Program

Date: 3 March 2025, 9AM to 6PM (Eastern Standard Time GMT-5)

*Each paper will take 10 mins oral presentations, and 5 mins Q&A.

* Location: (onsite) Room 111B and (online) https://unimelb.zoom.us/j/81354832494?pwd=ou7d5szbkAR2xzA9TaRPa5CAwBdX1k.1

Morning Session

08:55 AM Opening Remarks 🥳

09:00 AM LLM-based Corroborating and Refuting Evidence Retrieval for Scientific Claim Verification
Authors: Siyuan Wang (AHNU), James Foulds (UMBC), Md Osman Gani (UMBC), Shimei Pan (UMBC)*

SBS Figures: Pre-training Figure QA from Stage-by-Stage Synthesized Images
Authors: Risa Shinoda (Kyoto University, National Institute of Advanced Industrial Science and Technology (AIST))*, Kuniaki Saito (OMRON SINIC X), Shohei Tanaka (OMRON SINIC X), Tosho Hirasawa (OMRON SINIC X), Yoshitaka Ushiku (OMRON SINIC X)

Domain-specific Question Answering with Hybrid Search
Authors: Dewang Sultania (Adobe)*, Zhaoyu Lu (Adobe), Twisha Naik (Adobe), Franck Dernoncourt (Adobe), David Seunghyun Yoon (Adobe)

Zero-Shot Complex Question-Answering on Long Scientific Documents
Authors: Wanting Wang (London School of Economics)*

StructFormer: Document Structure-based Masked Attention and its impact on Language Model Pre-Training
Authors: Kaustubh Ponkshe (Mohamed bin Zayed University of Artificial Intelligence)*, Venkatapathy Subramanian ( Indian Institute of Technology Bombay), Natwar Modani (Adobe India), Ganesh Ramakrishnan (Indian Institute of Technology Bombay)

10:30 AM Morning Tea Break (30 mins) ☕🍵

11:00 AM Epistemic Bias as a Means for the Automated Detection of Injustices in Text
Authors: Kenya Andrews (University of Illinois at Chicago)*, Lamogha Chiazor (IBM Research)

Leveraging Structural Information in Tree Ensembles for Table Representation Learning
Authors: Nikhil Priyatam Pattisapu (Amazon), Siva Rajesh Kasa (National University of Singapore)*, Sumegh Roychowdhury (Amazon), Karan Gupta (Amazon), Anish Bhanushali (NA), Prassana Srinivasa Murthy (NA)

Judge a Book by its Cover: Investigating Multi-Modal LLMs for Multi-Page Handwritten Document Transcription
Authors: Benjamin Gutteridge (Oxford)*

Enhancing Vietnamese VQA through Curriculum Learning on Raw and Augmented Text Representations
Authors: Khoi Nguyen (Ton Duc Thang University)*, Linh Vu (AI VIETNAM), Thang Duong (AI VIETNAM), Thuan Duong (AI VIETNAM), Thanh Huy Nguyen (Université de Bourgogne), Vinh Dinh (AI VIETNAM)

12:00 PM Lunch Break (1.5 hours) 🍔🍕🍞🍝🍜🍛

Afternoon Session

13:30 PM KeyNote Speaker (Prof. Eduard Hovy, University of Melbourne and Carnegie Mellon University) - 1 hour

Prof. Eduard Hovy

Title: Document Structure in the Age of LLMs

Abstract: Generative LLMs produce their output incrementally, left-to-right. It is no surprise that early GenLLMs like GPT-2 were famous for their tendency to wander off topic and hallucinate wildly. But more-recent GenLLMs can do amazing things, including staying coherent around a topic and producing well-structured output, even when the output exceeds a page. How do they do so? Few if any humans can produce a well-structured coherent document of longer than a page without any long-range planning. This talk discusses what’s needed for coherent structure, especially for scientific documents, and explores some possible ways GenLLMs do what they do.

14:40 PM Break (20 mins)

15:00 PM Fine-grained and Explainable Factuality Evaluation for Multimodal Summarization
Authors: Yue Zhang (UTD)*, Jingxuan Zuo (SDU), Liqiang Jing (UTD)

Enhancing Document Key Information Localization Through Data Augmentation
Authors: Yue Dai (The University of Western Australia)

ChatGeo: Multimodal Retrieval-based Meta Prompting in Language Models for GeoScience
Authors: Yihao Ding (The University of Melbourne), Caren Han (The University of Melbourne), Zhe Huang (ByteDance), Kaixuan Ren (The University of Sydney), Yuzhong Ma (The University of Sydney), Wei Liu (The University of Western Australia), Eun-Jung Holden (The University of Melbourne)

16:00 PM LeaderBoard Award Ceremony

LeaderBoard - Winner#1
Authors: Team DAKiet

LeaderBoard - Winner#2
Authors: Team ctyun-ai

LeaderBoard - Winner#3
Authors: Team ABCD

17:00 PM Closing 📪

Page updated

Google Sites

Report abuse