Link analysis (6.7.2011)
Formal Metadata
Title 
Link analysis (6.7.2011)

Title of Series  
Part Number 
12

Number of Parts 
13

Author 

Contributors 

License 
CC Attribution  NonCommercial 3.0 Germany:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and noncommercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor. 
Identifiers 

Publisher 

Release Date 
2011

Language 
English

Producer 

Production Year 
2011

Production Place 
Braunschweig

Content Metadata
Subject Area  
Abstract 
This lecture provides an introduction to the fields of information retrieval and web search. We will discuss how relevant information can be found in very large and mostly unstructured data collections; this is particularly interesting in cases where users cannot provide a clear formulation of their current information need. Web search engines like Google are a typical application of the techniques covered by this course.

00:00
Presentation of a group
Group action
Randomization
INTEGRAL
Multiplication sign
Execution unit
Combinational logic
Insertion loss
Special unitary group
Permutation
Neuroinformatik
Information retrieval
Mathematics
Web service
Military operation
World Wide Web Consortium
Link (knot theory)
Moment (mathematics)
Sound effect
Web crawler
Measurement
Shooting method
Process (computing)
Phase transition
Ranking
Arithmetic progression
Web page
Asynchronous Transfer Mode
Link (knot theory)
Patch (Unix)
Connectivity (graph theory)
Similarity (geometry)
Online help
Mathematical analysis
Content (media)
Event horizon
Scalability
Revision control
Goodness of fit
Operator (mathematics)
Queue (abstract data type)
Data structure
Compilation album
World Wide Web Consortium
Addition
Distribution (mathematics)
Scaling (geometry)
Content (media)
Planning
Division (mathematics)
Grass (card game)
Subject indexing
Film editing
Computer animation
Search engine (computing)
Calculation
Video game
Local ring
10:34
Web page
Group action
Link (knot theory)
Algorithm
Length
Direction (geometry)
Authoring system
Mathematical analysis
Icosahedron
Morley's categoricity theorem
Time domain
Information retrieval
Exploratory data analysis
Different (Kate Ryan album)
Query language
Energy level
Ranking
Physical law
Lie group
Data structure
World Wide Web Consortium
Domain name
World Wide Web Consortium
Algorithm
Link (knot theory)
Artificial neural network
Web page
Interactive television
Computer network
Menu (computing)
Binary file
Word
Voting
Process (computing)
Computer animation
Software
Commitment scheme
Information retrieval
Case modding
Website
Video game
Game theory
Data structure
Wide area network
15:05
System call
Confidence interval
Strut
File format
Hyperlink
Design by contract
Mereology
Special unitary group
Neuroinformatik
Data model
Sign (mathematics)
Different (Kate Ryan album)
Semiconductor memory
Query language
Collaborationism
Link (knot theory)
Theory of relativity
Web page
Interior (topology)
Sound effect
Price index
10 (number)
Type theory
Gaussian elimination
Computer science
Website
Endliche Modelltheorie
Quicksort
Asymmetry
Web page
Observational study
Link (knot theory)
Software developer
Algorithm
Average
Graph (mathematics)
Adjacency matrix
Maize
Quicksort
Directed graph
World Wide Web Consortium
Standard deviation
Matching (graph theory)
Computer network
System call
Voting
Computer animation
Software
Personal digital assistant
Universe (mathematics)
Vertex (graph theory)
Video game
Family
Address space
20:03
Point (geometry)
Classical physics
Web page
Link (knot theory)
Graph (mathematics)
Multiplication sign
View (database)
Control flow
Heat transfer
Mereology
Special unitary group
Field (computer science)
Data model
Order (biology)
Sign (mathematics)
Degree (graph theory)
Different (Kate Ryan album)
Natural number
Pressure volume diagram
Data structure
Recursion
World Wide Web Consortium
Area
World Wide Web Consortium
Focus (optics)
Graph (mathematics)
Artificial neural network
Classical physics
Content (media)
Counting
Computer network
Subgraph
Price index
Flow separation
Demoscene
Connected space
Degree (graph theory)
Computer animation
Software
Function (mathematics)
Universe (mathematics)
Website
Endliche Modelltheorie
Quicksort
Whiteboard
Recursion
28:13
Point (geometry)
Web page
Link (knot theory)
Constraint (mathematics)
Multiplication sign
Characteristic polynomial
Disk readandwrite head
Equivalence relation
Total S.A.
Special unitary group
Neuroinformatik
Usability
Number
Data model
Mathematics
Vector space
Vertex (graph theory)
Endliche Modelltheorie
Summierbarkeit
Physical system
Condition number
World Wide Web Consortium
Mobile app
Graph (mathematics)
Real number
Structural load
Graph (mathematics)
Physical law
Sound effect
Line (geometry)
Symbol table
Degree (graph theory)
Number
Computer animation
Universe (mathematics)
IRIST
Website
Condition number
Right angle
Task (computing)
Recursion
35:22
Information management
Graph (mathematics)
Multiplication sign
1 (number)
Set (mathematics)
Insertion loss
Mereology
Mathematics
CAN bus
Roundness (object)
Uniformer Raum
Different (Kate Ryan album)
Set (mathematics)
Moving average
Circle
Mathematician
Physical system
Area
World Wide Web Consortium
Collaborationism
Link (knot theory)
Structural load
Maxima and minima
Distance
Buffer solution
Order (biology)
Computer science
Website
Whiteboard
Mathematician
Flux
Web page
Point (geometry)
Trail
Link (knot theory)
Collaborationism
Maxima and minima
Distance
Number
Centralizer and normalizer
Goodness of fit
Radius
Condition number
Graph (mathematics)
Graph (mathematics)
Cartesian coordinate system
Number
Radius
Computer animation
Software
Personal digital assistant
Gravitation
43:52
Point (geometry)
Existence
Link (knot theory)
Graph (mathematics)
Connectivity (graph theory)
Mereology
Special unitary group
Number
Measurement
Centralizer and normalizer
Spherical cap
Dedekind cut
Telecommunication
Computer configuration
Term (mathematics)
Adjacency matrix
Vertex (graph theory)
Tunis
World Wide Web Consortium
Focus (optics)
Dialect
Observational study
Link (knot theory)
Graph (mathematics)
Theory of relativity
Block (periodic table)
Graph (mathematics)
Computer network
Term (mathematics)
Measurement
Componentbased software engineering
Number
Arithmetic mean
Computer animation
Personal digital assistant
Right angle
Table (information)
Matrix (mathematics)
47:16
Point (geometry)
Cluster sampling
Algorithm
Scaling (geometry)
Multiplication sign
Gene cluster
Singular value decomposition
Distance
Mereology
Special unitary group
Computer icon
Number
Measurement
Sign (mathematics)
Hypermedia
Different (Kate Ryan album)
Matrix (mathematics)
Ideal (ethics)
Spacetime
MiniDisc
Pairwise comparison
Area
Link (knot theory)
Scaling (geometry)
Theory of relativity
Cellular automaton
Closed set
Point (geometry)
Computer simulation
Similarity (geometry)
Plane (geometry)
Mathematics
Subject indexing
Process (computing)
Computer animation
Telecommunication
Personal area network
Euklidischer Raum
Matrix (mathematics)
Resultant
Spacetime
51:10
Building
Group action
Context awareness
Multiplication sign
Execution unit
Hyperlink
Special unitary group
Area
CAN bus
Roundness (object)
Query language
Electronic visual display
Descriptive statistics
Metropolitan area network
World Wide Web Consortium
Link (knot theory)
Hyperlink
Measurement
Type theory
Uniform resource name
Website
Quantum
Reading (process)
Spacetime
Web page
Point (geometry)
Link (knot theory)
Mathematical analysis
Content (media)
Distance
Event horizon
Frequency
Voting
Term (mathematics)
Representation (politics)
Gamma function
World Wide Web Consortium
WebGraph
Classical physics
Physical law
Computer network
Symbol table
Word
Hypermedia
Event horizon
Computer animation
Software
Personal digital assistant
Revision control
Video game
Object (grammar)
Cuboid
YouTube
Renewal theory
56:32
Web page
Point (geometry)
Link (knot theory)
Algorithm
Home page
Directory service
Mathematical analysis
Special unitary group
Food energy
Neuroinformatik
Number
Twitter
Revision control
Software
Utility software
Maize
Data structure
World Wide Web Consortium
World Wide Web Consortium
Algorithm
Link (knot theory)
Structural load
Web page
Electronic mailing list
Computer
Coma Berenices
Total S.A.
Term (mathematics)
Computer animation
Search engine (computing)
Web service
Computer hardware
Mixed reality
Heuristic
Row (database)
1:01:06
State observer
Multiplication sign
Icosahedron
Special unitary group
Total S.A.
