Papers by Abhijnan Nath
2*n is better than n2: Decomposing Event Coreference Resolution into Two Tractable Problems (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for training coreference systems sample from a largely skewed distribution, making it difficult to learn coreference beyond surface matching. |
| Approach: | They propose a heuristic to efficiently filter out a large number of non-coreferent pairs and a training approach on a balanced set of coreferent and non- coreferente mention pairs. |
| Outcome: | The proposed approach significantly reduces compute requirements on two popular ECR datasets while reducing the computational complexity. |
A Generalized Method for Automated Multilingual Loanword Detection (2022.coling-1)
Copied to clipboard
Abhijnan Nath, Sina Mahdipour Saravani, Ibrahim Khebour, Sheikh Mannan, Zihui Li, Nikhil Krishnaswamy
| Challenge: | Loanwords are words incorporated from one language into another without translation . authors present a method to automatically detect loanwords across language pairs . |
| Approach: | They propose a method to automatically detect loanwords across language pairs . they incorporate edit distance, semantic similarity measures, phonetic alignment . |
| Outcome: | The proposed method outperforms existing methods on single-pair loanword detection tasks and can generalize to unseen language pairs with sufficient data. |
DPL: Diverse Preference Learning Without A Reference Model (2025.naacl-long)
Copied to clipboard
Abhijnan Nath, Andrey Volozin, Saumajit Saha, Albert Aristotle Nanda, Galina Grunin, Rahul Bhotika, Nikhil Krishnaswamy
| Challenge: | Existing methods to direct preference alignment do not utilize diversity in preference annotations which limits their applicability. |
| Approach: | They propose a reference-model-free method that learns a baseline desirability in LLM responses while being robust to the diversity of preference annotations. |
| Outcome: | The proposed method learns a baseline desirability in LLM responses while being robust to the diversity of preference annotations. |
Okay, Let’s Do This! Modeling Event Coreference with Generated Rationales and Knowledge Distillation (2024.naacl-long)
Copied to clipboard
| Challenge: | Recent work shows that generative large language models (LLMs) can be used to solve cross-document coreference problems. |
| Approach: | They propose rationale-oriented event clustering and knowledge distillation methods for event coreference scoring that leverage enriched information from the FTRs for improved CDCR. |
| Outcome: | The proposed model achieves SOTA B3 F1 on the ECB+ and GVC corpora without additional annotation or expensive document clustering. |
Multimodal Cross-Document Event Coreference Resolution Using Linear Semantic Transfer and Mixed-Modality Ensembles (2024.lrec-main)
Copied to clipboard
Abhijnan Nath, Huma Jamil, Shafiuddin Rehan Ahmed, George Arthur Baker, Rahul Ghosh, James H. Martin, Nathaniel Blanchard, Nikhil Krishnaswamy
| Challenge: | Existing methods for cross-document coreference resolution do not provide images for all mentions of events. |
| Approach: | They propose a multimodal cross-document event coreference resolution method that integrates visual and textual cues with a simple linear map between vision and language models. |
| Outcome: | The proposed method improves on a popular ECB+ and AIDA datasets. |
“Any Other Thoughts, Hedgehog?” Linking Deliberation Chains in Collaborative Dialogues (2024.findings-emnlp)
Copied to clipboard
Abhijnan Nath, Videep Venkatesha, Mariah Bradford, Avyakta Chelle, Austin Youngren, Carlos Mabrey, Nathaniel Blanchard, Nikhil Krishnaswamy
| Challenge: | Recent advances in generative AI have raised the possibility of systems that follow and interact with multiparty dialogues. |
| Approach: | They propose a graph-based framework for probing questions in collaborative dialogues that models causal relations between probing and causal utterances and the links between them. |
| Outcome: | The proposed framework compares to baselines and stronger coreference approaches and establishes a standard of performance in this novel task. |
Frictional Agent Alignment Framework: Slow Down and Don’t Break Things (2025.acl-long)
Copied to clipboard
| Challenge: | Common preference alignment methods excel in static settings, but struggle in dynamic collaborative tasks where explicit signals of interlocutor beliefs are sparse and skewed. |
| Approach: | They propose a Frictional Agent Alignment Framework to generate precise, context-aware friction that prompts deliberation and re-examination of existing evidence. |
| Outcome: | The proposed framework outperforms existing methods in producing concise, interpretable friction and in OOD generalization. |
AxomiyaBERTa: A Phonologically-aware Transformer Model for Assamese (2023.findings-acl)
Copied to clipboard
| Challenge: | AxomiyaBERTa is a novel BERT model for low-resource languages . Transformers require extensive computing resources and suffer in low-compute settings . |
| Approach: | They propose a novel BERT model for Assamese, a morphologically-rich low-resource language of eastern India that is trained on a simple masked language modeling task without the NSP objective. |
| Outcome: | The proposed model performs well on token-level tasks and on “longer context” tasks with the aid of embedding disperser and phonological signals. |