Research Profile of Sudeshna Das

Research Interests
NLP for Social Good | AI for Healthcare

Selected Publications

Emerging Trends of Self-Harm Using Sodium Nitrite in an Online Suicide Community: Observational Study Using Natural Language Processing Analysis
Sudeshna Das, Drew Walker, Swati Rajwal, Sahithi Lakamana, Steven A Sumner, Karin A Mack, Wojciech Kaczkowski, Abeed Sarker
JMIR Mental Health, Volume 11, Issue 1, 2024
TL;DR | Paper | BibTeX | Code

There is growing concern around the use of sodium nitrite as an emerging means of suicide, particularly among younger people. We studied posts made to an online suicide discussion forum, “Sanctioned Suicide”. Posts pertaining to sodium nitrite show a rise in popularity, and there were statistically significant correlations between real-life use of sodium nitrite and suicidal intent when compared to data from CDC WONDER (⍴=0.727; P<.001) and the National Poison Data System (⍴=0.866; P=.001). Our proposed transformer–based source and location classifier can detect potential sources of sodium nitrite with an accuracy of 72.92%, showing consumption in the United States and elsewhere.

@article{das2024emerging,
title={Emerging trends of self-harm using sodium nitrite in an online suicide community: observational study using natural language processing analysis},
author={Das, Sudeshna and Walker, Drew and Rajwal, Swati and Lakamana, Sahithi and Sumner, Steven A and Mack, Karin A and Kaczkowski, Wojciech and Sarker, Abeed and others},
journal={JMIR Mental Health},
volume={11},
number={1},
pages={e53730},
year={2024},
publisher={JMIR Publications Inc., Toronto, Canada}
}
Gender Tagging of Named Entities using Retrieval-assisted Multi-context Aggregation: An Unsupervised Approach
Sudeshna Das and Jiaul H Paik
Journal of the Association for Information Science and Technology, Volume 74, Issue 4, 2023
TL;DR | Paper | BibTeX | Dataset

Existing approaches toward name gender identification rely exclusively on using the gender distributions from labeled data. In the absence of such labeled data, these methods fail. We propose a two-stage model that is able to infer the gender of names present in text without requiring explicit name-gender labels. We use coreference resolution as the backbone for our proposed model. To aid coreference resolution where the existing contextual information does not suffice, we use a retrieval-assisted context aggregation framework. We demonstrate that state-of-the-art name gender inference is possible without supervision.

@article{das2023gendertag,
title={Gender tagging of Named Entities using Retrieval-assisted Multi-context Aggregation: An Unsupervised Approach},
author={Das, Sudeshna and Paik, Jiaul H},
journal={Journal of the Association for Information Science and Technology},
volume={74},
number={4},
pages={461-475},
year={2023},
publisher={Wiley}
}
Context-Sensitive Gender Inference of Named Entities in Text
Sudeshna Das and Jiaul H Paik
Information Processing & Management, Volume 58, Issue 1, 2021
TL;DR | Paper | BibTeX | Dataset

Gender tagging of named entities has traditionally been database-reliant and insensitive to context. The same gender is assigned to "Alex" in both "Alex is a good boy" and "Alex is a smart girl". We propose a novel context-sensitive supervised approach based on the transformer network to identify the gender of named entities. We also create four open-source datasets from well-known NER corpora and make them publicly available.

@article{das2021context,
title={Context-Sensitive Gender Inference of Named Entities in Text},
author={Das, Sudeshna and Paik, Jiaul H},
journal={Information Processing \& Management},
volume={58},
number={1},
pages={102423},
year={2021},
publisher={Elsevier}
}