Arm
Measurement
Independence (probability theory)
Pointer (computer programming)
Coefficient of determination
Web service
Uniformer Raum
Forschungszentrum Rossendorf
Query language
LipschitzStetigkeit
World Wide Web Consortium
Algorithm
Web page
Interior (topology)
Computer
Measurement
Hand fan
Type theory
Macro (computer science)
Energy level
Clef
Navigation
Row (database)
Web page
Software engineering
Frame problem
Page, Larry
Link (knot theory)
Control flow
Electronic mailing list
Event horizon
Theory
Term (mathematics)
Googol
Motion blur
Ranking
Maize
Summierbarkeit
World Wide Web Consortium
Newton's law of universal gravitation
Information
Server (computing)
Content (media)
Line (geometry)
Computer animation
Software
Search engine (computing)
Hybrid computer
Physicist
Information retrieval
Universe (mathematics)
Local ring
1:05:57
Meta element
User interface
Structural load
Multiplication sign
Direction (geometry)
Zoom lens
Set (mathematics)
Semantic Web
Mereology
Special unitary group
Formal language
Facebook
Roundness (object)
Hypermedia
Different (Kate Ryan album)
Metropolitan area network
World Wide Web Consortium
Algorithm
Link (knot theory)
Web page
Moment (mathematics)
Open source
Staff (military)
Bit
Formal language
Demoscene
Arithmetic mean
Process (computing)
Ring (mathematics)
Phase transition
Interface (computing)
Natural number
Right angle
Ranking
Data structure
Resultant
Sinc function
Web page
Server (computing)
Link (knot theory)
Observational study
Online help
Event horizon
Scalability
Googol
Energy level
Ranking
Data structure
World Wide Web Consortium
Focus (optics)
Scaling (geometry)
Information
Key (cryptography)
Interface (computing)
Projective plane
Independence (probability theory)
Subject indexing
Computer animation
Search engine (computing)
Abfrageverarbeitung
Musical ensemble
Family
1:13:29
Web page
Point (geometry)
Multiplication sign
Patch (Unix)
Control flow
Mathematical analysis
Icosahedron
Special unitary group
Number
Neuroinformatik
Measurement
Independence (probability theory)
Revision control
Data model
Sign (mathematics)
Goodness of fit
Different (Kate Ryan album)
Query language
Ranking
Endliche Modelltheorie
World Wide Web Consortium
World Wide Web Consortium
Algorithm
Multiplication
Link (knot theory)
Graph (mathematics)
Web page
WebGraph
Basis <Mathematik>
System call
Measurement
Number
Computer animation
Integrated development environment
Personal digital assistant
Website
Data structure
Resultant
1:16:48
Randomization
State of matter
Equals sign
Multiplication sign
Chaos (cosmogony)
Bit rate
Special unitary group
Data model
Type theory
Bit rate
Endliche Modelltheorie
Area
Covering space
World Wide Web Consortium
Link (knot theory)
Web page
Hyperlink
Term (mathematics)
Control flow
Bookmark (World Wide Web)
Degree (graph theory)
Type theory
Website
Quicksort
Navigation
Resultant
Point (geometry)
Web page
Ocean current
Trail
Random number
Link (knot theory)
Gauge theory
Web browser
Event horizon
Number
Revision control
Degree (graph theory)
Term (mathematics)
Random walk
Ranking
Data structure
World Wide Web Consortium
Default (computer science)
Graph (mathematics)
Planning
Computer animation
Revision control
Video game
Musical ensemble
1:24:08
Web page
Axiom of choice
Point (geometry)
Suite (music)
Random number
Group action
Randomization
State of matter
Multiplication sign
Tape drive
1 (number)
Disk readandwrite head
Special unitary group
Number
Data model
Goodness of fit
Hypermedia
Set (mathematics)
Adjacency matrix
Lie group
Address space
World Wide Web Consortium
Execution unit
Web page
Type theory
Computer animation
Personal digital assistant
Game theory
Musical ensemble
Freeware
Metric system
Matrix (mathematics)
1:31:41
Web page
Dataflow
Random number
Group action
Simulation
State of matter
Multiplication sign
Decision theory
Web page
State of matter
Coprocessor
Data model
Computer animation
Video game
Maize
Clef
Metric system
Simulation
Matrix (mathematics)
Capability Maturity Model
World Wide Web Consortium
1:34:04
Point (geometry)
Dataflow
Random number
Group action
State of matter
Multiplication sign
Special unitary group
Data model
Causality
Vector space
Row (database)
Vertex (graph theory)
Maize
Summierbarkeit
World Wide Web Consortium
Point (geometry)
State of matter
Theory
Limit (category theory)
Number
Process (computing)
Stochastic
Algebra
Computer animation
Vector space
Software
Personal digital assistant
Right angle
Simulation
Matrix (mathematics)
1:37:07
Point (geometry)
Discrete group
Ocean current
Axiom of choice
Group action
Run time (program lifecycle phase)
State of matter
Multiplication sign
Virtual machine
Set (mathematics)
Special unitary group
Chain
Mathematics
Vector graphics
Vector space
Finitestate machine
Codec
Row (database)
Condition number
World Wide Web Consortium
Distribution (mathematics)
Matching (graph theory)
Point (geometry)
Discrete group
State of matter
Length
Markov chain
Stochastic
Computer animation
Chain
Matrix (mathematics)
Electric current
Spacetime
1:43:08
Point (geometry)
Web page
Group action
Eigenvalues and eigenvectors
State of matter
Multiplication sign
Letterpress printing
Set (mathematics)
Special unitary group
Chain
Pi
Mathematics
Vector space
Matrix (mathematics)
Multiplication
Condition number
World Wide Web Consortium
Distribution (mathematics)
Matching (graph theory)
Eigenvalues and eigenvectors
Weight
Physical law
State of matter
Markov chain
Category of being
Arithmetic mean
Computer animation
Chain
Video game
Endliche Modelltheorie
Right angle
Musical ensemble
Metric system
Matrix (mathematics)
Resultant
1:49:29
Point (geometry)
Random number
Group action
Eigenvalues and eigenvectors
State of matter
Workstation <Musikinstrument>
Mereology
Number
Data model
Vector space
Negative number
Uniqueness quantification
Ranking
Algebra
Position operator
World Wide Web Consortium
Algorithm
Distribution (mathematics)
Eigenvalues and eigenvectors
Web page
Markov chain
Computer animation
Network topology
Theorem
Game theory
Metric system
Matrix (mathematics)
Row (database)
1:53:21
Wechselseitige Information
State of matter
Multiplication sign
Programmable readonly memory
Boom (sailing)
Database
Special unitary group
Arm
Thomas Kuhn
Expected value
Query language
Moving average
Physical law
LipschitzStetigkeit
Drum memory
Arc (geometry)
Library (computing)
Metropolitan area network
World Wide Web Consortium
Software bug
Web page
Interior (topology)
Topological vector space
Shared memory
Physicalism
Menu (computing)
Measurement
Software development kit
Digital photography
Multiagent system
Uniform resource name
Computer cluster
Order (biology)
Convex hull
Right angle
Quicksort
Ranking
Resultant
Wide area network
Web page
Point (geometry)
Trail
Asynchronous Transfer Mode
Personal identification number
MUD
Page, Larry
Maxima and minima
3 (number)
Electronic mailing list
Emulation
Wave packet
Goodness of fit
Latent heat
Term (mathematics)
Googol
Ranking
Elearning
Data structure
Gamma function
Quantum
Conditionalaccess module
Page, Larry
World Wide Web Consortium
Newton's law of universal gravitation
Information management
Raw image format
JustinTimeCompiler
Electronic data interchange
Dataflow
Twin prime
Uniqueness quantification
Content (media)
Coma Berenices
Exclusive or
CAN bus
Feature structure
Computer animation
Search engine (computing)
Network topology
Universe (mathematics)
Optics
1:58:54
Web page
Point (geometry)
Randomization
Eigenvalues and eigenvectors
Link (knot theory)
State of matter
Maxima and minima
Electronic mailing list
Mereology
Special unitary group
Number
Database normalization
Mathematics
Iteration
Vector space
Set (mathematics)
Ranking
Maize
Endliche Modelltheorie
Addressing mode
Summierbarkeit
Position operator
World Wide Web Consortium
MIDI
Execution unit
Structural load
WebGraph
Web page
Length
Power (physics)
Computer animation
Vector space
Personal digital assistant
Phase transition
Website
Iteration
Matrix (mathematics)
Resultant
2:03:33
Web page
Group action
Functional (mathematics)
Eigenvalues and eigenvectors
Algorithm
Materialization (paranormal)
Control flow
Valuation (algebra)
Mereology
Special unitary group
Field (computer science)
Number
Neuroinformatik
Iteration
Different (Kate Ryan album)
Vector space
Personal digital assistant
Energy level
Ranking
Vertex (graph theory)
World Wide Web Consortium
Area
World Wide Web Consortium
Information management
Stapeldatei
Algorithm
Graph (mathematics)
Eigenvalues and eigenvectors
Web page
Lemma (mathematics)
WebGraph
Physical law
Content (media)
Bit
System call
Power (physics)
Number
Wave
Computer animation
Website
Iteration
Right angle
Ranking
Metric system
Perfect group
2:09:08
Euclidean vector
Multiplication sign
Workstation <Musikinstrument>
Mereology
Special unitary group
Neuroinformatik
Mathematics
Different (Kate Ryan album)
Query language
Search engine (computing)
Endliche Modelltheorie
Area
World Wide Web Consortium
Algorithm
Link (knot theory)
View (database)
Web page
Sound effect
Physicalism
Mereology
Distance
Degree (graph theory)
Type theory
Vector space
Different (Kate Ryan album)
Ranking
Resultant
Web page
Point (geometry)
Random number
Link (knot theory)
Algorithm
Variety (linguistics)
Connectivity (graph theory)
Control flow
Twitter
Sound effect
Frequency
Googol
Energy level
Ranking
Data structure
Metropolitan area network
World Wide Web Consortium
Focus (optics)
Physical law
System call
Componentbased software engineering
Computer animation
Search engine (computing)
Physicist
Form (programming)
Address space
2:17:10
Web page
Multiplication sign
Decision theory
3 (number)
Maxima and minima
Set (mathematics)
Special unitary group
Arm
Theory
Emulation
Neuroinformatik
Frequency
Insertion loss
Arithmetic mean
Different (Kate Ryan album)
Term (mathematics)
Ring (mathematics)
Query language
Ranking
No free lunch in search and optimization
Summierbarkeit
Pairwise comparison
Newton's law of universal gravitation
World Wide Web Consortium
Metropolitan area network
Pairwise comparison
Link (knot theory)
Closed set
Web page
Interior (topology)
Measurement
CAN bus
Computer animation
Convex hull
Right angle
Musical ensemble
Cycle (graph theory)
Whiteboard
Game theory
Ranking
Resultant
2:20:03
Web page
Link (knot theory)
Multiplication sign
Execution unit
Insertion loss
Special unitary group
Computer icon
Facebook
Different (Kate Ryan album)
Number theory
Googol
Oval
Ranking
Data structure
Extension (kinesiology)
World Wide Web Consortium
Link (knot theory)
Web page
Coma Berenices
Bit
Hyperlink
Process (computing)
Voting
Computer animation
Website
Family
Data structure
2:22:52
Googol
Web page
Ranking
World Wide Web Consortium
00:00
on way to unthinkable patients and and 1 that would unexpected provides a back so Humula some types after and will continue to pursue lecturers homage measure dream search engines the talk about patch rank and to the famous methods to execute who exploding structure present and Groff's so to find what page is a really important and were not so before account event
00:34
now as low as discussed the homework of some of the excess as the song goes well where the and methods basic moment of operation he the her mind a new need to of such a rental squad from the collector in confined to analyze and and give it to you next of the new additions so that is a queue you of quality this summer pugilist as a good way chromatophore should to cross by now D with a Gambale set up lack of by it means that his so and what do need to rest no so that are standout surge to dip in and out of the she pages should look like so that we should be kept robustness will use the money to time accounts of the latest on the Web such is so most people look at you and you can do not care about and the chamber must be stressed just kind of a shame really wanted implementing the absurd restaurant computed will because a bar its the base some stunned not so busy group for gross be prepared for anything from a bad additional features such because of a groyne celebs quite lapser at PM PM at the end of friends as a major issue in Scalability distribution because we get to make this kind of calculation last week that many of you want to catch every page 1 c unit of shoes on pages a 2nd service cuts for year and and the and Distributed compilers Scalability picture which is not a tribute ask them to some of credit works well quite nice is the only but doing this on a large scale are and a good integration with NexCen and or combine of components is very very difficult survive and the amount of ideas from a happy to that by the way across the border crossing at to cheque for deprecate year highs and how do they do it the at all and of need look at it there yen why should yet up a as the should yesterday stem CEBR edgy good but D a day smoke on picketed other to prop up the Yemen we need to have a problem with that of the campaign and have been group local TV means the before almonds no work and Distributed entry into the world of stocks and most of the most for a