What Quora’s VP Reveals About Building Real‑World Recommender Systems

Introduction

Xavier Amatriain, VP of Engineering at Quora, presents the experience of building Quora’s recommender system (recsys) and the key lessons learned.

Quora’s Data

Quora benefits from massive high‑quality textual content and extensive data relationships, providing a rich foundation for recommendation algorithms.

Quora’s Recommender System

Recommendation is embedded throughout Quora’s product, influencing many user‑facing features.

Lessons Learned

Implicit signals beat explicit ones (almost always).

Be thoughtful about your training data.

Your model will learn what you teach it to learn.

Explanations might matter more than the prediction.

If you must pick one approach, matrix factorization is the best bet.

Everything is an ensemble.

Building recommender systems is also about feature engineering.

Understanding why you should answer questions is crucial for your recsys.

Data and models are great, but the right evaluation approach is even better.

You don’t need to distribute your recsys; distributed architecture isn’t always necessary.

Conclusion

The presentation summarizes practical insights that can guide engineers building large‑scale recommendation engines.

Slide deck: http://www.slideshare.net/xamat/recsys-2016-tutorial-lessons-learned-from-building-reallife-recommender-systems