papers from EACL 2021 that shaped research clusters.

Foundational

Influential

Papers

Title
Cross-Cultural Similarity Features for Cross-Lingual Transfer Learning of Pragmatically Motivated Tasks
Cross-lingual Contextualized Topic Models with Zero-shot Learning
Cross-lingual Entity Alignment with Incidental Supervision
Cross-Lingual Transfer with MAML on Trees
Cross-lingual Visual Pre-training for Multimodal Machine Translation
Cross-Topic Rumor Detection using Topic-Mixtures
cs@DravidianLangTech-EACL2021: Offensive Language Identification Based On Multilingual BERT Model
CTC-based Compression for Direct Speech Translation
C-Test Collector: A Proficiency Testing Application to Collect Training Data for C-Tests
CUSATNLP@DravidianLangTech-EACL2021:Language Agnostic Classification of Offensive Content in Tweets
DanNet2: Extending the coverage of adjectives in DanNet based on thesaurus data (project presentation)
Data Augmentation for Hypernymy Detection
Data Augmentation for Voice-Assistant NLU using BERT-based Interchangeable Rephrase
Data Strategies for Low-Resource Grammatical Error Correction
Debiasing Pre-trained Contextualised Embeddings
DeepBlueAI at WANLP-EACL2021 task 2: A Deep Ensemble-based Method for Sarcasm and Sentiment Detection in Arabic
Deep Multi-Task Model for Sarcasm Detection and Sentiment Analysis in Arabic Language
Deep Subjecthood: Higher-Order Grammatical Features in Multilingual BERT
Dependency Parsing Evaluation for Low-resource Spontaneous Speech
Dependency parsing with structure preserving embeddings
Detecting Extraneous Content in Podcasts
Detecting Inappropriate Messages on Sensitive Topics that Could Harm a Company’s Reputation
Detecting Post-Edited References and Their Effect on Human Evaluation
Detecting Scenes in Fiction: A new Segmentation Task
Development of Conversational AI for Sleep Coaching Programme
dhivya-hope-detection@LT-EDI-EACL2021: Multilingual Hope Speech Detection for Code-mixed and Transliterated Texts
Dialect Identification in Nuanced Arabic Tweets Using Farasa Segmentation and AraBERT
DiaLex: A Benchmark for Evaluating Multidialectal Arabic Word Embeddings
Dialogue Act-based Breakdown Detection in Negotiation Dialogues
Dictionary-based Debiasing of Pre-trained Word Embeddings
Disambiguatory Signals are Stronger in Word-initial Positions
Discourse-Aware Unsupervised Summarization for Long Scientific Documents
Discourse Understanding and Factual Consistency in Abstractive Summarization
Discriminating Homonymy from Polysemy in Wordnets: English, Spanish and Polish Nouns
Disentangling Document Topic and Author Gender in Multiple Languages: Lessons for Adversarial Debiasing
Disfluency Correction using Unsupervised and Semi-supervised Learning
DISK-CSV: Distilling Interpretable Semantic Knowledge with a Class Semantic Vector
Diverse Adversaries for Mitigating Bias in Training
DLRG@DravidianLangTech-EACL2021: Transformer based approachfor Offensive Language Identification on Code-Mixed Tamil
Document-level grammatical error correction
Does She Wink or Does She Nod? A Challenging Benchmark for Evaluating Word Understanding of Language Models
Does the Order of Training Samples Matter? Improving Neural Data-to-Text Generation with Curriculum Learning
Does Typological Blinding Impede Cross-Lingual Sharing?
Domain Adaptation for NMT via Filtered Iterative Back-Translation
Domain Expert Platform for Goal-Oriented Dialog Collection
Do Multi-Hop Question Answering Systems Know How to Answer the Single-Hop Sub-Questions?
Don't Change Me! User-Controllable Selective Paraphrase Generation
DOSA: Dravidian Code-Mixed Offensive Span Identification Dataset
Do Syntax Trees Help Pre-trained Transformers Extract Information?
Do we read what we hear? Modeling orthographic influences on spoken word recognition

« prev Page 4/14 next »