popular ride on the back of a so due to cheque for the picket continent why on to the bench for this and how does showing where cannot help you she with the 1st version and there has a to cheque from public a continent or it sometimes people just make the same kind accessible owned by up different 2 and of merger determined that using just 0 you're not rise your automatic so you division to cheque the continent eyed so that this is really important and where she can be used to do this by titled work and what problem doesn't souls put group busker are looked for a while that he wrote the the put it over the wall the end and the and of of BBC tried to just visited the Dome long lasting mathematical because was for plant ephemera Bialek like some of them were prepared trouser disco questions pretty or so the and in the last will usually which unit of extracts you respond to the document the Williams or move his and then you try to compute overlap agreement textually set to documents typographers and and this is where the similarity of content for business a new Pench you care compared to loss existing pages in your index progressive and would goal nomadic its job during thousands of compose which time pages company page comes in answer to shooting and you can simply of represent each document by sketch to derive by change growth and taking random permutations Phillips are Maria 100 200 and footage and pension plan to compute the minimum value of some of which are set the sketches indecomposable this not and 1 can show that 2 documents are similar if they are should of the same name are so it is probable way of fee of 200 to 200 sketches sketches of the 200 thousand tonnes of delegates that you can approximately from Jack are coefficient by just comparing the me and you can download this season where by building Tremblant starring or the sketches intent and then Computing similarity off new documental existing documents and very very effective way life is a wave of at last the grass that was chronic so that I was due to end where would be needed for a home for P grew up with D the Solar speculated that based on the idea only 4 of those links seem to be most promising almost rather than to the topic you want to index I focus credit and the 1st phase and classified we had a lot of them don't and just be happy some some nice guy last week that a beautifully does work so very interesting technique if you just want to collect or documents threatened to
10:35
some topics as interesting to you are right so
10:40
that was on the to don't and either going on with which we are in and the connexion to the length structure so all want to do today is kind of you with the structural where because every did in information retrieval was usually using the word using the textbook continent off of a page of the book your and and the mixing but with the bat we have a very big difference with respect to the collections of some of of of documents and that is a lie which between the hype and this is what we want
11:25
to look at today's so 1st social to a brief glance at the link for way and that no 1 to introduce to algorithms of that are of much we noticed that the page Frank algorithms which will remain the of build and and actually make who will about what it used to be 1 of the very for of influential of and of both like the the and the and how to explain it so
11:58
from links between document are in the in the sense of the web and the broader than you would pay for example of 1 papers of like scientific paper side of but it's rather well they've network of socially into wrecked between the people behind the with of the people creating about it because if you see something like I'm like a like the academic life you might have papers on this difficult document that you find it and then you have kind of offers of paper and they cost of something they worked together on some and thus place tight each other and by offering the same papers for of from papers that kind of commitment Graseby social interaction by for example in the in the movie domain directing and acting on so there are those that have worked together of all at the top level of game and the way it was for the kind of network and you could say that the network structure the idea that somebody called for the paper that to a new whose well then the topic that those offers are interested in must be the same because they were from the same paper the salt movies but the Act a usually does must be the same As Lavoro of this collectors they stop together in a film a movie and obviously they both data for the same job but the idea that in the technique of and and a seat in the genre of the topic of research group of some of by the looking at with who they are connected the battle for example somebody did very many films with an action stop typical act what the tell me about the 1st of all he seems to work in action movies 1 within this the way of thinking this is the way off of of the writing inflammation that the talk the difference in the way where everything and the and the and and where kind of like a linked to some of the site is the references is the vote for the site and the well off and and social networks so
15:08
far this basically applies to all area of of the case that began musicians you can talk of thought and a friend of the family because of the relations between different country's either go by by by trading contract where it is not and you know I assume that there is such a relations the drank of the relations may be different but the topic of the relation of the bulk the interest from same rose
15:45
for 12 people make phone calls for people from the effect know it all network full the network from
15:57
scientific papers veridical of new no like but and here is a study in the UK later somebody in the air introduced the type and love of life from citing this means was what does this guy work on wealth of the of the ballet at least he did 1 paper about the me its main topical something but the more that the more citations we collect the more secure a we become in predicting his favourite of of the this is at the idea that is
16:44
also in in the way of pages site each other and each citation each hyperlink is a vote of the about of confidence that the bomb the become tens of the link to page is adequate is a sign of quality so you believe what you on the and it has something to talk but the to do with what you were talking about because of a wife why should be linked to the maybe some sites about the of well linked so for example you will everybody who will so that might be linked to Willem and the fact that I'm usually it has some some some public health up indications profoundly look at the website of the Institute Web but I would find links to the 2 computer signs of computer science at the end of university officials from for above as such because the topically connect and the 2 0 2 tomorrow this sought to exploit the new 1st need to hold knife and womanising at this quite easy to do it because you have resold just the sort of stuff websites or of people carrying infections all are part of family members or offers the academic standards all white at and these are the notes of a crack and as to whether there is a high pulling citations somebody infected of whatever it may be a collaboration and the way you add a link between but and the morning behind the whole network the memory of the think you get the across consisting of edges and vertices load that is something that we can handle quite efficiently in match because we can represent described by so called Jason seem efforts were we just say of is a link goes from no wonder too 2 we know just say there is and that taste of of these 2 sides taste and if there is no bowling from 1 with no free redress say no 1st noticed that the but the magic is still the gravest direct and so you linked to with something of a site somebody and it not really custom Africa matches but of calls them may be back playing in which case it would too much of a part of the beauty of some of book in general if the asymmetric match that match and
20:07
not all the social and Ellis's of such a network has a lot of classical questions as to why not westcentral point of focus point at time which offers on a high cited for example which starts work together with many different people and this is all ways a sign of high with ball by the some cited very off and but different people that may be an indication that his work was over a high quality and she may be in a different area of many people thought the the lot you can also say talk about connected other points in your graph that well connected other points the isolated these your graph coherent in some ways is a negative for the consisteth several of subgraphs isolate with of the grid of things that you can find out by by by the book and the and also other some central points out that some points but could make the world look that act as mediated between different part of world and which accounts to be remove the the network break the thought and so this would be an example of such a point that the move it is that to suffer that nothing from just because you remove 1 1 but point and the structure of the rough is Israeli interest and has for below Pope of
22:06
research at bombed what we want to do is look at the 1st to use of the phrase the of of paid just for the sake of playing well if something highly cited something that might be linked to the quality and you will of of this site seems to be high idea that these of something more the books but seems but seems about was so it's itself out and everybody can put up a Web site and and put it on a Web so that people but if other people find the actually to take the effort of adding a link on their side to the other side then this is kind of the board of the trust sign of comfort you seem to be happy with the quality of what is written the for use seemed to find it interesting see what the ballot so this is the scene the content on the site seems to me that are in some way the and some Toby unconnected side that nobody have links to of false 50 year's on might be the reason that the new site is that it has just become available as perfect incontinent multi wonderful in all but opinion of more often than not you the new page that would sell what you do is basically a set off bomb that led that the the 1st indicator of 1st each use the indignity of the note that is any note that is pointed to very of you seems to be yes and pull PM plug seems to be the high quality of because all these guys over you take the effort of reading the and adding a link but seem to be a sign of the 1st the greedy just means that account from many point you have to assign at this stage it doesn't matter and of comedy point out of the of see the incoming linked and if we call them the can get quality criteria of of some of the prestige and the but it did actually have been quite some while the Web was not even invented yet that people investigating such networks structure in the area of the field to became a where wealth the in the green actually is not which the very good in that it has something to do with with if handing over yes but it is also something to do with the 1 at the of the people of site where they may be but the view though it that found people that both for 1 page and found people that love but different how would decide which 1 of them is more of but it up for other health could emboldened yes and yes but is it could be of the will of all but what if the thousand people pointing to 1 side all University professor at last be beauty and the other by a the other thousand people are persons just you usually Street persons that have no connexion with the public what would you say that sort of cloth he would and everybody has to leave the university professor because they know right from wrong the and this is it this is what it was sociologist I'm very early on realised we it it has a certain recursive nature so if I'd on last my boat counts of more than somebody wealth was just of the street and with no credibility so the high are my street cred the more important and this is of cost because because up if I'd have high street cred I'm must have a high in the league Eileen to some papers I'd do not only increase in degree of this page but I'd Trent's for of some of my credibility with and the high on my credibility the high of the credibility of a transfer to the more for it becomes this son of a progressive so far in
28:16
