Papers by Sarthak Sarthak

38 papers
An Empirical Comparison of Instance Attribution Methods for NLP (2021.naacl-main)

Copied to clipboard

Challenge: Influence functions provide machinery for identifying training instances that may have led to a specific prediction, but are computationally expensive and prohibitive in many cases.
Approach: They evaluate the degree to which different potential instance attribution agrees with respect to the importance of training samples.
Outcome: The proposed methods exhibit desirable characteristics similar to more complex methods, but are computationally expensive.
Learning to Faithfully Rationalize by Construction (2020.acl-main)

Copied to clipboard

Challenge: Neural models dominate NLP but it remains difficult to know why they make specific predictions for sequential text inputs.
Approach: They propose a model to produce faithful rationales for neural text classification by defining independent snippet extraction and prediction modules.
Outcome: The proposed model produces faithful explanations even when the model is complex and complex.
Identification of Fine-Grained Location Mentions in Crisis Tweets (2022.lrec-1)

Copied to clipboard

Challenge: Recent studies have focused on identifying informative tweets by individuals affected by a crisis, without considering their specific types.
Approach: They assemble two tweet crisis datasets and manually annotate them with specific location types to facilitate progress on the fine-grained location identification task.
Outcome: The proposed model performs well in both in-domain and cross-domain settings.
An Evaluation of Progressive Neural Networksfor Transfer Learning in Natural Language Processing (2020.lrec-1)

Copied to clipboard

Challenge: Fine-tuning suffers from catastrophic forgetting, a problem exacerbated in natural language processing (NLP).
Approach: They propose to use progressive neural networks to re-use previously learned knowledge when learning new tasks.
Outcome: The proposed approach improves on common NLP tasks across a range of architectures, datasets, and tasks.
Does BERT Pretrained on Clinical Notes Reveal Sensitive Data? (2021.naacl-main)

Copied to clipboard

Challenge: Pretraining large (masked) language models over EHR data has yielded consistent performance gains across tasks.
Approach: They propose to use large Transformers to release pretraining models over EHRs . they propose to recover patient names and conditions associated with them .
Outcome: The proposed models recover patient names and conditions associated with patients . the proposed models share the model parameters for use by other researchers .
Multi-task Learning to Enable Location Mention Identification in the Early Hours of a Crisis Event (2021.findings-emnlp)

Copied to clipboard

Challenge: Social media is a platform for people to share their concerns and report information as eyewitnesses of events.
Approach: They propose a multi-task learning approach to leverage available annotated data for several related tasks from the crisis domain to improve performance on a main task with limited annotation.
Outcome: The proposed approach improves performance on a task with limited annotated data.
Scalable and Robust Self-Learning for Skill Routing in Large-Scale Conversational AI Systems (2022.naacl-industry)

Copied to clipboard

Challenge: Existing methods to enable skill routing do not scale in terms of the number of skills and skill on-boarding.
Approach: They propose a model-based approach to enable natural conversation by allowing frequent policy updates . they propose an annotation-based system, rule-based model, and bandit-based learning .
Outcome: The proposed method is scalable and cost-effective, the authors show . they show that it can improve the user experience without abrupt policy changes .
Stance Detection in COVID-19 Tweets (2021.acl-long)

Copied to clipboard

Challenge: a global pandemic of COVID-19 has forced major changes in our daily lives . a new stance detection dataset is being used to track the stances of Twitter users .
Approach: They use Twitter stance data to collect stances on topics related to the pandemic . they train models to take advantage of large amounts of unlabeled data .
Outcome: The proposed model improves on existing stance detection datasets and unlabeled data.
Empirical Evaluation of Active Learning Techniques for Neural MT (D19-61)

Copied to clipboard

Challenge: Several active learning (AL) algorithms for machine translation (MT) have been well-studied for phrase-based MT.
Approach: They propose to use a phrase-based algorithm to compare different AL methods in a simulated AL framework to demonstrate how unsupervised pre-training and paraphrastic embedding can be used to improve existing AL methods.
Outcome: The proposed method outperforms existing methods in the context of phrase-based MT and is based on a simulated phrase-driven dataset.
CASE: Efficient Curricular Data Pre-training for Building Assistive Psychology Expert Models (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing methods to identify mental health disorders rely on limited availability of psychologists.
Approach: They propose to use forum posts to analyze text data to identify mental health issues . they propose to utilize readily available curricular texts for pre-training pipelines .
Outcome: The proposed pipelines achieve an f1 score of 0.91 for Depression and 0.88 for Anxiety compared to existing pipelines.
Learning to Relate from Captions and Bounding Boxes (P19-1)

