Dunamu said on Jan. 26 that its machine learning team’s paper was accepted to the demo track at the international AI conference Association for the Advancement of Artificial Intelligence (AAAI) and was successfully demonstrated.

AAAI is a major conference where AI researchers worldwide present the latest technologies and research results, and is considered one of the world’s three leading AI conferences.

Dunamu’s machine learning team paper was selected for the highly competitive demo track. The demo track goes beyond theoretical presentations, with a working system demonstrated to assess practical effectiveness, requiring research results to be proven through applied cases.

Hee-soo Park (박희수), a researcher on Dunamu’s machine learning team, presented a paper titled "Market-Aware Event Timeline Summarization: Integrating Price Signals to Improve Financial News Understanding" and demonstrated an in-house system.

The research describes a system that combines news data with fluctuation data from digital asset price charts to select and provide only the key news that caused price changes.

Previously, there were limitations in identifying the key news that affected actual price changes and in immediately understanding the reasons for price rises and falls. To address this, Dunamu’s machine learning team proposed new modelling that combines large language models (LLM) with volatility indicators.

The system automatically extracts digital asset-related events from news feeds, selects only the events that caused volatility at times of high volatility, and has an LLM summarise the events and background knowledge needed to understand them. It then visualises them in a timeline format alongside charts to help investors intuitively understand the background behind chart movements.

Dae-hyun Kim (김대현), Dunamu’s chief data officer, said the AAAI presentation was meaningful in that Dunamu’s AI technology gained global recognition and demonstrated practicality that can resolve information asymmetry. He said the company will continue to use AI technology to provide more valuable information to investors and contribute to improving market transparency.