This page was machine-translated and may differ from the original. View original
Saltlux Dreams of a Future Powered by Artificial Intelligence with "Adam"
Equipped with core technologies based on big data and artificial intelligence
We have various success stories with domestic and foreign companies.Before Google DeepMind's AlphaGo wowed us with its Go game, there was IBM Watson. In 2011, Watson competed on the quiz show Jeopardy! and took first place, beating the all-time champion and championship record holder, demonstrating the wonders of artificial intelligence.
Accordingly, in Korea, we began developing artificial intelligence technology called Exo Brain SW in 2013. Exo Brain is an artificial brain outside of my body, and it is a SW that allows computers to understand human language, or natural language, and provide answers to questions.
There is a company that participated in this project, such as ETRI and KAIST. It is called 'SALTLUX'. Since its founding in 2000, Saltlux has been securing AI core technologies, including natural language processing, semantics, and inference. It sells enterprise search solutions and translation software based on its natural language processing engine.
Saltlux's core technologies include 'BigO' based on big data and 'ADAMs' based on artificial intelligence.
Saltlux's 'Big O' platform
BigO, short for BigOcean, encompasses six proprietary engines: Storm, a knowledge storage engine; Rainbow, a knowledge visualization engine; Blue Bolt, a big data analysis engine; Tornado, a knowledge asset collection engine; LEA, a natural language processing engine; and Discovery, a semantic mining engine. Covering the entire analytics value chain, from big data collection, transformation, analysis, visualization, and decision support, BigO is built on core foundational technologies such as natural language processing (NLP), machine learning, and distributed parallel processing, along with over 8 billion social and public big data records.
Big-O's strength lies in its ability to optimize its big data analytics platform and LEA by integrating them with open-source big data frameworks. It integrates with the latest big data frameworks, including Hadoop2 and HBase, as well as Impala, Spark, and Kafka-Storm, making it suitable for a variety of production environments. It offers real-time analytics capabilities based on Storm and Spark, and integrates with BlueVault to provide CEP and real-time anomaly detection capabilities. This platform addresses the limitations of open frameworks, such as lack of stability, functionality, and performance, through over 800 use cases.
Adam, Korea's first commercialized AI platform, boasts Asia's largest knowledge base and offers over 60 APIs for knowledge, language, and visual recognition. It excels at knowledge learning and complex reasoning, achieving a 94% accuracy rate in everyday Q&A. It consists of Adam Intelligence, Adam Analytics, and Adam Data Hub. To build a large, high-quality knowledge base, it uses a "dual-spiral" methodology, where machines and humans collaborate.
Adam Assistant Implementation Case Study
Adam Assistance is a chatbot-integrated consulting system that proactively answers customer questions through channels like chat and email, and then asks questions in return. It enables not only repetitive tasks but also in-depth Q&A and conversations.
Japanese electronics company DNP has developed "Mina," a virtual counselor for Saltlux and Mizuho Bank. Additionally, Monex Securities, Sompo Insurance Japan, and All Nippon Airways (ANA) have developed virtual counseling services using Adam. In Korea, Shinhan Bank, Woori Bank, and Nonghyup Bank are also working on projects.
A Saltlux representative explained, "When Adam first began commercialization, it was at a similar level to other platforms in a third-party in-depth QA quality assessment conducted by S Electronics. Through five months of domain knowledge learning, we improved it from 30% to 70%. This is a level that surpasses Amazon's Alexa."
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.

.png)












