Explore in Timeline : from a ML pipeline perspective -English version-

Hochul Kim / LINE Plus

■ Session Overview
Explore in Timeline is a Machine Learning(ML) based service that offers content that users may be interested in based on their behavior in various LINE services. During the recommendation process, users’ preferences and content categories are extracted and analyzed in various ways, and in turn are used as features for service. In general, providing a ML-based service not only requires ML algorithms, but also integration with other technology areas such as Frontend/Backend, DB, network, and distributed computing. Each task has dependencies with one another in terms of being put into service. The ML-Pipeline is responsible for designing and managing the overall ML service by organizing and deploying these tasks and corresponding physical resources. Explore is a ML-service that provides services by managing tasks and required resources using pipelines. In this talk, I will share experiences of implementing and optimizing pipelines that are used in the Explore service.


■ Official Site #linedevday
https://linedevday.linecorp.com/2020/...

■ Category
AI&Data - ML