AIConfPaper
← Search

Tharindu Cyril Weerasooriya

7 accepted papers

2026

ProRefine: Inference-Time Prompt Refinement with Textual Feedback (Student Abstract)

Deepak Pandita, Tharindu Cyril Weerasooriya, Ankit Shah, Isabelle Diana May-Xin Ng, Christopher M. Homan, Wei Wei
AAAI 2026technical

Agentic workflows, where multiple AI agents collaborate to accomplish complex tasks like reasoning or planning, play a substantial role in many cutting-edge commercial applications. These workflows depend critically on the prompts used to provide the roles models play in such workflows. Poorly desig

Cited by 0Source ↗PDF ↗Scholar ↗
2025

ARTICLE: Annotator Reliability Through In-Context Learning

Sujan Dutta, Deepak Pandita, Tharindu Cyril Weerasooriya, Marcos Zampieri, Christopher M. Homan, Ashiqur R. KhudaBukhsh
AAAI 2025technical

Ensuring annotator quality in training and evaluation data is a key piece of machine learning in NLP. Tasks such as sentiment analysis and offensive speech detection are intrinsically subjective, creating a challenging scenario for traditional quality assessment approaches because it is hard to dist…

Cited by 0Source ↗PDF ↗code ↗Scholar ↗
2025

ARTICLE: Annotator Reliability Through In-Context Learning (Student Abstract)

Sujan Dutta, Deepak Pandita, Tharindu Cyril Weerasooriya, Marcos Zampieri, Christopher M. Homan, Ashiqur R. KhudaBukhsh
AAAI 2025technical

Ensuring annotator quality in training and evaluation data is a key piece of machine learning in NLP. Tasks such as sentiment analysis and offensive speech detection are intrinsically subjective, creating a challenging scenario for traditional quality assessment approaches because it is hard to dist…

Cited by 0Source ↗PDF ↗code ↗Scholar ↗
2024

Rater Cohesion and Quality from a Vicarious Perspective

Deepak Pandita, Tharindu Cyril Weerasooriya, Sujan Dutta, Sarah K. Luger, Tharindu Ranasinghe, Ashiqur R. KhudaBukhsh +2
EMNLP 2024finding

Human feedback is essential for building human-centered AI systems across domains where disagreement is prevalent, such as AI safety, content moderation, or sentiment analysis. Many disagreements, particularly in politically charged settings, arise because raters have opposing values or beliefs. Vic…

Cited by 0Source ↗PDF ↗code ↗Scholar ↗
2023

Disagreement Matters: Preserving Label Diversity by Jointly Modeling Item and Annotator Label Distributions with DisCo

Tharindu Cyril Weerasooriya, Alexander Ororbia, Raj Bhensadadia, Ashiqur KhudaBukhsh, Christopher Homan
ACL 2023findings

Annotator disagreement is common whenever human judgment is needed for supervised learning. It is conventional to assume that one label per item represents ground truth. However, this obscures minority opinions, if present. We regard “ground truth” as the distribution of all labels that a population…

Cited by 8Source ↗PDF ↗Scholar ↗
2023

Subjective Crowd Disagreements for Subjective Data: Uncovering Meaningful CrowdOpinion with Population-level Learning

Tharindu Cyril Weerasooriya, Sarah Luger, Saloni Poddar, Ashiqur KhudaBukhsh, Christopher Homan
ACL 2023long

Human-annotated data plays a critical role in the fairness of AI systems, including those that deal with life-altering decisions or moderating human-created web/social media content. Conventionally, annotator disagreements are resolved before any learning takes place. However, researchers are increa…

Cited by 6Source ↗PDF ↗code ↗Scholar ↗
2023

Vicarious Offense and Noise Audit of Offensive Speech Classifiers: Unifying Human and Machine Disagreement on What is Offensive

Tharindu Cyril Weerasooriya, Sujan Dutta, Tharindu Ranasinghe, Marcos Zampieri, Christopher M Homan, Ashiqur R. KhudaBukhsh
EMNLP 2023long main

Offensive speech detection is a key component of content moderation. However, what is offensive can be highly subjective. This paper investigates how machine and human moderators disagree on what is offensive when it comes to real-world social web political discourse. We show that (1) there is exten…

Cited by 0Source ↗code ↗Scholar ↗