sociology the Christie's was of more of the follows you say what you have a note in the graphs any of the press teach of the note this just some some some some for the number the higher the number the more critique of moving low of a number of less than amount you go and represent the Teach for as a better exactly 1 entry for each No and the 1st each of each note should be precaution not to lose some of the 1st use of the notes linking it put it so the more people linked to them the higher the which will be the better the more Opus T just that the people off linking them up the higher the of each of the and review this is just way that some 40 can face as well you can Cambio Pagés's not very off linked to but if your you people thinking to use all very high quality you can never less beat the system and to show that more people the cost of raw of across to the new 1 university professor than the open your head of the people on the street for walk with the but that or something like that of the UK and so this is the basic ideas about of caused by a brief but if you consider the at a syntagmatic that we have in the greedy of each note and determined that if you want to press to use of the load pointing to it to consider the open note pointing book and determined degree of appointing no if the ease of the note pointing to the note pointing to pay and so on and so on so this is kind of a progressive and and what you do is take the Jason the metric and and you use you to fix point computations so you start you stop might applying the adjacent the metric such as that of the condition old rest teach is some fact as the Press seat and if that doesn't change any more than the point at which of for this country and the and the idea behind the and we will see in in a sure while but this can actually done by to pick them up and with few out what for of the of the characteristics of the the tricks of the difficult to be such that have the right of such a six point competition for what we want is to want to consider all the Fifties of sites pointing to further with the 1st I saw 1st season and at some point at the end a means with of
31:55
but saw with the symbol model of Christie's we have the chase and the methods of the graph and banned just take the Trent those match but and this is just and both my pretty something of let up and but columns and the lines of those of the exchange and we might stop with some of the siege of the sites that are kind of connected with the notes here but this is the 1st each of the 5 days of the of 65 of the beach of the Royal like the 5 and a bit of of the 1st of bombed the Eurotunnel but like this site nobody point to the fact that the home of point for some but it points to this just want so this is kind of the idea of where we say OK it has something to do with 2 pointing this site only 1 pointing to this site gnawn pointing to this site and it is true that some of that by the way up and taking just the need which is a good 1st time is a good 1st approximate some of what we want to have a hint of the up and Oregon all says about found the link coming from this site down here should probably the worst less the link coming from the though this notes because they are more perfect but several of the link coming from here for basically be worth nothing but both sides of the beaches though this page as to in the but also these pages of you only have to in the we feel but the point of these pages should be higher because more for the point of and if we take the piss teach as such as they were and will take the 1st tee to the idea that we will get no pressed Shia and and the lower the and the effect that the second lien some of the note for came from a very shot he signed from the side that is not really far were the because nobody points to the side of law which or and it would take the fact that of the 4 4 6 2 we reached the pick up so everybody and find work of the was
35:22
the basic ideas behind another different interesting idea about graphs as the notion of the centrality so how central is an old for the network it under are on the up on the on the all time area of some work or is it in the middle of the network and what can we do and we can we can investigate networks not only which suspect to some kind of Coughlin network and the Web over what ever but there is actually a rather than other this kind of the discipline in computer science and mathematics but the bell investigated and the and the people were interested in some such notions because they could be used for different application area but was the idea of central website and in the way to St from the feed that seems to be very popular and many many people point but it from a different Area so what would be a very simple way website for but exactly buffer turned a page book called the police and the and board of Wikipedia might be a very simple paid because of the loss of the folk not of different pages linked to it at the very central sent off because it really love it the Pope it with it exactly they seem to have all they seem to exhibit higher centrality than than the ones and and the order of the so far we need a couple of conditions here we will talk about distance between notes which means 2 notes on a certain that shows that distance if there is a number of links between them so that they have the power of from 1 of the other and in the case that are much of apocalyptic the for of the distance between 2 of the smallest part of cost with news that the number of between now and then you can find a kind of consider well by if the graph is connected you can say the of radius of an old given by how far is the 1st point and the network that you can reach from the UK for the of going round in circles or something of that really taking taking taking the lead in 4th to the rim of the book and then you can say about the centre of gravity is the 1 that has a minimum of radius of the for example look at this graph of the was the distance between this note and his No 2 there is no than this note also want having you could go down via but that would be the 4th cost to go appeal of Kent good at was the distance of the note here 2 because of the slow way of getting them with just 1 or and was the radius of the note this but radiances and yet again the the distance to the most distance know that if we look at it again this which this noted whom this not in whom this not and 1 this want this note into of the radio of point to fall out the point over here but the previous but with the guests were brought with that on something but it 2 3 4 effective 1 2 3 4 UK and that the only way to get to this no because that's no direct after nothing way can take shorter so I'm the most distance note is really for and thus the radio of this Load Olivia for and the idea now is taking the 1 point or the set of 1 with whom the ladies this is exactly the ones that up very welcome at the and and in the centre of the track and the for the both but it if you look
41:16
at the scientific citation graph and I say Well basically time if 1 paper sites another that comes as a link of this paper over side this paper and so we editing and in the and and the papers that have small radio are likely to be very influential Bacalso they are not only cited by a lot of different but they are cited from papers from different area the thick and some of that they can be reached from almost every area of the graph within the you and if you don't take the citation rough of collaboration with of people were to get who actually for the paper this is something that is not done in mathematics very very off called the abolition of that was by at mathematicians of American mathematicians who was amazingly influence of the work of a lot of public and he worked with a lot of people and to find out how fringing you up for the days for young you on the addorsed number is kind of the distance of any note in the network to the famous mathematicians part of on the idea is basically that if you cost of paper with somebody who bought the paper with and the Evershed number to the disputed if you call the paper for the with somebody who for people with some of paper was poet at of number of the best things that below us at his number that you can get these days is that 2 or 3 things something like that of a can of at number 1 in the mouth of the with has said of the system the chemical off papers the with them but the but some of his collaborators are still around so if you want a small Adish number of I'm you you should for the people with whom you like to a and the and County
43:53
and that some there are some other options to be fined to for public and and look at caps on which basically means that you just go through the RAF and try to find out about what is the number of links but connect the to part of the ground and you will find that of a well to disconnect these 2 graphs by would have to cut throat not of links of and but if are wanted disconnect these 2 components cutting easy because the septicemia singling or a does have to take out this work and the graphics full of this can also be on notion off centrality is a that the existence of some of the central for a collection graph because if you remove the note to collect with the idea and has been very of applied to the epidemic of espionage of something like that on the telephone the work with the 1st of clothes and a point for the focus point out that the and by writing the disease interesting to see how but
45:30
what measures coal citations a two document fights to other documents and the 2 other documents seemed to be in 1 of the nett income collection of calls the with your estate in out means that could be a paper thing about this paper he has definitely nothing to do with this paper all of that but as the table usually say OK so we could be looking at the problem of the block and the foundation was later in the last and the walls over joined by a block of and then he would have the right to between the like of the and if document cited by many documents this seems to be a strong public relations the both but only by a few document but that 1 of the cases they about like this this is set to the tune of some of the region's for some reason or the other will of that I'm in terms of the day systematic that began the link to the document by to some paper of the edge in the collection wrapped and the number of documents cause fighting the other 2 document is the entry in the chase and the magic of ghost from the case of the metric and basically of with the idea of the number of book humans fighting both the height of the number of those of the goals of and bomb the
47:22
entry of the of this day since I Matrics corresponding to be collection between the document is the so called co citation index and but it still related of not created by of the of time and you can begin to allow of things with some the proposed use of last during of finding opaque like we did with the late them and the next thing know I'm and and said called damaged scaling here but it's actually very similar to the living the thing about because of the need for the public at time but the basic idea is that you have the simulator to metric and was a simpler to metric you M the different documents and in the space the idea of much of the of the year at some and the scale of the relations between the 2 sold a 5th on anti beauty media but in Italy the at the beauty excess and firm if you have look at the at the cost of the results from the both citation you will find that those document the last that really have publicly something to do with review how to save and
49:00
point and if you do it for example for a job to go I'm based Lako citation I'm here coupled or took a million of general wanted published in 2 thousand and every point represents a Journal and his he how all the different time of the from a of both sides each other to find that there are topic area of for example is that of the social sciences at the sociology of only signs that history of the the with a geography communications so if it is all part of a similar and and the distance is much smaller than the distance for example to Bolangi biochemistry which again of very close to a the other but are further away from the social sciences at the end of the of the of the ideal of the the and that it into the the space somehow and by people of this world and then you can see looking at the cost of the ticket close to look at the of you can't find common between the paper with the number and the and distance between clusters of actually means something me I'd act icon of say that the difference between history and economy is kind of the same as that the distance between the Computer Science and Andrew bought it just because the kind of the same distance away but there is a tendency that large of the distance the more this command the area it took a puppet the cell Solheim
51:14
