Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Mercari Tech Conf 2018 Keynote (Dr. Mok Oh)

mercari
October 04, 2018

Mercari Tech Conf 2018 Keynote (Dr. Mok Oh)

Mok Oh, PhD
CTO, Mercari US

Mercari JP CTO Suguru Namura, Mercari US CTO Mok Oh, and Merpay Director Keisuke Sogawa will discuss the events that have occurred at their respective companies in the past year. They will also talk about the goals and technical challenges that they plan to tackle in the future.

mercari

October 04, 2018
Tweet

More Decks by mercari

Other Decks in Technology

Transcript

  1. Seller Buyer Item Item Name Price Category Brand Description Photos

    Date Size Weight … Name Email Balance Followers Following Coupons Badges Items listed Items sold Items bought Likes Social Payment … Name Email Balance Followers Following Coupons Badges Items listed Items sold Items bought Likes Social Payment …
  2. All Items ~100M Unique Product “SKUs” f Item matching Retail

    price Size Weight Category Brand Description Photos URLs …
  3. All Items All Sellers All Buyers f Listing optimization f

    Demand optimization f Pricing optimization f Similar items f Category & brand normalization f Item sellability f Item buyability f Fraud items f Item genome f Supply optimization f Seller cancellation f Top sellers f Similar sellers f Fraud sellers f Seller genome f Top buyers f Similar buyers f Buyer cancellation f Fraud buyers f Buyer genome f Sentiment analysis f CS auto classifier f CS chatbots fML
  4. f

  5. f ( ) output = Item Name Price Category Brand

    Description Photos Date Size Weight …
  6. [ ] n-dimensional vector f ( ) = x1 x2

    x3 … xn Item Name Price Category Brand Description Photos Date Size Weight … Genome
  7. - f ( ) = f ( ) - f

    ( ) = f ( ) fsimilarity ( , ) = 0.78 fsimilarity ( , ) = 0.12 “If you liked , then you may also like .”
  8. Data Science Analysis - Female Buyers Technology Mom Buyer Women’s

    Fashion Women’s Accessories Home Technology Beauty Vintage Mom Buyer Submarkets Identified Women’s Fashion Women’s Accessories Submarkets with highest growth rates, STR, and GMV High Growth Submarket Opportunities:
  9. Data Science Analysis - Male Buyers Submarkets with highest growth

    rates, STR, and GMV Technology Submarkets Identified Technology Home Men’s Fashion Collectibles Women’s Accessories High Growth Submarket Opportunities: Home Collectibles
  10. f ( ) Sale probability within 72 hours = i.e.

    What is it’s Sellability Score?
  11. f ( ) Counterfeit? = f ( ) Churn? =

    f ( ) Own this item to list? = f ( ) Buy this item? = …
  12. Design PM iOS Android Web Backend ML Data Eng QA

    Growth Search Conversion Completion Support Foundation Teams Scrum Themes