andrei alexandrescu keynote at csdn 2007 in beijing, china

19

Upload: andrei-alexandrescu

Post on 19-Jan-2017

237 views

Category:

Technology


0 download

TRANSCRIPT

1 / 19

Beyond Web 2.0

Andrei Alexandres u

Introdu tion

Introdu tionThe Next Web?Web 2.0 Re apThe Wisdom ofCrowdsThe Computer isEverywhereSyndi ating andCombining DataThe Symbioti WebCan We Do It?2 / 19

The Next Web?3 / 19

n Web 2.0 approa hes maturityn Not a parti ular te hnology or te hnologi albreakthrough sin e Web 1.0

n Question: what will the next Web look like?u One that does embed a te hnologi alrevolution?

Web 2.0 Re ap4 / 19

n \The Wisdom of Crowds"u User input ( ontent and opinion) is essentialu Bidire tional dialog instead of one-way deliveryn Content on more than omputersu Cell phones, players, PDAs...n The Web is a programmable platformu Huge amounts of easily-a essible datau Maps, books, produ ts, movies. . .u Combining data reatively is keyn Web 2.0 is really \Web done right"

The Wisdom of Crowds5 / 19

n Anyone an publish anything on the Webn People sele t the good from the bad through votingn Without votes: an amorphous anar hy of good andbad ontentn With votes: a hierar hy of quality throughpopularityn Issue: not everything popular is goodn Issue: ma hines and humans annot easily\ ooperate" over the same data

The Computer is Everywhere6 / 19

n Computers are getting smaller, more apable, andless re ognizablen Web 2.0 means a ess of the same data on alldevi esn Issue: input methods

Syndi ating and Combining Data7 / 19

n Progress in bandwidth, storage, and omputing havemade it possible to amass huge amounts of datan Issue: Mu h data is either for humans or ma hines;there's no synergy!u Sear h youtube. om for something that hasbeen said?u Figure out if an opinion (on a produ t, movie,person. . . ) is positive or negative?u Summarize many pages into a short digest?u Sear h foreign-language pages for information?

The Symbioti WebIntrodu tionThe Symbioti WebVisionNatural LanguagePro essing (I)Natural LanguagePro essing (II)Spee h (I)Spee h (II)Spee h (III)Virtual RealityCan We Do It?

8 / 19

Vision9 / 19

n Today: killer Web apps (email, IM, blogs, RSS)in lude humans as the essential parti ipantsn The ma hine just relays data!n Even Google's massive automation relies on humansstatisti ally reating \interesting" links

n Natural progression: ma hines will be an a tiveparti ipant to the Web thus adding value to itn The Web will be ome a ollaborator, not only aplatform, in our everyday tasks

Natural Language Pro essing (I)10 / 19

n Progress in statisti al ma hine learningu Requires huge amounts of data for trainingu Computationally intensiven We're at a parti ular point in omputing history:u Huge quantities of data availableu Powerful inter onne ted omputers

n Future: Computers will \read" and \understand"pages intended for humans better and better

Natural Language Pro essing (II)11 / 19

n Amazon 1.0: \This book is ranked 3.5" (dire tintegration of user s ores)n Amazon 2.0: \Be ause of your and others' shoppinghistory, you may like. . . " (simple data mining)

n Future: \This book's style may appeal to you, andyour blog reveals your keen interest in dystopi s i-�. Here's a synopsis. . . "u Issue: resolve the fear that the omputer followsevery step we make

Spee h (I)12 / 19

n Spee h 1.0: Person yelling in front of a huge omputern Spee h 2.0: Person risply di tating to a desktop omputern Future: Dialog with the Web over the ell phoneu Dialog systems are already usable and will be ommoditizedu Cell phones be ome more wearableu All buttons in ell phones be ome unne essary

Spee h (II)13 / 19

n Calendar, email, news, messaging, trading|all anbe handled over the ell phone via dialogn The Web is already a repository of our datan Now the Web be omes a virtual se retary that helpswith routine tasksn The ell phone's disadvantage be omes a strategi advantageu Most people would rather talk than on a smallkeyboard

Spee h (III)14 / 19

n Spee h be omes just another input method over amesh of Web servi esn Telephony will be ome \programmable"n Translation servi es will be ome ommoditizedn Real-time onversation between an English speakerand his Chinese in-laws, with automated translation

Virtual Reality15 / 19

n Rendering te hnologies already existentn Virtual assistants reate a ompelling appli ation ofVRn Expe t more VR intera tion on the Web

n Issue: onne tion speedu Bandwidth on the risen Issue: input methodu Area of a tive resear h

Can We Do It?

Introdu tionThe Symbioti WebCan We Do It?IssuesA Ra e AgainstTimeThe Con lusion...16 / 19

Issues17 / 19

n Software Engineering makes slow progressn Computing power may stop short of a hieving trueintelligen en Global problems might slow us down or even stop usn We live on a planet with limited resour es

n We are ra ing towards a solution that would help ussolve our growing problems

A Ra e Against Time18 / 19

n Will we make an intelligent Web in time?

or. . .n Will we run out of smarts?n Will we run out of omputation?n Will we run out of fuel?n Will we run out of Earth?

The Con lusion...19 / 19

. . . is left to you.