The Google Age The Path: Search Engines goto Artifical Intelligence Professor Dr. Eduard Heindl.

The Google Age

The Path: Search Engines goto Artifical IntelligenceProfessor Dr. Eduard Heindl

Eduard Heindl, Heindl Internet AG

Topics

Googles history The Google technology Why is Google special Why is Google so powerfull The future of Google


Google Stone Age 1995 meeting of

Sergey Brin (23) and Larry Page (24)

1996 BackRub System starts at Stanford University

100.000$ support by Andy Bechtolsheim

7. September 1998 Start Google Inc.

*September 21, 1999, the beta label came off the website.


Googles Philosophy

perfect search enginedefined by co-founder Larry Page

as something that:

"understands exactly what you mean and gives you back exactly what you want."


Live of a Query

Quelle: http://www.google.com/corporate/query.html


The PageRank Google sorts by PageRank The more links point to a

document, the higher is the rank

But not all links are equal, the PageRank of the referee counts too!

A recursives problem „solving an equation of

more than 500 million variables and 2 billion terms“ (source: Google)

Link

AA

LL MMNN

GG

OO

FFEE

CCBB

HHDD

II KK


The Link MatrixA B C D E F G H I K L M N O

A 0 0 0 0 0 0 0 1 0 0 0 0 0 0

B 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C 0 0 0 0 1 0 0 0 0 0 0 0 0 0

D 0 0 0 0 0 0 0 0 0 0 0 0 0 0

E 0 0 0 0 0 1 0 0 0 0 0 0 0 0

F 0 0 0 0 1 0 0 0 0 0 0 0 0 1

G 0 0 0 0 0 0 0 0 0 0 0 0 0 0

H 0 2 0 0 0 0 0 0 0 1 0 0 0 0

I 0 0 0 1 0 0 0 1 0 1 0 0 0 0

K 0 0 0 0 0 0 0 0 0 0 0 1 0 0

L 0 0 0 0 0 0 0 0 1 0 0 0 0 0

M 0 0 0 0 0 0 0 1 0 0 1 0 0 0

N 0 0 0 0 0 0 1 0 0 0 0 1 0 0

O 0 0 0 0 0 0 0 0 0 0 0 0 1 0

Link

AA

LL MMNN

GG

OO

FFEE

CCBB

HHDD

II KK


What is Intelligence Know the best source Google's technology uses the collective

intelligence of the web to determine a page's importance1

There is no human involvement or manipulation of results1

'The ultimate search engine would be smart; it would understand everything in the world,' says Page.2

[1] http://www.google.com/corporate/tech.html[2] http://www.aaai.org/AITopics/assets/AIalerts/alert.12.18.02.html


Why is Google special


Domainnames of the top 5oo Yahoo Go Goo Gooooal Cool Room Moon Wanadoo Football Book Cartoon OO Objekt Oriented

Goodday tool School Choose Look Kategorie Gold Gov Pogo Bingo God Google

http://www.alexa.com/site/ds/top_500


The largest Engine The computational power of Google: a cluster with

100.000 nodes using 6 PetaByte harddisc storage (The largest computational capacity on earth)*

Google stores „every“ document of the web, more than 30 TeraByte

4.285.199.774 documents, why? 232=4.294.967.296 limit for longinteger!

(2005: changed) >1.000 querys per second First time in history, a company name is used for a global

verb: „googeln“

Footnote: a PetaByte = 250 Byte = 1024 TeraByte* John Markoff in der NYT vom 13. April 2003


Research Incredible fast

growth of R&D expense!

More than 100 Ph.D.´s work at Google, „industry's most unorthodox portfolio of human capital “*

*NYT, June 6, 2004

Research


Zeitgeist Google knows the

trends Displayed by

county Displayed by topics Statistical value

worth 200.000.000 querrys a day

www.google.com/press/zeitgeist.html


Google News A robot reads the newspaper and writes

Google News sorts using 2400 sources top news within different areas

Search within the news is available

Best, just behind Washington Post, online journalismus EPpy Award


The Robot is not perfect

Car driving, car race and a injury


A small Difference

Technik und Techno


Google Ads

Advertisement made simple Everybody can place an

advertisement related to any word Pay per click Words with a low clickrate are

stopped More often clicked ads are higher

ranked


Good Ads bad Ads This ad was clicked

by 1,2% of the user!

And that one received 50% more! (1,9%)

Ads optimization for less money than a pizza


Another example

Which ad sourced more visitors?

0,7% 2,8% The data are highly

significant, reason: 3000 clicks were counted


Google Adsense The easy way to

make money Pay per click Return up to XX€1 per

thousand visitors Presentation of ads

depends on content Significant higher

click through rate than classic banner ads

[1]Google does not allow content-partners to present there income


Und so lautet der Beschluß

Daß die Maschine

etwas lernen mußAnd the decision is:

the machine has to learn something


Learning

Learning means: give results and get better

WWW Search EngineUser


Search Engine Version 0.1

Keywords before 1995 FIZ Karlsruhe Patent search

ContentSearchengine

User

Bibliothekar


Search Engine I. Generation

Full text search ~ 1995 Lycos Altavista

WWWSearchEngine

User


Search Engine II. Generation

New Algorithm Link structur Text cluster


Pre-processing


Search Engine III. Generation

Text understanding feedback Neuronal Algorithm


IQreadunderstand


Nutzer

Nutzeruser

Forces

Search Engine user Web pagegives link use link

unfaitfulSearchEngine

Optimized

happy

gives link

use link

Not satisfied

returns

SEO Search Engine

Optinizer

NewSearch Engine

improve algorithm

Receives advertisement

money


Search System

Cycle

improve algorithm

Smart user

Receiveadvertisement

money

Contentdistributer


Querrys Is the whole world represented

within the WWW? All documents are a human view

of the world A lot of documents are incomplete

or only copies It is hard to validate the content

by context But there is no other huge

digital source of knowledge

WWW


Information within the Internet Multilingual content

Few languages are relevant, 50% is english

Data are highly redundant Advantage if inconsistent

Multimedia data (Images, movies) needs complex analysis

Image-text relation Allows system to learn from image


How to read

Simple reading Problem: many pages use

complex, inkoherent structure (Table!)

Problem: fast changing content

Datebase should use the link structure of the WWW

read


Understand the World? Does the system need

background knowledge? Can the system learn by

user habits? Is it neccessary to

understand the data structure?

Which algorithm is efficent for learning?

understand


What is Intelligence

Knowledge for successful actions

Knowledge processing Best knowledge usage Knowledge expand by

additional information New knowledge

production

IQ


Future

Hard to predict


The Google Wall

Supplier Customer

Media

Advertisementtunnel

GoogleAdsense

Adwords


Efficient Markets

Interface to Information

Supplier Customer

Optimized contribution


Strategic Risic What happens if: Google stopps?

Hacker attack Physical attack

Some countries receive manipulation Censorship Results interchanged Change of Ownership (e.g. Microsoft)

Darkness within the Internet!


The Stone Age

An AgeIf there is a special matter between man

and material

Information age: there is a special system between man and information, how should we call this epoche?


Google goes Public

DON’T BE EVIL Don’t be evil. We believe strongly that in the long term, we will be better served—as shareholders and in all other ways—by a company that does good things for the world even if we forgo some short term gains. This is an important aspect of our culture and is broadly shared within the company.

Risks Related to Our Business and Industry We face significant competition from Microsoft and Yahoo.

2.718.281.828 Shares, price 0,01 $


How much is Google worth

Aproximation (short term) 200.000.000 Search results a day Value per result 5ct Anual return $ 3,6 G

Maximum value (long term) 500 Mio. Google user Save 5 min a day = $ 1,0 Anual return = $ 182,5 G

+++ current value +++ about 80 G$ +++ stock


Additional Reading This Lecture: heindl.de/google

google.com/about.html google.com/ads google.com/adsense labs.google.com labs.google.com/papers.html google.indicateur.com searchenginewatch.com

TThhee EEnndd

The Google Age The Path: Search Engines goto Artifical Intelligence Professor Dr. Eduard Heindl.

Documents

google specialwhy

google specialdomainnames

google agethe path

writesgoogle news sorts

philosophyperfect search

cofounder larry page

largest computational

ultimate search engine