RecSys 2022: Recap, Favorite Papers, and Lessons
Eugene Yan
RecSys 2022, held September 18-23 in Seattle, received 50% more industry submissions than 2021 with 15 industry papers integrated into the main track. Key papers covered sequential recommendations using recency-based sampling, off-policy bandits for industry applications, efficiency improvements in ads recommendation models, and unified frameworks for multiple recommendation tasks. The conference emphasized reproducibility, fairness, privacy, and real-world applications, with notable findings that model implementation significantly affects results and dataset selection matters for performance comparisons.
Why it matters
My three favorite papers, 17 paper summaries, and ML and non-ML lessons.
