Exploring an Optimal Online Model for New Job Recommendation
Masahiro Sato, Koki Nagatani, Takuji Tahara · 2017
The RecSys Challenge 2017 is a competition of recommender systems including both the offline and online phases. The task of the challenge is recommending new jobs to users considering the satisfaction of both the users and the job recruiters. This paper presents our approach to this challenge. We built content-based recommendation models using gradient boosting decision trees. The models were finely tuned for both the offline and the online phases using different internal evaluation protocols. Our approach achieved 6th place in the preliminary offline phase and 3rd place in the final online phase.