less of above the Web again we wrote considering the document types used in classical at up and of Costco citations and stuff like that applies to the but when reuse was basically the idea of not the not so much a co citation and and familiarity with the but rather of I'm well what words to occur in a human from Hull symbol of popular with the youth of what they are covered by the Mara this transit values of and we come to the beach document to be a self contained unit and describe it basically by of representation the single the with display smaller we could look we we opened up the access for every terms and then look 8 at of command and terms space depending on how much weight of of the of of the use of or for the collection of measures that can be used to the idea what not up but we kind of represented at the document by the works of her we did consider document was being symbol when the distance in the space of more that means they basically at the same to a used to say of similar took sentiment now with the way with something more well we do would have the describing 1 10th of the website of a we this was shingling so we could consider the seminary to the of objects annoyed with the fate of them basically it's a good idea to return all the similar pages based on what we did with with the company cheque but we also have the notion of somebody took the effort to create a link from the UK and the page creating the link is more trustworthy well the collected for more from a law and this brings us back to or from network and so why don't we apply the idea from the book and the way round and the building of the recommendations of something that makes a website more important in the morning they are all so the higher ranked the page become and of qualified left to mix it with the kind of fate because of by a can of just say Well are it doesn't remember what your read is Iwo allways return but the most popular type of why yes it very popular but it doesn't have anything to do with my period bad idea for you to consider the Quantum of both of the of the pitch never below that has to be a mixture of and was very often time dumbasses using the banque said As description because you use your point to something by saying what it is to be than the common referred to the site you use the ANC attacks or useless around in the link and and find out what seemed to be but but
55:18
saw the 1st assumption that we are making is hyperlink is a signal of qualities of a parent and that it like a way of life like recommendation and what you do with it on a few things well from what is the link about but obviously if the bulk of advertisement of the by the company that was prepared for those who global book the before event and the and just looking at this and context tell you actually launch about the site the points to won Amin may also be the case just a week he which is that year's we actually need to tell me anything about their the quantity of of the of the pages of the book so you can and do kind of of take this around the text of the link also and how they could do it because of the this
56:34
assumption to is kind of the anchor text of link although strongly text describes the top 3 in the version of the device link owned by BM from the book by be amid affect of both computer services of a local for funding of before you with a lot of blood but the but with some of the verdict the text to be mix of the page but the linked to the use of and the and and this year it link to from the record of link to fund and if we actually look at Ibn OPEC work computer it is not meant that doesn't seem to be very interesting for by to have computer because they have smog and energy and they have dry shrinks things like talks load of knowledge Tring's the utility bills by so it's all about reuniting of the but since but expect but anybody looking at the idea mweb but know that it's about the team that the computer a can of for not even mentioned but it greenlighting but they can't talk of what my edgy and the next thing you try things that you it was kind of the idea sometimes the and the text of actually more than the actual page the number 2 assumptions that we want to know what to make for this today's lecture
58:25
so we will say Fault basically we will analyse linked and we know the assumption of the heuristics but don't have full ways to through the year with lot and I think of all the great I'm all of I'm might be linked to some page because it's a very bad example something from this is that total spam page camp and you thought for half dozen usually the if you take the of the interviews and you want to say I'm so far and that the way we have to be on point to be on the safe side we have to be a where that this is the tourist but Touristik do without a lot of this has also been in show from and to the best the 1st assumption links are kind of quality of falling salt recommendations as full 2 algorithms are especially the 1st on the picture and the of has found that for laid the foundations for an empire and that is what we been beyond the wildest dreams of people creating because it was these 2 guys over the and there smiling so much because they were rescued in the stand for the work of the work and the and came up with a trend that will be used for the new Web search engines of the of the at this point and the mock across the UK at a list of all the big on to the land of the long like some of the many but and you don't even know the names of any more and in the middle of it almost 30 lined up full of 1 of big Palmer's away such and but these 2 guys came with the leader of the what if just about the club of also Fault links structure of and if we mix it some of it would be landed and and an a in tribute ranking based on HBO and the total quantity of what Web search will be better and that proved to be true so
1:01:06
found was immediately kind of accepted by the people who will fill in for more stop the basically in 19th 69 by later patients have sprint I'm and that the idea is to have few the independent measures of each for each work with salt and last if you to a Web search based on the theory that it can also take into account the page ranked quality of the data for the final and the 2nd over the worst of basically developed around the same time at the idea Mohammed and some by dog time whose at physicist that I'm and made its way into a into a Computer Science and and contributed by quite a lot of time in the really amazing to see his record of what he invented I'm from Hutapea the networks navigability and and Hutapea a network of up to a year its algorithms and some his idea was basically a that there are 2 types of of of of papers that I'm everywhere research has can be a hybrid for an offer to the house designed to because things like to with the full it not pointing to fix and all off the pages actually delivering the inflammation and this is far more inspired by the social networks that it is and than the patrons because I'm Adenauer and in the end the your friendless enough you might have some people but no everybody of the world and if your problems but slightly on specific your eye somebody who knows about my secret and you don't know the exact person you ask your wellconnected friend do you go somebody Reviews with them on the other hand if you know somebody with the perfect all party on your topic that you interested he will directly with for ball and in this sense every 5 could be kind of them high school and of 40 for the of 40 up rather than in terms of content the hot off basically red event in collecting links to of all but
1:03:55
and this is made up of 4 break because we go into the history of Web search for year so last week we talked about these 3 off the weapons sense of combat timber slipped and doing research in Geneva Holland Nokia Physics and have now become through a Web search and find information in the wake of the duration before the 93 was quite some well actually the memorandum large those you need to have a hand in the search engines are and it was growing up time mooted tenderness himself maintained this morning so so we created some kind of a victim at age Jimmy Page fans are the best time this was modest looked like the best of the new post of EU institutions having some observers set up and down the stage is set to available so you use you know to this day to see if oyster interest only because the continent has updated since late night to so around this time he realised that the idea people need to know more Scalable way of finding the information the web of the Senate realised by the local services Germany so we have shops of storey about that in some weeks ago so I on the idea of the impact and of link everything on the line large map of Germany where concede which universities mostly provide Web so that you can make of it and then you go up the of the Sex and the and the basic was so young and a soon realised that the via the to many mezzos to present on to assume that and around this time of the
1:06:00
1st stage in zooms came up so that 1st taste of the Web search where Kelly these Tesco engines like a sudden of this exciting Tomahawk part of chief and somewhere out the physical a demented information fever algorithms mapping published the music the millions in the Seventies and Eighties and design somewhere across the US and the collect information from the Web started it in a big way index and just round information few technologies on it so we looked rather where up and now the focus of research was on how to get through these techniques to large document collection as can find them on the Web another updated the collections frequently saw the but could credits in August so that quality the West and you do what it wants from from my or times and nobody really catapult that of the US media problem of collecting all the information you need to sell and 1990 6 3 all that are said to be in late page are worked at stand out specialist in this and the way the project about where were which it wants to be a large scale is that in the end so something like this may be the moment with the focus on Scalability and some are due in the next few had this idea of using the structure and the way for measuring teams and finding out but they which pages of that and other pages independent of the pages continent and serve and the device ranking system that took the lives of thousands amazement account page continents and league structure so and get that found that the search engine has been quite successful so that they are doing too dangerous to about them must as agreed but stiff decided to found the company had to 2 is that the new companies so that was the beginning of goodwill 98 so the start with some help from the Stanford University and implemented a off a lot of work not do that at the time and the and people noticed that Google said and that worked with a bed at the the stuff up here because they use the structure and of his said the only other companies realised what would prove to be music could less of most people in the West or even of used to before the searches of these companies are broke a basic this is not the duty its staff half to beat them the United and the key for that was that they trade idea using link information for breaking the results of today's search engines businesses and its more and so many Microsoft and to beat competitors he to to go with a few Microsoft the of the car business and goes is kind of big player
1:09:22
so I'm since Introduction approach rank on the Web search hasn't changed that much so that have been some kind of a loose making the user interfaces integrating the information be finding the ranking for the lot which is a bit trade secretive who will be doing on this kind of stuff but the has been known to take ideas since and so the question is what could be the next big thing in which such a some ideas something could be classed ring that really works we already have some of their most of some weeks ago that usually doesn't worked well and the nice to have it so event agreed that is a big in some way and the with such an yet due to meet at the of the company the through of the what I'm up but it doesn't have now periscope interesting mantra language Processing can enter your questions in a way it up the way you and people really off questions and not using dusky and the search engine is an able to understand what you really looking for a good bet for a few if information need has a high deep as it is is a high level of detail and the need takes place in some way with a new duty to search engine that is able to understand you this could be a good way yet where 1 that has been the big idea of the semantic wept also are proposed by 2 bonelessly rammed in which she webpages ball constructed information and the the search and his reasoning about the Web so for some but at the moment and while this kind of connexions to the public but entities and the way that while could exploited of buys appointing Search cruise but the set Rio to you that it really doesn't work because it's too complicated the way it is designed by many they are some ways to make it to make it easier to make a bet on business that could be a nice things for example if it really knows that on no came in for Computer Science stuff as long as climate work that would be good to focus my research my mind the said and the results and into this directions it is done and and and some some way actually could be improved may be that in some efforts to to create Olmsteads search engines with scenes the are projected on that has been are discontinuous some some use ago where idea was let the community engine and and everyone can truly the ranking family lap and with the wisdom of the crowd we can create best search and there was so actually also this didn't work and meant as to become the next McGinty study of combining the results of different search engines and then to the dedicated the results quality quality of the also but that does not work for the good right now so Federated so it also combining different different search engines and make their way many of the key is to use a interfaces using nannygate through the where pre of this Caddick get access just 3 ideas are is also could be a good be approach to to to with that of search results and many something out nobody knows that many many ideas around half that could be improved by for most of them nobody really knows what that will see them going on day all 1 of the trick all of the key problems to be soft so usually currently Web search means are improving the upsurge means all you have make many making some slightly loosen here tweaking the use and the phase of integrating new integrating new information like to be tough on the Facebook social networks in some way that was published on to David nobody knows a Web search will look and feel your best but a at the time to
