Yiming Lu

Yiming Lu

CS PhD student @ Emory

Publications

You can also find my articles on my Google Scholar profile.

Conference Papers

DeFine: Enhancing LLM Decision-Making with Factor Profiles and Analogical Reasoning

Published in ACL Findings, 2025

This paper presents DeFine, a novel approach that bolsters LLM-based decision-making through factor profiles and analogical reasoning. By systematically identifying and comparing key factors across contexts, DeFine promotes deeper context comprehension and more robust inferences. Our experiments on finance-related reasoning tasks demonstrate its ability to deliver both higher accuracy and enhanced interpretability, with broader applications to various decision-making scenarios.

Download Paper

STRUX: An LLM for Decision-Making with Structured Explanations

Published in NAACL, 2025

This paper presents STRUX, which is a novel LLM framework that enhances decision-making through structured explanations. By distilling complex information into key facts and employing self-reflection steps, it categorizes and prioritizes factors affecting decisions. We demonstrate its effectiveness in stock investment decisions based on earnings call transcripts, making LLM decision-making more transparent and practical.

Download Paper

Preprints Papers

Communication to Completion: Modeling Collaborative Workflows with Intelligent Multi-Agent Communication

Published in Under Review, 2025

In this work, we introduce Communication to Completion (C2C), a scalable multi-agent framework that enhances task oriented collaboration through structured communication. C2C features the Alignment Factor (AF), a novel metric quantifying task understanding, and a Sequential Action Framework that enables cost aware communication decisions. Evaluated on realistic coding workflows across varying team sizes and complexity tiers, C2C reduces task completion time by 40%, establishing both theoretical foundations and practical utilities for communication efficient multi-agent systems.

Download Paper

Journal Articles

iSMOD: an integrative browser for image-based single-cell multi-omics data

Published in Nucleic Acids Research, 2023

This paper presents iSMOD, which is the first integrative browser for collecting and analyzing FISH and nucleus proteomics data. By integrating multi-omics studies from 20,000+ published papers, it provides powerful tools for exploring cellular mechanisms and advancing biological research through comprehensive data visualization and analysis.

Download Paper