Large-Scale Behavioral Targeting (The Art of Scaling Up Simple Algorithms)

0
5312

AUTHORS

Ye Chen(Microsoft Corporation | yeoechen@ahoo.com)

Dmitry Pavlov (Yandex Labs | pavlov@yandexteam.ru)

John F. Canny (Computer Science Division, UC Berkeley | jfc@cs.berkeley.edu)

 

PUBLICATION YEAR

2009

ABSTRACT

Behavioral targeting (BT) leverages historical user behavior to select the ads most relevant to users to display. The state-of-the-art of BT derives a linear Poisson regression model from fine-grained user behavioral data and predicts click-through rate (CTR) from user history. We designed and implemented a highly scalable and efficient solution to BT using Hadoop MapReduce framework. With our parallel algorithm and the resulting system, we can build above 450 BT-category models from the entire Yahoo’s user base within one day, the scale that one can not even imagine with prior systems. Moreover, our approach has yielded 20% CTR lift over the existing production system by leveraging the well-grounded probabilistic model fitted from a much larger training dataset.

Specifically, our major contributions include: (1) A MapReduce statistical learning algorithm and implementation that achieve optimal data parallelism, task parallelism, and load balance in spite of the typically skewed distribution of domain data. (2) An in-place feature vector generation algorithm with strict linear-time complexity O(n) regardless of the granularity of sliding target window. (3) An in-memory caching scheme that significantly reduces the number of disk IOs to make large-scale learning practical. (4) Highly efficient data structures and sparse representations of models and data to enable fast model updates. We believe that our work makes significant contributions to solving large-scale machine learning problems of industrial relevance in general. Finally, we report comprehensive experimental results, using industrial proprietary codebase and datasets.

FULL TEXT

 

About The Author

Ye Chen

Research leadership in the areas of statistical machine learning, data mining, and large-scale algorithms. An impressive career trajectory at top industrial research organizations including Microsoft, Yahoo!, and eBay. A superb track record of deployments of behavioral targeting, recommender systems, click and conversion prediction, real-time bidding, for display advertising, sponsored search, and online marketplaces. Designed and implemented the next-generation behavioral targeting system for Yahoo!, the next-generation recommender system for eBay, click and conversion prediction, impression value prediction, and real-time bidding algorithms for Microsoft. First-authored patents in behavioral targeting, sponsored search, large-scale user clustering, recommender systems, real-time bidding for performance ads, Hadoop and grid implementation. First-authored publications in NIPS, KDD, SIGIR, WWW, among others. Our behavioral targeting paper won the Best Application Paper Award at KDD ’09, and our recommender system paper won the Honorable Mention Award at SIGIR ’11. (Source: LinkedIn.com)

 

Leave a Reply