Copied to clipboard

Challenge: Existing methods for classifying images without supervision are limited.
Approach: They propose a top-down attention mechanism to align entities in captions to objects in the image and leverage the syntactic structure of captions for alignment.
Outcome: The proposed model achieves a recall@50 of 15% and recall@100 of 25% on the relationships present in the image and predicts relations that are not present in captions.
Influence Functions for Sequence Tagging Models (2022.findings-emnlp)

Copied to clipboard

Challenge: Named Entity Recognition, Part-of-Speech tagging, and Semantic Role Labeling are standard tasks in NLP, but there has been little work on interpretability methods for sequence taging.
Approach: They propose to extend influence functions to sequence tagging tasks by identifying noisy annotations in NER corpora.
Outcome: The proposed methods are able to identify noisy annotations in NER corpora and are scalable.
Learning Disentangled Representations of Texts with Application to Biomedical Abstracts (D18-1)

Copied to clipboard

Challenge: a method for learning disentangled representations of texts that encode distinct and complementary aspects is proposed . a classic problem in distributed representation learning is that it is difficult to determine what information individual dimensions encode.
Approach: They propose a method for learning disentangled representations of texts that encode distinct and complementary aspects by a adversarial objective based on the (dis)similarity between triplets of documents with respect to specific aspects.
Outcome: The proposed method can be used to perform aspect-specific retrieval on biomedical abstracts.
SVeritas: Benchmark for Robust Speaker Verification under Diverse Conditions (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing benchmarks only evaluate a subset of potential conditions, missing others entirely.
Approach: a new benchmark suite evaluates speaker verification models under a variety of stressors . a san francisco-based team evaluates models under natural and background conditions .
Outcome: a new benchmark suite evaluates speaker verification models under stressors under a variety of conditions . the results show that some models perform better under stress conditions than others .
Jointly Learning to Align and Translate with Transformer Models (D19-1)

Copied to clipboard

Challenge: Existing word alignment models are not accurate for word alignments.
Approach: They propose a method to train a Transformer model to produce accurate translations and alignments.
Outcome: The proposed model outperforms GIZA++ trained models on translation and alignment tasks while maintaining translation accuracy.
SciREX: A Challenge Dataset for Document-Level Information Extraction (2020.acl-main)

Copied to clipboard

Challenge: Conventional datasets and methods for information extraction focus on within-sentence relations from general Newswire text.
Approach: They propose a document-level IE dataset that integrates automatic and human annotations to annotate entities and document- level N-ary relation identification from scientific articles.
Outcome: The proposed dataset extends state-of-the-art IE models to document-level IE.
Combining Feature and Instance Attribution to Detect Artifacts (2022.findings-acl)

Copied to clipboard

Challenge: In this paper, we evaluate use of different attribution methods for aiding identification of training data artifacts.
Approach: They propose hybrid methods that combine saliency maps and instance attribution methods to aid in identifying training data artifacts.
Outcome: The proposed methods can be used to efficiently uncover artifacts in training data when a challenging validation set is available.
Permutation Invariant Strategy Using Transformer Encoders for Table Understanding (2022.findings-naacl)

Copied to clipboard

Challenge: Existing methods for encoding text in tables require additional training and require additional pretraining.
Approach: They propose a novel encoding strategy that preserves the critical property of permutation invariance across rows or columns.
Outcome: The proposed approach outperforms state-of-the-art methods on three table interpretation tasks: column type annotation, relation extraction, and entity linking.
Improving Tool Retrieval by Leveraging Large Language Models for Query Generation (2025.coling-industry)

Copied to clipboard

Challenge: Large Language Models (LLMs) have shown great promise in common sense language understanding, conversational fluency, and reasoning.
Approach: They propose to use Large Language Models to generate a retrieval query and embed it into the prompt to find relevant tools via a nearest-neighbor search.
Outcome: The proposed method improves retrieval for in-domain (seen tools) and out-of-domain settings.
DQA: Diagnostic Question Answering for IT Support (2026.acl-industry)

