#smxlondon Everything You Need to Know About How GraphSearch Works in 15-ish Minutes
Upcoming SlideShare
Loading in...5
×
 

#smxlondon Everything You Need to Know About How GraphSearch Works in 15-ish Minutes

on

  • 1,522 views

acebook’s recently unveiled Graph Search is currently being rolled out as a small private beta, but over time all billion-plus users will have access to it, so it’s something all online marketers ...

acebook’s recently unveiled Graph Search is currently being rolled out as a small private beta, but over time all billion-plus users will have access to it, so it’s something all online marketers should start thinking about. While it’s currently not a direct threat to Google, it’s very good at people search, local & vertical search and entertainment search. Come hear our panel discuss how graph search is unique, how it will change SEO, and the opportunities and challenges it’s likely to present for advertisers and marketers.

Statistics

Views

Total Views
1,522
Views on SlideShare
1,357
Embed Views
165

Actions

Likes
5
Downloads
11
Comments
0

2 Embeds 165

https://twitter.com 164
http://pinterest.com 1

Accessibility

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

#smxlondon Everything You Need to Know About How GraphSearch Works in 15-ish Minutes #smxlondon Everything You Need to Know About How GraphSearch Works in 15-ish Minutes Presentation Transcript

  • Everything You Needto Know About HowGraphSearch Worksin 15-ish MinutesKelvin Newman@kelvinnewman
  • Or if this presentationhas a sub-title...
  • Edges, Nodes and aFrickin’ Unicornhttp://www.escapefromcubiclenation.com/
  • Strategy DirectorSiteVisibilityA digital agency specialising in retail,travel and financial services
  • OrganiserBrightonSEO/ContentMarketing ShowTwo Free (and awesome) Conferences
  • Co-FounderClockwork TalentDecent Digital Recruitment
  • Not here to convinceyou GraphSearch willcatch on but...
  • If the area of this sliderepresents all thetraffic on the internet
  • This muchis Facebookhttp://mashable.com/2010/11/19/facebook-traffic-stats/
  • And every thingin grey is the restof the internet
  • Google, YouTube,Wikipedia, TheDaily Mail, etc.
  • your website, mywebsite, herwebsite etc.
  • we’re fightingover the scraps
  • If anyone canbuild a Google-Killerit’s Facebook...
  • There’s a fundamentaldifference betweenFacebook & Google
  • is about...
  • documents and linksJD Hancock
  • is about...
  • things and relationshipsJD Hancock
  • this difference issubtle buthuge
  • but I think it worksbetter for the webas we know itJD Hancock
  • Facebook’s datahas a far moreexplicit structurethan traditionalweb textD Hancock
  • it’s not that trickyfor Google to parse“I Like Nerf Guns”porkist
  • they could even have a go at “I was atCattlegrid in Leeds forLunch Yesterday”**if you mark it up in the right wayR_Savvy
  • but has a much harder jobunderstanding “Kelvin ismarried to Carolyn”
  • Facebook knows thathappened in 2007
  • And who attendedthe ceremony
  • And when we gotengaged
  • etc.
  • On GraphSearch you’renot really making asearch.You’re just filtering astructured database of allthe data Facebook has.
  • The Problem
  • But it’s a bloodybig databaseJD Hancock
  • 1 Billion UsersEvery Month
  • 240 Million Photo’sPer Day
  • 2.7 Billion LikesEveryday
  • People share billions ofpieces of contenteveryday
  • One trillion connectionsof a thousand differenttypes
  • 1,000,000,000,000
  • The Solution?
  • The AforementionedFrickin’ Unicorn
  • But before we get into theunicorn,let’s take a step back anddefine some terms
  • Edges & Nodes
  • Nodes are Nouns
  • Edges are Verbs
  • Every User, Page, Photo,Post & Place is a NodeJD Hancock
  • Every friendship, checkin,tag or like is an EdgeJD Hancock
  • Each Node has Meta-Datalike description, this howthe old FB Search“worked”
  • GraphSearch Allows yousearch the Edges as well as theNodesJD Hancock
  • Back the the Unicorn
  • Unicorn is and“inverted index system”
  • an inverted index (alsoreferred to as postings fileor inverted file) is an indexdata structure storing amapping from content, suchas words or numbers, to itslocations in a database file,or in a document or a set ofdocuments. The purpose ofan inverted index is to allowfast full text searches, at acost of increasedprocessing when adocument is added to thedatabase.
  • The main components ofUnicorn are:■ The index -- a many-to-many mapping from attributes(strings) to entities (fbids)■ A framework to build the index from other persistentdata and incremental updates■ A framework to retrieve entities from the index based onvarious constraints on attributes
  • Suppose your friend has fbid 1234 and lives in NewYorkand likes Downton Abbey.The index corresponding toyour friend will include the mappings:             friend:10003 → 1234            lives-in:111 → 1234            like:222 → 1234Here, we assume your fbid is 10003, and the fbid’s of NewYork and Downtown Abbey are 111 and 222 respectively.In addition, friend:10003, lives-in:111, and like:222 maymap to other users that share these attributes.
  • Unicorn makes it easy to find nodes that areconnected to another node by searching for anedge-type combined with an input node. E.g.:■Your friends:  friend:10003■People who live in new york: lives-in:111■People who like downtown abbey: like:222
  • ‘Facebook use query-independent signals to come upwith a numeric value for importance.This value is called the “static rank” of the entity.’JD Hancock
  • What makes up static rank is still upfor debate, but sensibly could beinformed by the elements ofEdgerankakathe newsfeed algo
  • Affinity
  • Weight
  • Decay
  • But what do I do?
  • The value oflegitimate likes fromwell connectedpeople just increased
  • Mark Up using theOpen GraphProtocolhttp://ogp.me/
  • You need an‘AffinityAcquisitionApproach’
  • Constantly BuildAffinitysubodh_chettri
  • Ask Questionsfontplaydotcom
  • Have a voteChodHound
  • BaitTextderekGavey
  • Tease
  • Get people totag you
  • Do good SocialMarketing
  • tl;drGraph Search is prettyawesome but works completelydifferently to Google rankingsrely exclusively on theconnections between the userand the entity ranking, so youneed do ‘good’ Facebookmarketing with a real focus onbuilding affinity.