Papers by Sandipan Sikdar
When Facts Change: Temporal Knowledge Conflict Resolution in LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models are increasingly used in retrieval-augmented generation systems to reconcile knowledge conflicts between parametric memory and contextual inputs. |
| Approach: | They propose to use mutability to resolve temporal misalignment in large language models to compare stable and recently updated facts from Wikidata to determine if mutable models can serve as a mediating signal in this process. |
| Outcome: | The proposed model can produce reasoning for facts that actually changed but rarely for stable ones, whereas smaller models rarely detect conflict, while larger models detect it but fail to act on mutability judgments. |
Rethinking Evaluation Methods for Machine Unlearning (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Machine *unlearning* is a method for deleting information about training instances from a trained machine learning model. |
| Approach: | et al. propose to use random selection to obtain an unlearned model that behaves as closely as possible to the retrained model. |
| Outcome: | a new method is proposed to remove training instances from a trained model . the proposed method is infeasible due to the size of the dataset and the size . |
Integrated Directional Gradients: Feature Interaction Attribution for Neural NLP Models (2021.acl-long)
Copied to clipboard
| Challenge: | Existing methods for attribution of importance to features borrowed from cooperative game theory . success of Deep Neural Networks has led to their ability to learn from complex higher order interactions from raw features. |
| Approach: | They propose a method for attributing importance scores to groups of features . they propose axioms that any intuitive feature group attribution method should satisfy . |
| Outcome: | The proposed method captures the importance of features in a linguistic model using negations and conjunctions. |
SensePOLAR: Word sense aware interpretability for pre-trained contextual word embeddings (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing word embedding models lack interpretability for words . |
| Approach: | They propose to add interpretability to word embeddings by using a POLAR framework that enables wordsense aware interpretations for pre-trained contextual word embeds. |
| Outcome: | The proposed framework achieves comparable performance to existing embeddings across GLUE and SQuAD benchmarks. |
ACCESS DENIED INC: The First Benchmark Environment for Sensitivity Awareness (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models can process text from various document formats and facilitate user interactions through natural language queries. |
| Approach: | They propose a concept of sensitivity awareness (SA) which enables LLMs to adhere to predefined access rights rules. |
| Outcome: | The proposed model can adhere to predefined access rights rules while managing unauthorized requests while effectively addressing legitimate queries. |
StRE: Self Attentive Edit Quality Prediction in Wikipedia (P19-1)
Copied to clipboard
| Challenge: | Existing methods for content moderation and review rely on page reputation, editor activity or rule based heuristics. |
| Approach: | They propose a self-attentive revision encoder which leverages orthographic similarity of lexical units toward predicting the quality of new edits. |
| Outcome: | The proposed model outperforms existing models by at least 17% and at most 103% on a set of 21M revisions across 32K Wikipedia pages. |