1:13:29
make a five minute break mountain continue with a frank and its in Taylor sell of
1:13:37
lectures so what would be
1:13:44
discussing and the rest of the lecture will be to algorithms out the 1st the page of Britain's found and by age of 4 in the 2nd was that it was in fibre and
1:13:58
that the idea of a page algorithm is I'm how do you get this measure of press teach that each side some is assigned because of the way over of a large number of new you have the Web problem for every page need the pastiche number and from of cost you could you could just rank across of the results of a by by by popularity and a sign number propulsion for the popularity of that time and the questions of calls me how he met a popularity in away to the right Environment The after like the spate of white the all whatever you now I'm that usually that that is not the case all but you look at the incoming traffic Doha many people look at the site of a side that have a high counters so the number of accounts of the of the version but this is all rather theoretical to do that but it's a very easy I'm before the page to 2 to take the number that while at sold many calls and I'm so popular so it for you don't want to have something that is more of a ticket away and the patrons solution most lost quite simple and there were patches of the fact that this is 1 of the wonderful ideas and and computer side that simple ideas Spetchley taking from from into the sea it worked so well that it for the basis of multi billion dollar a multi national company and say what they did was they just stole the model of also show that is and applied it to the Web graph and what they did what they said about the number of in late is correlated to the 1st each and billing from good results a should become more than the league from that 1 of was exactly like we saw the 1st each of a site is basically the number of the press to you just from different signed for the with some multiplicity just from different side that point to the size of the the more Side point the by the 1st each Eastwood at the height of the Christie shuffle size pointing to hit the high of 52 of and the and fact and the
1:16:52
more behind that most of called of a random more but say what if I'm engine a Web so far the dust randomly walks through the and the Web so far ahead of the possible began navigate from page to page into following the though you go to 1 page look at lanes that other it too far you go then you could type in the new name of the website and to some of the different area of the we look to basic Navigation's with so you can make a random hyperlinked the you can type in a random your off and very off you will you will follow you would navigate the where he will follow the link structure and ready types of the state 94 the time you just navigate 10 per cent you of some and the pay trading band is the long term visit rate of each no so consider a random served had probably is that it will visit you a sort well the mornings point to use and the more links point to the site pointing to a new site for the more probability that but randomly costs for use by show but this is exactly the ingredients that we want we want the number of in the US and we want the press teach of the pages pointing to us which is the number of innnings for these pages and thus the remnants of a more cover exactly what we need in terms of pay track and found that the amount of kind of crude obviously you now Lycoming nobody's weblike like that the random NASA's not given at the end in a in reality because of cost you will served from topically related websites to to websites of other topics found that are similar out and and you will make your way but in a more focused of the of the remnants of I'm that it's it is the useful to consider the way because you used to have the idea of of what happened methodically from well some of the things that are the browsers out not consider so for example you might with a book might which are very of Lake of the Random typing of your life that you have the of the not too bad but about the fact that but not all of 18 like and in the more than all of this is not really a good model wept so but it is a sufficient model for modeling the 1st each of with the
1:20:12
plans and so and so that you more details version of the model of we stop at a random page a band at slip a quiet but scholes had Altria probability of 94 sent and if a child had and the current picture of default degree after follow 1 of the leading from the disciplinary gauge Evans and 98 per cent of the for my because my point shows at if the pointillist Teo's all the current rate of Nauplion served to a random we were pretty Oka this beautiful made and that the next this kind of the wicket the in tuition what happened so we have to grab the where graph the from the whupped opec I'm in the house so if we have a Webbrowsable we just to wire and they Wu and stops at the event and them with the point and say Well if the Conchos might said than 0 for low rent a blink this my of and I'm on this page 2nd time lot more than idle the same thing again and they were both at the point where all of the 10 per cent so I'd do again the random thing and and the so every point on the Web everywhere page has the Chaand to be addressed unit for but then if the pages many inmates page book yet that some of them out it as many indicates that chimes that 1 of this pages of cholo randomly stopping from it high because discrete out of the hole where just 1 or of the holder and the chaos that when you stop on this page you navigate of this page is than also high so the probability that this page is accessible by all remnants is higher because 1st mornings point which the results of which will be randomly opened the high end of costs if these falsus are well interlinked from on us all of the deaths have a high risk teach them not but the times higher that randomly by end up in 1 of these and again at the time of getting to 1 of notes the me in the next step the ability of mind set to this not for this the idea of of of
1:24:08
determining the and the US to use of the time and if you look at example and look at and Jason see metric of this very small way off over his head and a lead of 4 miles of the 0 2 just look at the state Systematics then I'd have transition metric that will say that are for example but there is an initial probability that that stops in every page and Lambert to depending on where in go so we find every game from page 1 in page bomb and want to navigate with 75 per cent probability to some pay to the only try of the half was going to pick a fight of it so in 75 per cent of the pages by goal with starting in page want to pay 5 good way this is what this 1 made of it if there would be once also via the would be a lie down here band with 75 per cent would have to be distributed on both page and how would have to be distributed you may recall was it's a random choice so half of the 75 per cent would go to page free and half of them the and would go to page fine but good thing we don't have that so that was 75 go to I'm Page 5 in this book but the of the still a 25 per cent missing what happens with the 25 per cent well it 25 per cent of the cases like type in a random but the comedy possibilities to appetite in a random replica of but the they jumped the 5 possibilities a even type in the address of the pace that I'm on right now so so after Distributed even need and 25 per cent distributed of the 5 pages makes 5 per cent but this is 1 of the media Kent have cost revived 75 per cent to go to page 5 by navigation and the 5 per cent by going to page 5 by typing in the euro al than have 80 per cent of possibilities what you find it that 8 5 after have been paid for and I'd let for full starting point so for every starting point stop but every starting point by but it was the probability of ending the gold page in 1 of the yep that's look at it for example you have paid for full of and but but this went and page full what's the idea of also 25 per cent 5 pages everybody gets the 5 per cent Brodie 75 per cent navigating where do and they begin to open shoppe to get page 3 times to get the page 1 you Distributed even need but December but 5 per cent and point of them but this is the distribution OK and while the number of connexions between such as called case and the metric and from start to go along with the probability of getting there is called the transition matches and and every step that my random so far take the transition metrics shows the of ability of getting to beach possible next time so it by stopping for example nodal 1 or and then go to the nett and notify in the next step the probability of ending up in any of the notes will be well stocked of 5 and of the the is basically by
1:29:54
remnants of what so was transition Maddox that that I'd in to use the idea of time and saved well if by Adam at time she in any of the starting point the probability that in time plus 1 by will be that any of the well paid just is basically well quite ascent 80 per cent 5 per cent and so on and so a doubles to with bomb but stop this at a suite of entry sell if I'm stopping at page suite what if I'm staying at tapes treatments that what time she before but to you that in the next step will be the ones to 5 per cent 80 per cent by the end of the was a basic the behind and of cost told a probability that would be any of these pages for with some of 1 book and this is the basic good
1:31:44
now Wikinews simulation by think we stop and want and now want to know the probability that after tea steps have been made by you a certain page so was appalled here was appalled to hear of the year of the new year Publity the of processors something to do with the team because it was stabilise over time and 1st in the 1st step it is kind of rather random where go to the decision of where a went with the flow and well hiking go to into the so if we do it but it as stop and state want lend using the transition metric said as well with 5 per cent would be eye was state in state 1 because I'd just acted with 5 per cent would be a huge because of the with 80 per cent would be a state 5 because and navigate this that type of but now I'm in Sept 5 to want to do to be the next step after my life the value from the transition maturate for starting in step 5 and going to where by and on to the probability that I'd got there in the 1st place so to get the 2nd step is you mean that I'd went to state find in the step in the previous I'm by declined the value from
1:33:55
the transition mantri so with 5 per cent stopping and that by the end up in the fund with the 80 per cent of
1:34:06
getting into the 5 in the 1st step UK and and the and 5 ascent of 80 per cent of Oakland all not that is basically the idea that the way his side got into state 5 in the 2nd and it is up reach the only possible if I'd got 2 6 5 and then stated state 5 but the of also so some also possible if I'd off some of that too step to and then stated went to see the 5 or if I'd got flow and when 2 5 4 4 2 3 and when 2 5 and it does have much to do with the UK and and the probability began the whole rope is all 1 they would do well to from not times it seems that the values not change very of any more that stabilise quest of causes now is that if your and the method of stabilises in this case denied to move the right examples for South some some red or of converting at and it is in the
1:35:38
