Future-‐Shaping Problems That App Stores Face 1. Open or closed app distribu<on model? 2. App Stores do not enable users to ﬁnd apps they want. 3. App Stores are not aware of user’s situa<onal and social context.
From Desktops To Gadgets • Handheld hardware created handheld soJware. • A shiJ from consuming informa<on to using func<ons.
„Apps In The Browser” Not Likely To Succeed • For most people there is no return to the desktop. • We need „open hardware” ﬁrst!
App Discovery Is Unsolved Less than 0.1% of Apps Generates More Than 50% of Monthly Downloads in Jun 2012 5,000,000 Downloads in February 4,500,000 4,000,000 2013 Downloads in Jun 2012 3,500,000 3,000,000 Android 2,2 bln 2,500,000 2,000,000 Google Play iOS 1,87 bln Apple Appstore 1,500,000 1,000,000 500,000 0 0 2000 4000 6000 8000 10000 PosiCon in Monthly Top List
A Long Tail Of Apps That Are Never Found How it really looks like State of the art 5,000,000 4,500,000 4,000,000 1. Most apps are never Downloads in Jun 2012 3,500,000 downloaded 3,000,000 2,500,000 2,000,000 2. People are not ﬁnding 1,500,000 what they want. 1,000,000 500,000 3. SoluCon to this 0 0 50000 100000 150000 200000 250000 300000 350000 400000 450000 500000 problem will shape the PosiCon in Monthly Top List future app store.
Content Discovery – Books & Papers • Fundamentally books & papers are informa<on. • We have thousands of years of experience in books discovery;> • Classiﬁca<on is becoming more automa<c.
Content Discovery -‐ Music • Hundreds of self-‐ proclaimed music genres exist. • Music is very social and self-‐organising which is leveraged by last.fm and similar services. hap://slycoder.ﬁles.wordpress.com/2010/01/meow.pdf
Intelligent App Store • App is a new type of content – An app is a piece of soJware that carries on a very speciﬁc ac<on in a short <me. – An app is deﬁned by what it does. • App Store must learn what apps can do for humans and how apps can be linked together. • With close to 2 mln. apps exis<ng and 85 000 new apps each month this process must be automa<c.
Finding What Apps Can Do Machine learning algorithms will: • Find all possible app func<ons. • Automa<cally assign each app to one or more func<on.
Showing What’s Out There • What’s out there in music?
Search Box Is Too Hard For Average User How People Search For Apps? The Implica<ons Speciﬁc AcCon Queries “Inspire Me” Queries “crop photos”, “block calls”, “games”, “fun”, “free” “view movies” Most users look for general app categories 5% 15% Some users want to be inspired: to ﬁnd a cool app or a new game A small minority look for a speciﬁc func<on 80% General Category Queries “music”, “movies”, “chat” Source: 2 years of XYO’s query log data
A Rich User Context Is Available • The device you are using knows what you do and where you are. • Social services know who you are and what and whom you like. • You are typically giving this informa<on away, disregarding any privacy concerns.
Facebook Graph Search: An Early Example Of Social Content Discovery
User As A Query What if we map what you and your friends like to what apps in app store can do for you?
App Store 2015 • A mul<tude of „Walled Gardens” • With deep understanding what humans can do with apps. • With intelligent algorithms which make long tail apps available. • With deep knowledge of user’s situa<onal and social context that delivers apps seamlessly.
Thank You! …and see all the above in ac<on hap://next.xyo.net/betasignup