Presented By O'Reilly and Cloudera
Make Data Work
March 28–29, 2016: Training
March 29–31, 2016: Conference
San Jose, CA

Building a marketplace: Eventbrite's approach to search and recommendation

John Berryman (Eventbrite)
1:50pm–2:30pm Wednesday, 03/30/2016
Average rating: ****.
(4.00, 7 ratings)

Prerequisite knowledge

Attendees should have a basic understanding of the domains of search and recommendation.

Description

Historically, Eventbrite has been a platform that allows event organizers to easily market and manage their events, but recently, it has taken an active role in connecting potential attendees to events that they will enjoy. Right now, Eventbrite is focused on an effort it calls “data-driven discovery”—rapidly developing the infrastructure and algorithms that will allow it to model its users and make event search and recommendation a highly personalized experience.

But this is no small undertaking, and Eventbrite has some interesting challenges to overcome. A chief concern is that, unlike movies or consumer goods, events are an unusually short-lived type of product. Netflix and Amazon use customer interactions to build rich recommendation models of their products, but when an event is published to Eventbrite there is no user-interaction data. By the time the event is finished, Eventbrite may have only started to adequately understand how the event matches to users. To address this issue, Eventbrite is implementing a hybrid recommendation methodology that starts with purely content-based recommendations and then incorporates collaborative-filtering recommendations as information becomes available. What’s more, as search and recommendation are cut from the same fabric, search too will be customized to Eventbrite’s individual users’ tastes, allowing for serendipitous event discovery.

Topics include:

  • The nitty-gritty details of Eventbrite’s problem space.
  • Eventbrite’s approach. In particular, is a search engine a good technology for serving up recommendations?
  • Eventbrite’s pain points and how it overcame them. What did it try? What failed? How did it scale? How did it determine the quality of its recommendations?
Photo of John Berryman

John Berryman

Eventbrite

John Berryman’s first career was as an aerospace engineer, but after several years in aerospace, he found that he most loved his job when he was either programming or working on a good math problem. Eventually, John cut out the aircraft and satellites and started working full time with software development, infrastructure architecture, and data science. These days, John works at Eventbrite, helping to build out the event discovery and data platform.