Upstage ranks 12th in global AI recommendation competition... highest position among domestic companies

[Asia Economy Reporter Kang Nahum] AI startup Upstage announced on the 2nd that it ranked 12th, the highest among domestic companies, in the 'ACM RecSys Challenge 2022' competition hosted by the world's most prestigious conference in the AI search recommendation field.

The ACM RecSys Challenge, held annually since 2007, addresses AI recommendation problems in various fields such as music, social networking services (SNS), and commerce. Universities researching AI recommendation worldwide, as well as global companies, participate to validate recommendation models. Previous winners include Alibaba from China, Yandex from Russia, and NVIDIA.

In this competition, Upstage achieved higher performance by utilizing a transformer-based neural network model that tracks relationships within sequential data and learns context and meaning, instead of a boosting model that first selects candidate items to be recommended and then ranks them.

The transformer model applies continuously evolving mathematical techniques to detect subtle changes in the meanings of data elements that are distant from each other depending on their relationships. Upstage used this to understand the relationships between click histories on a page and directly generate products that users are likely to purchase, achieving differentiation.

Seonghoon Kim, CEO of Upstage, said, "With the highest domestic ranking in the challenge held by ACM RecSys, the most prestigious conference in recommendation AI, Upstage's recommendation technology has once again been recognized as a global top technology. Our proven AI Seargest (search recommendation) technology will help all companies and industries that require hyper-personalization."

Meanwhile, Upstage plans to release the no-code/low-code solution ‘Upstage AI Pack’ by the end of this year, which enables customers to utilize AI technologies such as OCR technology that extracts desired information from images and natural language processing search technology in a customized manner all at once.

This content was produced with the assistance of AI translation services.