Research findings unveiled on exchange wallet identification and virtual asset classification
and on-chain fund tracing

Dunamu, which operates the virtual asset exchange Upbit, announced on November 17 that three of its research papers had been accepted at ICAIF 2026.


The conference is an international academic conference specializing in artificial intelligence (AI) in finance, organized by the Association for Computing Machinery (ACM), an academic organization in the field of computer science. Companies including BlackRock, JPMorgan and Visa participate. More than 1,000 papers, the most ever, were submitted to this year's conference, and the acceptance rate for the main conference was 22%.


The accepted papers address research aimed at improving the accuracy and reliability of virtual asset data analysis, including AI-based exchange wallet identification, virtual asset classification and blockchain fund tracing. The findings from the exchange wallet identification research have been put to practical use in Upbit Data Lab's blockchain analysis service.


The paper accepted for the main conference is titled "Classifying On-Chain Behavioral Characteristics of Ethereum-Based Centralized Exchange Wallets Using Proof of Reserves." It presents a machine learning (ML)-based methodology for identifying wallets held by centralized exchanges through analysis of blockchain transaction data.

From left, paper authors So Minseop, head of Dunamu's FDI team; Park Seongchan, a researcher on Dunamu's Quant team; Kim Seongyong, a researcher on Dunamu's FDI team; and Jeon Hyeonho, head of Dunamu's Quant team. Three of their papers were accepted for presentation at the ICAIF 2026 academic conference. Dunamu

From left, paper authors So Minseop, head of Dunamu's FDI team; Park Seongchan, a researcher on Dunamu's Quant team; Kim Seongyong, a researcher on Dunamu's FDI team; and Jeon Hyeonho, head of Dunamu's Quant team. Three of their papers were accepted for presentation at the ICAIF 2026 academic conference. Dunamu

View original image

The specialized workshop paper, titled "Semantic Transformation for an Institutional Digital Asset Taxonomy," presents a method for analyzing smart contract code with large language models (LLMs), identifying the practical uses of virtual assets and classifying them accordingly. The technology is also slated for use in further developing Upbit's virtual asset classification system.


Another workshop paper, titled "Estimating the Range of Deposited Funds Used in Incomplete Transaction Records," examines the information needed to analyze blockchain fund flows even when transaction records are incomplete. The research is expected to be used in AI-based fund tracing and financial crime analysis technologies.



Kim Daehyun, chief data officer (CDO) of Dunamu, said, "This research is significant because it confirms the potential to use AI technology to analyze blockchain data more precisely and increase the reliability of the results. We will continue research combining AI and blockchain technologies to lead technological innovation in the digital asset industry."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing