Kakao Enterprise’s AI Model Tops Korean Machine Reading Comprehension Test | Be Korea-savvy

Kakao Enterprise’s AI Model Tops Korean Machine Reading Comprehension Test


(image: Korea Bizwire)

(image: Korea Bizwire)

SEOUL, March 24 (Korea Bizwire)Kakao Enterprise Corp., an artificial intelligence (AI) unit of South Korean internet giant Kakao Corp. said Wednesday that its in-house natural language processing AI model LittleBird-large grabbed the top spot in the performance test of the Korean machine reading comprehension dataset KorQuAD 2.0.

KorQuAD 2.0 is a massive dataset of Q&As on the Korean language. In the performance test, content extracted from the dataset are presented as questions, and the AI model should find the correct answers.

Kakao Enterprise’s natural language processing model ranked first with an accuracy of 90.22 points.

The processing time was 6,157 milliseconds, about 1.7 times faster than 10,434 milliseconds set in September 2020 by Samsung SDS Co.’s SDS-NET v1.3.

Kakao Enterprise plans to continue to upgrade its AI model for use in services such as search, chatbots and the AI contact center Kakao i Connect Center.

Once it is added to the Kakao i Connect Center, the AI consultant can provide more accurate answers to clients’ questions and requests using the knowledge search system that is applied by the machine reading comprehension model.

Ashley Song (ashley@koreabizwire.com)

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>