Flitto: Korea's First TTA DQ Certification for Logical Reasoning Data

Flitto Opens New Horizons for AI Inference Learning: CoT Data Becomes First in Korea to Receive TTA Grade A Certification AI language data innovation company Flitto has set a new milestone in Korea's AI learning data market. Flitto's 'Logical Inference CoT (Chain of Thought) Data' has become the first in Korea to receive the highest data quality certification (DQ) of Grade A from the Korea Association for ICT Promotion (TTA), setting a new standard for Korean-based AI inference learning data. This certification from TTA, a credible institution responsible for the standards and quality of information and communication technology, holds deep significance. Specifically, it meticulously evaluated the validity and accuracy of annotations within the data, as well as the logical consistency and reasonableness of answers. Above all, it was praised for its excellence in terms of the design quality of the logical inference process and its practical applicability. This proves that Flitto's CoT data possesses a high-dimensional structure that goes beyond a mere collection of information, enabling AI to learn the processes of self-thought and judgment. Flitto designed this logical inference CoT data to enable AI to possess high-dimensional thinking capabilities, allowing it to 'think and infer' like humans, beyond mere information retrieval. Experts from five key domains — economics, science, technology, politics, and mathematics — directly participated in the project, dedicating their efforts to structuring complex problem-solving processes into step-by-step logical frameworks. This data is meticulously structured into a total of 11 items, ranging from the question to problem definition (task), approach, initial analysis, five stages of detailed thought processes (step-by-step 1-5), and the final response (output). This dramatically enhances AI's ability to clearly 'explain' and 'judge' the entire process of reaching a final conclusion for complex questions, not by providing fragmented answers, but by progressively defining the core of the problem, deriving intermediate hypotheses, and validating the logic's validity through real-world examples. Consequently, it presents core foundational data that will innovatively elevate the inference quality of Korean-based AI. Lee Jung-soo, CEO of Flitto, emphasized, "In an era where AI goes beyond simple learning to think and judge like real humans, CoT data is an indispensable high-precision linguistic asset for AI's intellectual growth." He further stated his ambition, "Using this TTA highest-grade certification as a stepping stone, Flitto will lead data quality standards for implementing Explainable AI and firmly establish unrivaled competitiveness in the global AI learning market." Flitto's achievement is expected to become a significant turning point in the advancement of Korean AI technology.