KISTI Showcases Core "AI for Science" Technologies at Global Data Mining Conference KDD 2026

- Accelerates global research innovation with science-specialized AI models and autonomous research platforms

□ The Korea Institute of Science and Technology Information (President Sik Lee, hereafter KISTI) announced that it participated in KDD 2026, the world’s premier academic conference in data mining and artificial intelligence, to showcase its next-generation technology achievements under the theme of "AI for Science."

□ KISTI introduced a range of cutting-edge solutions, including the development of hyperscale AI models specialized for science and technology, an autonomous research platform that automates and supports the end-to-end scientific discovery process, innovative agent technologies, and an ecosystem platform for Large Language Model (LLM) development. 

□ Key technologies featured include △ "AI Scientist," an autonomous agent that analyzes research literature and generates hypotheses, △ "Web Agent," a tool for document-grounded form filling that automatically populates web forms with personal CVs and research proposals, and △"KONI-Forge," a Large Language Model Operations (LLMOps) platform designed to support the efficient development and deployment of LLMs.

□ The "AI Scientist" navigates and analyzes vast volumes of scientific literature to synthesize knowledge, derive novel research hypotheses, and suggest promising directions for scientific breakthroughs. By supporting researchers through the labor-intensive process of manual literature review and hypothesis formulation, this technology significantly enhances the efficiency of scientific discovery.

□ The "Web Agent" streamlines administrative burdens by parsing multi-modal documents, such as CVs and research proposals, to automatically fill out complex web forms across research portals and grant management systems. By automating repetitive administrative tasks, it enables researchers to dedicate more time to core scientific endeavors. 

□ The LLMOps platform, "KONI-Forge" leverages High-Performance Computing (HPC) infrastructure to provide comprehensive support for the entire LLM lifecycle ranging from fine-tuning to model evaluation, dynamic serving, and optimization. It offers an environment where researchers and enterprises can efficiently build and utilize specialized scientific LLMs.

□ In addition, KISTI presented its progress in building a proprietary open science ecosystem for shared use by academia and industry, featuring a science-specialized Multimodal Large Language Model (MLLM) architecture for complex document understanding, advanced fine-tuning techniques, and a large-scale scientific dataset preprocessing pipeline, drawing strong interest from on-site researchers and global industry leaders. 

□ Dr. Kyong-Ha Lee, Director of the Large-Scale AI Research Center at KISTI, stated, "This showcase at KDD 2026 served as a pivotal opportunity to establish KISTI’s prowess in 'AI for Science' technologies, autonomous agents, and LLMOps platform achievements among global data and AI researchers. Moving forward, we will continue to advance the AI Scientist, intelligent agents, and the KONI-Forge platform to lead the national science and technology AI ecosystem and drive global research innovation."

KISTI Booth at KDD 2026

KONI AI Services