=Paper=
{{Paper
|id=Vol-2715/keynote
|storemode=property
|title=Three Challenges in Building Industrial-Scale Recommender Systems
|pdfUrl=https://ceur-ws.org/Vol-2715/keynote.pdf
|volume=Vol-2715
|authors=Sebastian Schelter
|dblpUrl=https://dblp.org/rec/conf/recsys/Schelter20
}}
==Three Challenges in Building Industrial-Scale Recommender Systems==
<pdf width="1500px">https://ceur-ws.org/Vol-2715/keynote.pdf</pdf>
<pre>
    Three Challenges in Building Industrial-Scale Recommender
                             Systems
                                                                 Sebastian Schelter
                                                             University of Amsterdam
                                                            Amsterdam, The Netherlands
                                                                 s.schelter@uva.nl
ABSTRACT                                                                       SPEAKER BIOGRAPHY
We have seen astonishing progress of machine learning research                 Sebastian Schelter is an Assistant Professor with the University of
in the last years. Unfortunately, it is often difficult to translate           Amsterdam, conducting research at the intersection of data man-
this academic progress into deployable applications, due to the                agement and machine learning. He manages the AI for Retail Lab
constraints and challenges imposed by production settings. In this             Amsterdam, and has a joint appointment as Research Fellow at
talk, I will present some of my recent research in the area of data            Ahold Delhaize, an international retailer based in the Netherlands.
management for machine learning, which tackles these problems.                 His work covers many aspects, such as automating data quality val-
Furthermore, I will put a special focus on three challenges that I see         idation, optimizing programs that combine operations from linear
for building industrial-scale recommender systems. In particular,              and relational algebra or tracking the lineage of machine learning
I will outline ideas on how to scale to datasets with billions of              pipelines. In the past, he has been a Faculty Fellow with the Center
interactions, understand the impact of response latency on the                 for Data Science at New York University and a Senior Applied Sci-
performance of a deployed recommender system, and make the                     entist at Amazon Research, after obtaining his Ph.D. at the database
"right to be forgotten" a first-class citizen in real-world ML systems.        group of TU Berlin with Volker Markl. He is active in open source
Reference Format:                                                              as an elected member of the Apache Software Foundation, and has
Sebastian Schelter. 2020. Three Challenges in Building Industrial-Scale Rec-   extensive experience in building real world systems from my time
ommender Systems. In 3rd Workshop on Online Recommender Systems and            at Amazon, Twitter, IBM Research, and Zalando.
User Modeling (ORSUM 2020), in conjunction with the 14th ACM Conference        ORSUM@ACM RecSys 2020, September 25th, 2020, Virtual Event, Brazil
on Recommender Systems, September 25th, 2020, Virtual Event, Brazil.           Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons
                                                                               License Attribution 4.0 International (CC BY 4.0).

</pre>