TMCnet News

S2W Provides Generative AI Platform 'SAIP' to Hyundai Steel
[May 20, 2024]

S2W Provides Generative AI Platform 'SAIP' to Hyundai Steel

IRVINE, Calif., May 20, 2024 /PRNewswire/ -- Data intelligence company S2W (CEO: Sangduk Suh) announced the provision of its generative AI platform, 'SAIP (S2W AI Platform),' to Hyundai Steel. Hyundai Steel will utilize SAIP in its internal knowledge information platform, 'HIP (Hyundai-steel Intelligence Platform).'

HIP aims to:

  • Provide a knowledge information system to its employees.
  • Improve efficiency in internal document searches.
  • Support employee tasks and enhance efficiency through a management support chatbot.

S2W has built a big data system tailored for the steel industry and applied an ontology-based SAIP to provide accurate responses. SAIP combines a secure structure with RAG (Retrieval-Augmented Generation), defending against data breache and internal threats, ensuring both accuracy and safety.

HIP is the first instance of utilizing an AI platform in the steelmaking and refining sector with LLM (Large Language Models). S2W anticipates the future application of SAIP across various industries, including finance, steel/chemical, and ICT.

Dong-Yun Han, the project manager for HIP at Hyundai Steel, stated, "Employees have expressed high satisfaction with response speed and accuracy. We expect to see increased work efficiency through the effective use of accumulated knowledge information."

Keun-Tae Park, CTO of S2W, commented, "We are delighted to have successfully launched HIP, which understands the language of the steelmaking sector, through our expertise in processing complex unstructured data."

S2W, a data intelligence company specializing in AI, big data, and security technologies, has been recognized for its technological prowess. It developed the AI language model for the dark web, 'DarkBERT,' and annually publishes papers on AI language models in prestigious conferences such as NAACL and ACL.


The concept of 'ontology' was defined in 1992 by Professor Gruber of Stanford University as "a specification of a conceptualization." Today, in the field of AI, it is discussed as a methodology for building AI that processes, shares, and reuses knowledge and relationships between various concepts, particularly in areas like natural language processing and the Semantic Web.

RAG (Retrieval-Augmented Generation)
RAG is a technology that optimizes the outputs of LLMs by utilizing reliable external data sources, thereby reducing hallucinations in generated responses.


Cision View original content to download multimedia:


[ Back To's Homepage ]