I the Paul bility vector that as country so if if you see the limit case and just that time go by what any initial Paul ability victory will be converted and this is some of this can be proven by by some of 3 D abroad and the fury of so fast processes I'm when we say we have a network of size and the with a note work and the probability victory is and Ndimensional victory only entries all the ability of getting to getting to the point of being point from some of entries want you to be in you the point you on of and the and the and the unknown negative so it and of appalled Bellouti that you have been took 1 or 2 of the 4 at the 0 from and stop us backtracked we will just take calm and and across and metric began the Rose that up to 1 but we have not transition reference you have to go somewhere but the probability of going somewhere comfy by of 1 of gifts to the lower end of the entries are more megachurch banned you can't say that you
1:37:10
can build a mock of change with the stuff because the mock of chain basically is a set point of states end of the half 6 matches but you value for the transition between the want to and your ability but you find state to offer if the and that will so kind of the transition from ability going from state 1 who stayed to of full from state 1 you could also look to the reuel you could stay in the state 1 of the and all these probabilities year after add up to 1 because you have to do something for Kent the basic idea of the amount of chain basically and states and a plastic magic at kind of reflects on to the roads and the column the starting states and the goal states and is a transition a new into the time as discrete depth of time in every timestep you don't something you go somewhere and the ball ability with which he go somewhere and that is exactly what you by the hand this match so the ability that you are off in a certain state at the next point and time you that you were at a certain stage at the previous point in time that is the transition ability between 2 St OK while basic the what what
1:39:19
happens is that you find that the most states that you have competition between just by assigning the before the abilities of the conditions to just don't say any more its at possible going there but you say called probable it is going that's the only difference between find the machine and the might of change and the PM the more you know about the current state of U mock of change by can be expressed ability of actors but every point time I'm no hope probability you it is to be in a certain state and the probability is just the some of the ways of getting there but the ability of the ways so if there is hardly any way off getting into some state at any point in time through a very improbable to be that state if there is a lot of ways of getting some of the the probability of being that some in the state at any point of will be very high this is what the find its state Moscheo of what mock of change expressed but the probability recto that we get yet is basically the idea of the EU at a certain time the the and if we let Runtime to infinity the ability of the will be stable and just give us the ability to that we are taking any point in day if we don't know the point I'm going back to Eixample we find that the current state of the of the changes and state you and this is kind of the starting of so that is kind of like I'm we know it is a place for you and we have L transitions ability to any other state from you the of the of the confirmed by the taste of the American of is the influenced by of choice of Hall we draw random states and hauled off we navigate and the higher the possibilities of every getting somewhere a higher before the political being so for example a bomb if we have probability that just 3 states and we have 20 per cent 54 and that it was sent but means that changed probability taking any point in time in a week that time run for for of while and to the random crosses and there at some point with in and look the and the probability that will find we are in state 1 as 20 per cent the probability that we find the chain has some Hall moved into space to 30 per cent of the public with a fine save is actually in a free another beach to of that we had think of it as the distribution of wealth that as a
1:43:09
by basic idea so I'm the state transitions came before last using the match expect a modification that starting distribution and to get to the next step by just might apply the starting distribution with the transition matches because that tells me for every new state what is before ability of actually getting from the current but the public is to be broken and this is what a good time and again because after let time wrong for print so what are the state probabilities pie at the next point and time it's basically by taking the ball that ahead before I'm might apply it with transposed transition Matthew them by my get a new 1 for Kent this might occasion means grow time summed so is basically this prop away if you take entry life are but look at small excelled at 1 page you the 2nd page of the of possibilities of going from 1 to the other so you want to know that what could stay in state law what an essay about T 1 1 and you want to you they must some up to 1 because they have to go either weight like up just do something Derby at are while to states 3 days the of the account you have to stick to the world of the book probabilities have to some than my saved well if I'd have a certain probability of being instead 1 of the state to the probability of being instead 1 and being instead to in the next step of time will be well the transition culpability of 1 1 given that was it in the long thought the transition Paul bility the 2 1 only 1 the probability of being in a one off but the either Ikeme from state 1 OK and state the This Britain you thought I'd have been in state to and went to stay what this is written this the Paul ability of being in state 1 in the 2nd similarly the probability of being in the 2 in the 2nd set of his a stop in state to and stayed there thorough ice thought that in the state 1 and went over starting 1 going over Starting in state to stay in UK and that is not the way UK that the so epigones it to 1 the it up good and
1:47:03
I'm is actually use of everything that we need to to talk about the convergence properties of of the Mockus change which can say that if we have some some initial probability and have of probability that does a certain point in time band for some we you reach fixed point where another transition step T because 1 of would not change that and it that is the chain of lectures off for last 3rd the chain of probability victims converges to a fixed right following that means that of well basically has to be favourite for the transition Matrics because the change in the way and of this is kind of the idea of buying a victory some this fixed point direct arrested is basically true for full not eigenvectors that Kevin of I'd value in the match but that is size what they don't change and so starting with any probability victory this croesus of might applying at with the condition metric multiplying with the condition that it might applying with the from the the results finally at the height of the to an hour to immediately status talk and transition metric has denied value to a 1 but it is that all waste true that has a high value on well rather on good as much but the but it off with the back the D but it's not that
1:49:36
easy to use the game says it said it signed up to what was left of it after a unique some some some heavy laid algebra of proving it and the light on to go into the details of the pay off will be as during a that the state a stop astigmatic obtaining only of the has won as a one off items so it is that the true that the full with might of what and last this algorithm of all ways might applying if it Polidori's I'm Elway's might applying that the transition metric the stuff astigmatic on to some stopping distribution will lead in the end the home to height of a this during also also claims that 1 of the largest like value of the metric and there is only 1 of the eigenvector by having this out like value is unique so we will end up with 1 and only 1 of the 2 role developed random tell point was appalled ability that the lot of the rope and Paul abilities canopy negativity of and the negative we do only of of the and the and trees and that is good from best we have I'd like the and his called stationary of the mouth of Red stalwart we
1:51:33
do it is to take a random so from all we take you need all we want to know the you need stationary ability that by end looking at the way notion of 1st each we find but this is exactly what we are looking for we want to recursively defined the 1st tee of the note by citations all recommendations although the for every point the of the only other notes weighted by their 1st this means basically stop somewhere you and then look at in what note that we will end up in the long run so using this part will definitely the that at some point was station and record of all the ingredients of the half the concession metric that has only positive entries South has a lot like a lot eigenvector of value 1 and it has a unique eigenvector the launch of the iPad value and a poet which we can get this at the convergence and the numbers that we get that after but all of that we had her reflex the idea of what you look for perfect movement As
1:53:24
we know from the petrenko's was invented by the by Larry Page and their friends it's it was not clear that it is whether it's named after their page and unique that somebody who deals with the pages really caught page of the page Frank is the of the that have to do something to do with the pages all is named after the attack and that the lack of the tree nobody really and it has happened to them and in as time as a method for note drinking in the database and of the idea of the Web search engine and so it has happened and the Patten was assigned to stand for what actually bomb and Google at the company as the USAF license right I'm when founded and will took off stand for receive for their for the licensing of the invention about 120 8 million shares of who will and that the shares were actually sold in 2 thousand 5 for all of 300 million dollar so if any of you in meant something clever just tell me and but on strike University that that will be liquid for the next couple of it with this that it set up that the amazing amount of money of the of
1:55:05
the enough and I'm I'm not quite sure it of 5 from the world of the best point out found so I'm I'm not quite sure what this amounts to what 1 point 8 billion to build a church with the worst at the peak of boom of the pavement and the highest price of what may be the 1st of so found that the economy interesting to see what they rank actually death and said UConn take page Frank on its own but what you have to do is get to lend that somehow with with the content of the of case but if we look at it in a bin in and with suspect to what happens if for example a search for the term University found you will find that by Hugh by off techniques the optical physics at the University of Oregon is the top priority because it very also mentioned the term University and some terms of the of are quite University and that it just as the structure of the document at amazingly rather well knowing the University of Oregon without of of nobody because of the very smolt plays of optical Physics is those very small DePalma and the very small place this Bobby not what you want when you for university the very specific result Powell would you deal with the track sold with a train on the European averse to the 1st page would be stand for University of who knows about stand for university well everybody but so popular your the sort measures by the teams but ranks at higher and what you have here is kind of like it in order by the champagne during the also very good University of the University of for in urban from the University of the photo Iowa State though it's not perfect but that he still has some of the the prominent you as University the popularity of the new universities among the 1st results 1st year with the Carnegie Mellon which has a very famous University by don't want to know whether West Lady University he is Collier University Shoreham reseller camp interested in move the rest of the month couple state universities definitely although relevant results if you take the fury University at face value because they all universities but if you also assuming that he you that something in mind that you have some some some expectations when asking for university called the prominent universe of the popular 1st along the good University where it may be that all the results of the patron results seems to be much better than the of the book good
1:58:56
