The Ever Expanding Taste Graph

THE EUER-EXPANDING TASTE GRAPH by hunch Hunch is building a 'Taste Graph' of the entire web, connecting every person on the web with their affinity for anything, from books to electronic gadgets to music, restaurants, fashion, games, or vacation spots. THE SOFTWARE THE HARDWARE Assembly was used since many computational cycles are done on large matrices which The Hunch Taste Graph is built using proprietary205 saftware written in C and Assembly. Hunch runs the Taste Graph on a super computer. overwhelm traditional 27,000 linear algebra libraries. 48 1 TERABYTE PROCESSORS OF MEMORY Lines of code went into the Taste Graph FLOATING POINT 100 BILLION OPERATIONS PER SECOND A PEEK INSIDE THE TASTE GRAPH URBANITE TOM FROM NEW YORK KIM FROM LONDON Hunch learns when two people like the same types of things, and then uses that kaowledge to make better predictions. LADY GAGA JOHN IRVING FOODE 6. NODE (USER EDGE MEINTY BETWEEN NODES) NODE TEM ORUSR ATTRIDUO DAY TWO Despite massive computing power, it takes 2 days for the Taste Graph 500 MILLION PEOPLE • 200 MILLION ITEMS • 30 BILLION EDGES WHERE THE DATA comeS FROM 1 THAY' QUESTIONS on HUNCH.Com 2 2 LITEM RATINGS ON HUNCH.Com LIKE DATA There are currently 2,000 THAYS (Teach Hunch Hunch draws upon the Taste Graph to make When users link their Hunch accounts to other content sites, Hunch learns about their ikes'. About You) questions in the Hunch question Tou) pool. New THAYS are continuously added. predicted ratings for each user, until the user submits an actual rating. MORE THAN RATNGS 80 MILLION THE AVERAGE USER HAS ANSWERED HUNCH USERS HAVE PROVIDED MORE THAN 15 MILLION BILLION 113 THAYS FACEBOOK LIKES THAYS HAVE BEEN ANSWERED ON MORE THAN 200,000 Hunch alsa indenes public review data from retail sites, review sites, travel sites, and more. Sample THAY IN A SURVIVAL SITUATION, DO YOU THINK YOU'D BE ABLE TO CONSUME HUMAN FLESH? INDIVIDUAL ITEMS SAMPLE RECOmmenORTIons 31% of users said yes 9% of users would look for anether option LOST IN SPACE ***** CHECH-IN DATA 5 SOCIAL CONNECTIONns ОРЕП АPI When users connect to Hunch from secial networks, the Taste Graph lears about their social connections. When users connect to Hunch from location-based services, the Taste Graph learns about their check-in preferences. Anyone can use Hunch's API to uplead data to the Taste Graph or get predictions from it. Applications make about THE TASTE GRAPH INCLUDES 400 MILLION USERS 400,000 EALS DAILY Hunch 2 MILLION VENUES 20 BİLLION FRIENDSHIPS There are FROM 2 MILLION AUTHENTICATED USERS OF THE API 100 MILLION USERS 3 BILLION FOLLOWS PEOPLE WHO CHECKED-IN Try it hunch.com/developers JUST HOW BIG? How big is 30 billion connections? Here it is compared to some other impressive figures. Hunch can make very accurate predictions based solely on a Facebook or Twitter name. Try it! Hunch.com/goodies HUNCH TASTE GRAPH number of edges 30 BILLION COLD START Because Hunch propagates known data to make inferences about unknown data, it has solved the 5o-called "cold start" problem. TWITTER number of follows GRAPH THEORY There are more than 30 billion 'edges (connections) in the graph between people and items. These connections represent the "affinity between a person and an item, as measured by likes, dislikes, ratings, check-ins, interactions, etc. The connections serve as the basis for Hunch's predictive ability. 4.5 BILLION GREAT WALL number of bricks 3.8 BILLION DNA number of pairs 3 BILLION MILKY WAY number of earth-like planets 2 BILLION hoduced may t e tats fempnies pther tha Munch are jl luing the websites fthe espectiue companies Hunch inc 201n