30
SmartAds: Bringing Contextual Ads to Mobile Apps Suman Nath , Microsoft Research Felix Xiaozhu Lin, Rice University Lenin Ravindranath, MIT Jitu Padhye, Microsoft Research

SmartAds : Bringing Contextual Ads to Mobile Apps

  • Upload
    yosefu

  • View
    57

  • Download
    1

Embed Size (px)

DESCRIPTION

SmartAds : Bringing Contextual Ads to Mobile Apps. Suman Nath , Microsoft Research Felix Xiaozhu Lin, Rice University Lenin Ravindranath, MIT Jitu Padhye, Microsoft Research. Mobile Apps and Ads. Ad spending proportional to time spent. Mobile is an exception. 1.8x Web In 2012. - PowerPoint PPT Presentation

Citation preview

PowerPoint Presentation

SmartAds: Bringing Contextual Ads to Mobile AppsSuman Nath, Microsoft Research Felix Xiaozhu Lin, Rice UniversityLenin Ravindranath, MIT Jitu Padhye, Microsoft ResearchMobile Apps and AdsAd spending proportional to time spent1.8x Web In 2012Mobile is an exceptionSources: VSS, Mary Meeker (KPCB), ComScore, Alexa, Flurry

Consumers say mobile ads are irrelevant

Green coffeeTrucking industry

Personal LoanSpray and pray ads just dont cut it on mobile

Contextual Ads on the Web

On Mobile Apps

Enabling Contextual Ads On Web

Advertising network crawls Web pagesExtracts advertising keywords offlineUse URL keywords mapping onlineOffline crawling of data inside app is challenging

Need to run/interact with apps for Cloud data

Data may change with location and time Online keyword extraction (in client-server) is tricky

Accuracy, efficiency, and privacy trade-offsOur goal: In-app Contextual Ad

Sports adsBars ads

Ringtone adsOutlineDoes contextual ads make sense for mobile apps?Measurements with 1200 Windows Phone Apps

How can we enable it?SmartAds system

How effective is contextual advertising?User study and measurementsMeasurement MethodologyAppPhonePageDataAdvertising Keyword ExtractorPage datakeywordsTop 1200 non-game Windows Phone apps (from overall top 2000)Salon, Haircut

Are there prominent keywords in app pages that match available ads in ad network?

One week bidding keywords trace from Microsofts ad network PhoneMonkeyAutomatically runs and explores apps

Scrapes any data shown to userPhone Monkey

Keywords in PageData PageData is a good source of ad keywordsContextual advertising has good potentialHalf the apps have >20 keywordsIs MetaData Good Enough?

PageData(PhoneMonkey)MetaData(App Store)85% apps have more keywords in PageDataPageData has more keywords than MetaDataPageData-based targeting has more potential than MetaData-based targetingPageData DynamicsHalf the apps have session similarity < 0.55Page data is dynamicNeed online keyword extractionOutlineDoes contextual ads make sense for mobile apps?Extract keywords from PageData, during run time

How can we enable contextual ads in apps?SmartAds system

How effective is contextual advertising?User study and measurementsSmartAdsSalon serviceadsdetails, haircut, up, to, salon, .. Online Keyword extraction

SmartAds Server

Ad Control

Offline Crawling(Ad, keyword) inventoryApp keywordsSmartAds GoalsAccurate: ads relevant to page content

Efficient: small memory and network overhead

Private: dont send sensitive page data out

Impossible to maximize all in a client-server design [Hardt, CCS13] AccuracyPageFrequencyCapitalizationFontSizeBidFrequencySalon services, haircut0.40.20.30.7

Use state-of-the-art ad-keywords extractor KEX[Yih, WWW06](See paper for our extensions)For each word:Where to extract keywords?Do in phone? Large memory footprint: ~100 MB dictionary of bidding keywordsDo in Server? Bad privacy: send page content to Cloud~5KB network bw per page

We do partly in phone, partly in serverAchieve a reasonable balanceAccuracyPrivacyEfficiencyPhoneServer

Accuracy + Memory efficiencyPartition the scoring functionDot product is partitionableFontSizeBidding Frequency

Bidding DatabaseFeature vectorWeight vectorAccuracy + Memory efficiency + Communication Efficiency + PrivacyPhone drops words that cannot be keywordsLocal PruningPhone drops word if local weight is too small

Correctness guarantee, with bounded weight and feature valuesBloom filterPhone maintains a filter with bidding keywords

Drop words if not in the filterBloom Filter ChallengesBloom filter size Memory overhead at client

Update on keyword changesNetwork overhead at client

False positivesAccidental leak of non-keywordsAnalyze Microsofts ad networkSize: 4 monthsUse one-way hashOutlineDoes contextual ads make sense for mobile apps?Measurements with 1200 Windows Apps

How can we enable it?SmartAds system

How effective is SmartAds?User study and measurementsPerformance measurementPrototype implemented for Windows Phone (client) and Windows Azure (server)

Performance measured on a Samsung Focus phoneCPU50ms at client,