Soulchild quest we have the went wrong very smart it was 5 faces and which of the following note this as fuller by patron what would you say it it but they other that it took I seems to be a very of for no arrests for the names from and the seem to be good in this because the also collected the seems to be not as good Booker but opened beloved in when looking at the oboes with 2 in the next year the 2 in the US yet the 1 day in mid which you but what it but part of it should be the same Jugnot because they both linked to eat and they just failings subgrid 0 they should get the same patrons were about the and the faultless when you go that school and 1 0 where and 1 year was the 1 in 100 so they should be the same they should be the same the Supreme the high just the of the 2nd to the 3rd phase the correct on book and this is something that we found out by load and in this case it is also about consistent with the number of of incoming being so it would have just 4 of them by a number of incoming think we would have the same result but that doesn't have to be the case if and assumed that he 1 link yet the thing perfectly all of it and this but
2:02:21
it is all now we still have the problem of palm how to compute a picture of me which seemed some that we have to basically take the remnants of a model the site for whole of the do we look at random pages of belief Load case you and then just to highlight to regime by shooting a random vector that is the initial called the and during iteration allways of my decline with the systematic at some point it will be based on 0 1 2 6 point which it will be state so we stop was offered free of actors we do iteration with position Maddox via then we look at the mobilisation of the and do it all again and at some point the victim become stable if the like that of change any more without it is a speciality of bombs
2:03:37
we are a bit proves that followed regime conversos to the idea eigenvector and in and of large Victor and from the top of areas during we know that large like 1 so this is that we want and actually the number of iterations that you need it is dependent on the number of innings of 50 the more of ability to have a more difficult to get out that actually it's quite of considerably from 300 million lanes of 160 million they will find that after a about 50 iterations of we are already quite close to the origin of so 50 iterations if we don't meet the perfect require any move wave just beautiful the lanky together with the content's called so if not to a halt to fact to bomb but of
2:04:54
calls we don't have smouldered Torode with 5 web pages and 300 million things but we do we have paid Frank computation of up 16 billion web pages with a lot of the how to compute the poet aeration yet even if the 50 they ought to do all the uses a I'm distributed algorithm and has a lot of computing centres all over the world that I'm only doing page Frank computations together with the IAAF style that the valuation of the field and somehow or in the pages nobody really did not want to sell some said it is a NoleNet the ranking function of will does not only consists of page Frank and the IAAF by part of the law different things so it is known that the over 70 all they claimed that they of over 70 greedy for this ranking page rank as 1 of them but the very of for what and how they actually do it is that business secret and so that we can only conjecture yet but it principle of it can be done this kind of what lies hobbled of adult that they know that they really materialise before where the RAF in the transition because there will be 16 billion cross 16 billion metrics the kind of off to the left to cut break it del somehow but it well and and probably the transition metric will rather be broke with a lot of the road here and if you choose the bloke right that has been shown in network analysis that the Web used to be at these small world graph Smallwood graph means that he of many communities that the heavy the interlinked and these communities are some of connected to found which each other by published falsely and if you Brett could be part the small worlds this might be a good idea of calculating the some of the of the patron because then you can restrict public to a small part of batch it in a way yes in a way so as if you can detect the small would part of the Website and the and kind of uses for this tribution and take so this is the 1st billion of Web sites and this is the patron a new compute that 4 different cautioned of the web but and this come before not 1 of them in the care of ideas compete with to the level with interested in New and and the US and it has to be something of this of this in this way but it for the love of for it by the book the
2:09:11
good of that the the about it but it would be a lot but in the infrastructure of 50 the and in the never algorithm to good it and of calls me as we had last lecture when we were talking about falling 1 c o finished crawling you can't stop all over again at the same but of to the patron of the patient but if finish building the page Frank vector the station of the of just take the next called and do it again this also qualify distributed over algorithm because if you cross is distributed pick level which may be the focus for and computing the station a victims for this bought from where it's much easier that can become lower they were set up by the in part of a trend that has sometimes been exaggerated and that that probably stems from the from from the impact on the market of who because the only thing that was Hewitt who was the 1st in the 1st use of the patron of the will of the wrong breaking idea using links with time and the same for this kind of offer before all the existing algorithms and all the existing search engine I'm you will very often find the opinion that this is the ingredient that made who of the will is there for it is still today the ingredient making Google search of bad and all the other searches and it has been patented stuff and but a positive and algorithms is all ways of very very difficult thing because how do you know whether somebody uses it all not and believe me that the search engines popular all the sensible up or on the web to the use links structure of the way of cost they don't use page Frank because they would have licensing if that had been since with the exclusive lights and the debt that can be done so they use some of Algorithm for the rest of so I'm so it that it looks the premises its component its and import of all is probably not the most simple of what because you need to the textual things you need to be very of consuming the and protect around in text the continent of the approximately of pages just that this is structured in the small book grabbed in these ideas and trilogy of the pages you need a lot of things to to to read to the rank and the who will use you really law of the said of the on a remember people with that over 70 and for the more of the top of law Frank to bomb and there are rumours that page Frank actually original for only as the man to of legend the negligible effect on ranking from link structure is a thought at some point in the rankings links for that is what that we can can be sure of 1 of the problem with when when people became a where of the 2 successful because it uses the link structure of cost of spam industry as well but you this linked structure for banque so why don't we create Paid that give who will be exactly the links structure that they want that they are looking for the bell ranking and in and and a highway and sold the competition between spammers and such and and that want to fill spam stolid off and the EU will have to the change algorithms see you what links talked of a kind of point to spam and and so on and it said it very difficult if you just rely on the fact that I'm you you but that and the way you treat them if to some good but there are
2:14:25
certain varieties of page Frank so off for example of I'm 1 of the big advantage of a drink is that it has a single also called for each way results which is the 1st each but you don't you have 8 each but you have a oppressed each was was picked something so far reaching that as the most beautiful girl in the cost of cricket at the at the end that doesn't have the felt somebody is the most clever personal somebody is the most Web on silicon be different types of Preston with prospective when you have a certain I'm a certain topic in mind you don't want to know who is good at something Toby LSE but you want to know who was good at this topic of for example the asking at a period in the area of Physics the and even by mediocre physicists would help you more than the most brilliant chemistry guy levels how ever pay trained is computed in and the of the US the most brilliant chemist would have a much that a patron than your mediocre because of this led to leave to the idea basically found that the patron could somehow be made topics said that and its basic but I find that I'm on popular topics of for for Microsoft far all politics Allwood about and the and euro use classification of residents to assign each were results to a certain degree of the topix and the and wage topic you do what you did and and and focused following so you build focus falling into the model though you computer topics said the patron by limiting the random tell what to pages of book and so that you time you detect the topic and use the patrons full that were computed pre computed with respect for the public but at these takes you physicists a chemist stop off at that no are much of it is but which can be done of has yet to
2:17:10
anticipate the topix that the period is that the people in the Soham if we take a few like bicycling and look at the different pages in public when sensitive page ranked if we just used the look patron no topic whatsoever and the best way to we get is the real right as a venture closing Florida cycling waypoint for full board of it may be time Company of 50 but if we introduced a topic before of off computer business games where the band the Rank your the pages strongly changed so for example was a business we would get books about likely that you can buy for you could have some of some of by building its with computer it would get a Jeep yes pilot of the device that tells you when cycling where you actually off if you by as its what no recreation you would get a travel companies to buy the to of stopping fighting building up of both by the it 15th camp boys whether of 50 of offers some cycling so that it really shows and influence on on the results if you think puppets the and
2:19:05
interesting and so on but if you take the comparison to buy what they trained with the decision at the end measure up for a set of a set of theories than you will find that for different furious manually defined the of the residents of 1 of the 1st 10 results that under the by both that some the widest patron the public sensitive they usually is much better in terms of prestige and the on board for the maybe different but situation of company and it every time but while leverage taking the beams puppets of picture of Road that was basically some want
2:20:07
wanted save of pop topic of sensitive a frank about the loss other extensions so for example you could eliminate navigational early on a page of the book referring to stop pages and that of cost is not a quality vote for the top page but it's just kind of the direct to re services and on my paycheck and have a lot of navigation as being said icon basically navigate from every page to every other page which does not mean that the pages of uptick to the interest so this is kind of and of what you would try to get out of it and you would also try to eliminate the cystic playing so I'm nepotism as kind of making your brother of giving the brunt of the job time and to you what you want I would of cost to this kind of this these link founding all of the debating Unite you to my page Eileen to your page and also what's happening on Facebook like who gets the most people following them whether they interested all not enough but you don't want that you want to get rid of the to the for example of if the same person or such different sites that very high Paul ability even if they are not publicly to well connected that he will introduce something overlooked that my other sites sales of the 1 before it but that is not what you want for a link between Friends of all of the side scrapped but he's good guys that is not what you want your picture and it's it's kind of difficult to find out what would be the up and the same applies to of them detection very hot to find out whether this is actually a bit of fraud and structure of a web page or if somebody did it ordered the designed it that way because of the very practical because it makes actual said and with that we
2:22:26
get to the next the to of looking at the Blue to woo by looking at the web pages of some night we don't we will attention and and sometimes the next week or so of them through egg have a nice lunch consumer next week to give it up