Copied to clipboard

Challenge: a diagnostic question-answering framework maintains persistent diagnostic state and aggregates retrieved cases at the level of root causes rather than individual documents.
Approach: They propose a diagnostic question-answering framework that maintains persistent diagnostic state . it aggregates retrieved cases at the level of root causes rather than individual documents .
Outcome: The framework achieves a 78.7% success rate under trajectory-level success criterion compared to a multi-turn RAG baseline . the framework reduces average turns from 8.4 to 3.9, compared with a single-turn baseline crim.
HatePRISM: Policies, Platforms, and Research Integration. Advancing NLP for Hate Speech Proactive Mitigation (2025.findings-acl)

Copied to clipboard

Challenge: Existing approaches to manage hate speech rely on reactive measures such as blocking or suspending offensive messages . despite regulations imposed by nations and social media platforms, hateful content remains a challenge .
Approach: They propose a framework for automated hate speech moderation based on different strategies . they examine hate speech regulations and strategies from three perspectives .
Outcome: The proposed framework could be based on a combination of country regulations, social platform policies, and NLP research datasets.
Attention is not Explanation (N19-1)

Copied to clipboard

Challenge: Attention mechanisms have seen wide adoption in neural NLP models.
Approach: They perform extensive experiments to assess the degree to which attention weights provide meaningful "explanations" they find that attention weighted inputs are often uncorrelated with gradient-based measures of feature importance .
Outcome: The proposed model is based on a distribution over attended-to input units . the findings show that attention weights are often uncorrelated with features .
ERASER: A Benchmark to Evaluate Rationalized NLP Models (2020.acl-main)

Copied to clipboard

Challenge: State-of-the-art models in NLP are opaque in terms of how they come to make predictions.
Approach: They propose to release a benchmark to measure the quality of rationales extracted by models and how faithful these rationale are to human annotators.
Outcome: The proposed benchmark will enable researchers to compare models and track progress on interpretable models for NLP.
Taxonomy Construction of Unseen Domains via Graph-based Cross-Domain Knowledge Transfer (2020.acl-main)

Copied to clipboard

Challenge: Existing taxonomies are either entirely absent or missing.
Approach: They propose a GNN-based cross-domain transfer framework for the taxonomy construction task.
Outcome: The proposed framework improves on benchmark datasets from science and environment domains.
Aligning to Constraints for Data-Efficient Language Model Customization (2025.findings-naacl)

Copied to clipboard

Challenge: General-purpose language models (LMs) are aligned to diverse user intents, but fall short when it comes to specific applications.
Approach: They propose a framework that uses constraints to automatically produce supervision signals for user alignment with constraints.
Outcome: The proposed framework can produce supervision signals for user alignment with constraints.
How Many and Which Training Points Would Need to be Removed to Flip this Prediction? (2023.eacl-main)

Copied to clipboard

Challenge: Existing methods to find St using brute-force are intractable.
Approach: They propose a fast approximation method to find St based on influence functions . they propose to identify a minimum subset of training data that one would need to remove .
Outcome: The proposed method can find St based on influence functions for simple classification models.
Automatic Taxonomy Induction and Expansion (D19-3)

Copied to clipboard

Challenge: Knowledge Graph Induction Service (KGIS) enables automatic taxonomy induction and human-in-the-loop curation.
Approach: They describe the features of the Knowledge Graph Induction Service (KGIS) KGIS allows the user to semi-automatically curate and expand the induced taxonomies through a component called Smart SpreadSheet .
Outcome: The Knowledge Graph Induction Service (KGIS) is an end-to-end knowledge graph induction system.
Scalable and Safe Remediation of Defective Actions in Self-Learning Conversational Systems (2023.acl-industry)

Copied to clipboard

Challenge: Off-Policy reinforcement learning has been used to improve conversational AIs, but in large-scale commercial environments it is challenging to balance between policy improvements and experience continuity.
Approach: They propose to curate and leverage regression incident reports to validate, safe-guard, and improve policies prior to the online deployment.
Outcome: The proposed method validates, safe-guards, and improves policies prior to the online deployment.
Mitigating Hallucinated Translations in Large Language Models with Hallucination-focused Preference Optimization (2025.naacl-long)

