Big Data @ Yelp -- taming the reviews & recommendations

Big Data @ Yelp

Yelp has more than 36 million reviews of local businesses in 20 countries. But reviews are only the tip of the iceberg that is Yelp's data. In January, more than 100 million unique visitors came to yelp.com (http://yelp.com/) and Yelp was used on 9.4 million unique mobile devices. Also in January, 4.6 million calls were made and 5.7 million sets of directions were generated through Yelp's mobile apps.

All of these interactions provide rich data for Yelp to analyze to make it even easier for people to connect with great local businesses. MapReduce (using mrjob and Amazon's Elastic MapReduce) is the workhorse that powers features like "review highlights", "people who viewed this also viewed...", and the statistics used in online scoring for search and ads. Realtime statistical analysis and monitoring within the ads system is built using Tornado and Scribe and an application specific Python framework that reuses offline reducers. Future ideas include adapting the MapReduce model to realtime stream processing.

In this session Jimmy will go over some of the technologies used in Yelp and answer your questions.

Agenda:
6pm - arrive, networking
6:30pm - session starts
7:30pm - more networking

This event is hosted by Samsung R&D
(Free Food & Drinks!)

About Jimmy Retzlaff:
Jimmy Retzlaff manages the Ads Team within engineering at Yelp. He joined Yelp in 2010 as an engineer and moved to management in 2012. Previously Jimmy worked at Amazon's Lab126 developing on-device content indexing and search for the first three generations of the Amazon Kindle. Jimmy has a B.S. in Mathematics from Harvey Mudd College and has been developing software professionally for nearly 25 years.