Ç¥ÁØÈ­ Âü¿©¾È³»

TTAÀÇ Ç¥ÁØÇöȲ

Ȩ > Ç¥ÁØÈ­ °³¿ä > TTAÀÇ Ç¥ÁØÇöȲ

Ç¥ÁعøÈ£ TTAK.KO-10.1595 ±¸Ç¥ÁعøÈ£
Á¦°³Á¤ÀÏ 2025-12-05 ÃÑÆäÀÌÁö 21
ÇѱÛÇ¥Áظí ÀΰøÁö´É ÇнÀ¿ë µ¥ÀÌÅÍ ±¸Ãà °øÁ¤ ±â¹ÝÀÇ ¸ÞŸµ¥ÀÌÅÍ
¿µ¹®Ç¥Áظí Metadata Based on AI Training Data Construction Process
Çѱ۳»¿ë¿ä¾à ÀÌ Ç¥ÁØÀº ÀΰøÁö´É ÇнÀ¿ë µ¥ÀÌÅÍ ±¸Ãà °øÁ¤¿¡ µû¸¥ ÇÁ·Î¼¼½º¿Í ǰÁú°ü¸® ÇÁ·¹ÀÓ¿öÅ©¸¦ ±â¹ÝÀ¸·Î µ¥ÀÌÅͼ¼Æ®ÀÇ »ç¿ë ¸ñÀû°ú ¹üÀ§¸¦ ¸í½ÃÇÔÀ¸·Î½á µ¥ÀÌÅͼ¼Æ®¸¦ ¹èÆ÷Çϰí Ȱ¿ëÇϱâ À§ÇÑ ¸ÞŸµ¥ÀÌÅÍ Ç׸ñÀ» Á¤ÀÇÇÑ´Ù. ÀΰøÁö´É ÇнÀ¿ë µ¥ÀÌÅÍ ±¸Ãà °øÁ¤Àº ±¸Ãà°èȹ ¼ö¸³ ´Ü°è, µ¥ÀÌÅÍ È¹µæ/¼öÁý ´Ü°è, µ¥ÀÌÅÍ Á¤Á¦ ´Ü°è, µ¥ÀÌÅÍ °¡°ø ´Ü°è, µ¥ÀÌÅÍ ÇнÀ ´Ü°è, ÃÑ 5´Ü°è·Î ÀÌ·ç¾îÁ® ÀÖÀ¸¸ç ÀΰøÁö´É ºÐ¾ß ¹× ¸Ó½Å·¯´× ºÐ¾ß¿¡¼­ ÀϹÝÀûÀ¸·Î »ç¿ëÇÏ´Â ÀΰøÁö´É ÇнÀ¿ë µ¥ÀÌÅͼ¼Æ®¿¡ ´ëÇÑ ¸ÞŸµ¥ÀÌÅÍ Ç׸ñ°ú ¼¼ºÎ ¿ä¼ÒµéÀ» Á¦¾ÈÇÑ´Ù.
ÀÌ Ç¥ÁØÀº ±âÁ¸ÀÇ ±â¼ú ¹× ÀÀ¿ë¼­ºñ½º °üÁ¡¿¡¼­ ¹þ¾î³ª µ¥ÀÌÅͼ¼Æ®¸¦ Ȱ¿ëÇÏ´Â ÀÌ¿ëÀÚ °üÁ¡¿¡¼­ ÀΰøÁö´É Åõ¸í¼º È®º¸¸¦ À§ÇÏ¿© µ¥ÀÌÅͼ¼Æ®ÀÇ ±¸¼º Á¤º¸¿Í »ç¿ë ¹æ¹ý»Ó¸¸ ¾Æ´Ï¶ó µ¥ÀÌÅͼ¼Æ® »ç¿ë½Ã ¹ß»ýÇÒ ¼ö ÀÖ´Â À§Çè, ÆíÇâ, ÇÑ°è µîÀÇ Á¦¾à»çÇ× Á¤º¸¸¦ Æ÷ÇÔÇÑ´Ù. ÀÌ¿Í °°ÀÌ À¯¿ë¼ºÀ» ¸ñÀûÀ¸·Î ÇÑ ÀÌ¿ëÀÚ °üÁ¡ÀÇ ¸ÞŸµ¥ÀÌÅ͸¦ Á¦°øÇÔÀ¸·Î½á µ¥ÀÌÅͼ¼Æ®¸¦ ¹èÆ÷ ¹× °ü¸®, ÀÌ¿ëÇÏ´Â ¸ðµç »ç¿ëÀÚµéÀÌ ÀΰøÁö´É °³¹ß ¹× ¼­ºñ½º ¾÷¹«¸¦ ¼öÇàÇÏ´Â µî ÀΰøÁö´É Ȱ¿ëÀû Ãø¸éÀÇ Åõ¸í¼º È®º¸¿¡ µµ¿òÀÌ µÉ ¼ö ÀÖ´Ù.
¿µ¹®³»¿ë¿ä¾à This standard defines metadata for distributing and utilizing datasets by specifying the purpose and scope of dataset use, based on the process and quality management framework for building AI learning data. The AI learning data building process consists of five stages: building plan development, data acquisition/collection, data refinement, data processing, and data training. It proposes metadata items and detailed elements for AI learning datasets commonly used in the fields of AI and machine learning.
This standard moves beyond the traditional perspective of technology and application services to ensure AI transparency from the perspective of users. It includes information on dataset composition and usage methods, as well as potential risks, biases, and limitations that may arise during dataset use. By providing user-focused metadata aimed at usability, it can help ensure transparency in AI utilization, including AI development and service tasks, for all users distributing, managing, and utilizing datasets.
±¹Á¦Ç¥ÁØ
°ü·ÃÆÄÀÏ TTAK.KO-10.1595_[1].pdf TTAK.KO-10.1595_[1].pdf            

ÀÌÀü
´ÙÀ½