Copied to clipboard

Challenge: Machine Translation (MT) systems based on fine-tuned large language models (LLMs) are at a higher risk of generating hallucinations, which can severely undermine user’s trust and safety.
Approach: They propose a method that intrinsically learns to mitigate hallucinations during the model training phase.
Outcome: The proposed method reduces hallucinations by 89% on an average across three unseen target languages while preserving translation quality.
Leveraging Synthetic Targets for Machine Translation (2023.findings-acl)

Copied to clipboard

Challenge: Using synthetic target data, training models on synthetic targets outperforms training on actual ground-truth data.
Approach: They propose a recipe for training machine translation models on synthetic target data by leveraging a large pre-trained model.
Outcome: The proposed model outperforms training on real-world translation datasets.
Open Knowledge Graphs Canonicalization using Variational Autoencoders (2021.emnlp-main)

Copied to clipboard

Challenge: Existing approaches to solve this problem generate embeddings for noun and relation phrases . ambiguous subject-relation-object triples are created by open knowledge graphs .
Approach: They propose a model to learn both embeddings and cluster assignments in an end-to-end approach . they propose CUVA to be able to group noun and relation phrases using embeddable features .
Outcome: The proposed model outperforms state-of-the-art methods over multiple benchmarks.
ClaimPortal: Integrated Monitoring, Searching, Checking, and Analytics of Factual Claims on Twitter (P19-3)

Copied to clipboard

Challenge: ClaimPortal is a web-based platform for monitoring, searching, checking and analyzing factual claims on Twitter from the American political domain.
Approach: They present a web-based platform for monitoring, searching, checking and analyzing English factual claims on Twitter from the American political domain.
Outcome: The proposed platform can monitor, search, check, and analyze English factual claims on Twitter from the political domain.
Probing LLMs for hate speech detection: strengths and vulnerabilities (2023.findings-emnlp)

Copied to clipboard

Challenge: Recent efforts to detect hateful or toxic language using large language models have not used explanation, additional context and victim community information in the detection process.
Approach: They use different prompt variations, input information and victim community information to evaluate large language models in zero shot setting without adding any in-context examples.
Outcome: The proposed models perform significantly better when included in the pipeline than baseline models.
GunStance: Stance Detection for Gun Control and Gun Regulation (2024.acl-long)

Copied to clipboard

Challenge: Social media, especially Twitter, has been a melting pot for such debates.
Approach: They propose to annotate tweets relevant to shooting events into three classes: In-Favor, Against, and Neutral.
Outcome: The proposed approach outperforms supervised, semi-supervised, and LLM-based zero-shot models on the dataset.
Compute-Efficient Churn Reduction for Conversational Agents (2023.emnlp-industry)

Copied to clipboard

Challenge: churn occurs when retraining models yields different predictions despite using the same data and hyper-parameters.
Approach: They propose a method that pairs semantic parses based on their “function call signature” and encourages similarity through an additional loss based upon Jensen-Shannon Divergence.
Outcome: The proposed method improves in academic, noisy, and industry settings.
Bilingual Lexicon Induction with Semi-supervision in Non-Isometric Embedding Spaces (P19-1)

Copied to clipboard

Challenge: Recent work on bilingual lexicon induction (BLI) relies on an assumption about the isometry of two embedding spaces.
Approach: They propose a semi-supervised approach that relaxes the isometric assumption while leveraging limited aligned bilingual lexicons and a larger set of unaligned word embeddings.
Outcome: The proposed method obtains state-of-the-art results on 15 of 18 language pairs on the MUSE dataset and does particularly well when the embedding spaces don’t appear isometric.
SUPP.AI: finding evidence for supplement-drug interactions (2020.acl-demos)

Copied to clipboard

Challenge: Dietary supplements are used by a large portion of the population, but information on their pharmacologic interactions is incomplete.
Approach: They propose an application to search evidence sentences extracted from the literature to identify supplement-drug interactions.
Outcome: The proposed model extracts supplement information and identifies interactions using labeled DDI data.
Modular Self-Supervision for Document-Level Relation Extraction (2021.emnlp-main)

Copied to clipboard

Challenge: Prior work on information extraction tends to focus on binary relations within sentences . practical applications often require extracting complex relations across large text spans .
Approach: They propose to decompose document-level relation extraction into relation detection and argument resolution, taking inspiration from Davidsonian semantics.
Outcome: The proposed method outperforms state-of-the-art methods in biomedical machine reading for precision oncology by 20 absolute F1 points.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations