Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Recommendation @ Meetic

1,031 views

Published on

Talk given by Wilfried Logerais, Meetic, during the RecsysFR meetup on February 1st 2017.

Published in: Internet
  • Be the first to comment

  • Be the first to like this

Recommendation @ Meetic

  1. 1. MEETIC Recommendation System Meetup 01/02/2017
  2. 2. Meetic – European online dating leader Services available on:
  3. 3. Some stats Millions of Monthly Active Users 2-3 millions messages exchanged per day between users 4 000 000 likes a day 20 000 000 profile views a day
  4. 4. Why we need recommendation at Meetic? Our Goal: Let’s any single meet the right person! • We have to help our members to focus their search to singles who meet their expectation (Gain time) • We have to ensure that interest could be mutual Too much eligible “prospect” for each user • They count on our experience in dating • Some people prefer recommendations instead of using the search feature Our users ask us to have recommendations • To ensure attention is well-distributed • To ensure engaged singles receive attention • Attention received (likes/mails) has an impact on conversion We have to master who receive attention
  5. 5. Our strengths for Recommendation at Meetic  Lots of data  Lots of features (roughly 10) on site and emails to push members  Encouraging POCs :  Collaborative filtering with Neo4j (triangulation)  Supervised modeling with all kinds of data improve matching rates High Qualified Profile • 30 profiles from Physics Criteria to Hobbies through Lifestyle Seek Criteria • We know what criteria our members are looking for! Behavioron Site • We know what our users liked (and disliked!)
  6. 6. Current Projects LARA: Create Smarter Bot to help our members in their search Similar Profile: Similar Face Approach
  7. 7. Algorithms at Meetic IT Architect Data Mngt / DBA Developers Back-end System Engineer Analytics Product If necessary, a temporary agile collocated team is created to get better synergy between teams - Propose Project - Data Modeling - Analyzed results (AB test) - Propose new projects/ New feature improve consumer experience - Manage projects - Define good architecture - Implement solutions in production
  8. 8. THANK YOU Enjoy the Meetup! https://careers.smartrecruiters.com/MeeticGroup

×