Putting people into models
This presentation is the property of its rightful owner.
Sponsored Links
1 / 35

Putting people into models Social networks and Bayesian networks PowerPoint PPT Presentation


  • 128 Views
  • Uploaded on
  • Presentation posted in: General

Putting people into models Social networks and Bayesian networks. Ingrid van Putten CSIRO – Marine and Atmospheric research (Hobart- Australia). Jacopo A. Baggio School of Human Evolution &Social Change, Arizona State University. What are networks and what are social networks.

Download Presentation

Putting people into models Social networks and Bayesian networks

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


Putting people into models social networks and bayesian networks

Putting people into models

Social networks and Bayesian networks

Ingrid van Putten

CSIRO – Marine and Atmospheric research (Hobart- Australia)

Jacopo A. Baggio

School of Human Evolution &Social Change,Arizona State University

What are networks and what are social networks

Where did it all start?

Small world, random, and scale-free networks

Fisheries example – quota trade market

Linking qualitative, networks and Bayesian networks

How do Bayesian networks work?


Putting people into models social networks and bayesian networks

Leonhard Euler

(1707 – 1783)

Networks: How did it all start?

Commentarii Academiae Scientiarum Imperialis Petropolitanae, vol. 8, pp. 128-140, 1736

The problem, which I am told is widely known, is as follows: in Königsberg in Prussia, there is an island A, called the Kneiphof; the river which surrounds it is divided into two branches, as can be seen in the figure, and these branches are crossed by seven bridges, a, b, c, d, e, f and g.

Concerning these bridges, it was asked whether anyone could arrange a route in such a way that he would cross each bridge once and only once.


Putting people into models social networks and bayesian networks

Networks: How did it all start?

Definitions (in modern words):

A network is a figure of points (vertices/nodes/actors) connected by non-intersecting curves (edges/links/ties). A vertex is called odd if it has an odd number of arcs leading to it. An Euler path is a continuous path that passes through every arc once and only once.

Theorems:

If a network has more than two odd vertices, it has no Euler paths;if it has two or less odd vertices, there is at least one Euler path.


Putting people into models social networks and bayesian networks

What is a network?

An arrangement of intersecting horizontal and vertical lines

1 a network of arteries WEB, lattice, net, matrix, mesh,

crisscross, grid, reticulum, reticulation; Anatomy plexus.

2 a network of lanes MAZE, labyrinth, warren, tangle.

3 a network of friends SYSTEM, complex, nexus, web, webwork.

What is a SOCIAL network?

A social structure that is made up of entities/ agents/ individuals/ or organisations that have ties/ relationships (interactions) between them.

Physical networks

Ties or relationships could be anything ….

Kinship

Friendship

Informationexchange

Marketexchange

[http://serialconsign.com/2007/11/we-put-net-network]


Putting people into models social networks and bayesian networks

Physical interaction networks

The Blogosphere

Biochemical networks

Gene-protein networks

Food webs: who eats whom

The World Wide Web (?)

Airline networks

Call centre networks

Paper citations

Social interaction networks

Friendships

Acquaintances

Boards and directors

Organizations

facebook.com

twitter.com


Putting people into models social networks and bayesian networks

node, vertex, actor

link, edge, tie

V = {v1…vn}

Graph G(V,E)

E = {e1…en}

1

1

4

2

4

2

3

5

3

5

1

1

2

4

2

4

3

5

3

5

Networks and Matrices

Undirected

Directed

Not symmetrical

Symmetrical


Putting people into models social networks and bayesian networks

How did interest in social networks start?

……………. Six degrees of separation …………………

Stanley Milgram (and other researchers) carried out what is now known as “The small world experiment”

The experiments are often associated with the phrase "six degrees of separation", although Milgram did not use this term himself.

The research was groundbreaking

human society is a network characterized byshort path (chain) lengths


Putting people into models social networks and bayesian networks

How did Milgramdo the experiment?

1

Information packets sent to "randomly" selected individuals around USA.

Packets had basic information about a target contact person in Boston (Boston stockbroker). This person is the end destination for the packet.

If the recipient personally knew the Boston stockbroker described in the letter, they should forward the letter directly.

2

If they did not know the Boston stockbroker personally, then the person was to think of a friend or relative he knew personally who was more likely to know the target. Could only send to someone with whom they were on a first-name basis

Recipient was directed to sign his name on a roster and forward the packet to the next person

3

When and if the package eventually reached the contact person in Boston, researchers could examine the roster to count the number of times it had been forwarded from person to person

4

20% of packets reached target

Chain length ' 6.5

CSIRO.


Putting people into models social networks and bayesian networks

Milgram’s experiment

John Guare wrote a play called Six Degrees of Separation, based on this concept. One of the main character’s lines (Quisa)

Chain length ' 6.5

“Everybody on this planet is separated by only six other people. Six degrees of separation. Between us and everybody else on this planet. The president of the United States. A gondolier in Venice… It’s not just the big names. It’s anyone. A native in a rain forest. A Tierra del Fuegan. An Eskimo. I am bound to everyone on this planet by a trail of six people…”

CSIRO.


Putting people into models social networks and bayesian networks

Erdős Number (Bacon game for the scientist)

Paul Erdős was an influential and itinerant mathematician (often living out of a suitcase boarding with his colleagues). He published more papers during his life (at least 1,525) than any other mathematician in history (with 507 co-authors)

Number of links required to connect scholars to Erdős, via co-authorship of papers

Paul Erdős (1913-1996)

Jerry Grossman’s (Oakland Univ.) website allows mathematicians to compute their Erdos numbers: http://www.oakland.edu/enp/

  • Connecting path lengths, among mathematicians only:

    • average is 4.65

    • maximum is 13


Putting people into models social networks and bayesian networks

Random Graphs --- or why does the “small world” phenomena exist?

N = nodes (individuals)

p= number of nodes with links

(A pair of nodes has probability p of being connected)

K=number of links

(Average degree, k ≈ pN)

p=1.0 ; k≈N

p=0 ; k=0

N = 12

N = 12

Now put in few random connections

Each person is connected to two neighbours either side

B

Number of steps to get from A to B reduced to two

A

Takes three steps to get from A to B


Putting people into models social networks and bayesian networks

small-world network

L= avg shortest path length

C = avg clustering coefficient


Putting people into models social networks and bayesian networks

Most networks are not random but are ‘scale free’

  • Tend to have a relatively few nodes of high connectivity (the “Hub” nodes – or “broker” nodes)

Our world complies with the Pareto principle

(also known as the 80–20 rule, the law of the vital few)


Putting people into models social networks and bayesian networks

Degree Distribution & Power Laws

Albert and Barabasi (1999)

Many real-world networks exhibit a power-law distribution (also called “Heavy tailed” distribution)

P(k)

Number of nodes with k links

Lots of nodes with only a few links

(k)

Number of links

Power laws in real networks:

(a) WWW hyperlinks

(b) co-starring in movies

(c) co-authorship of physicists

(d) co-authorship of neuroscientists

(e) Distribution of wealth

Power-law distributions are straight lines in log-log space


Putting people into models social networks and bayesian networks

Power Laws ….. Scale-Free Networks

CSIRO.


Putting people into models social networks and bayesian networks

Power Laws ….. What happens if you take out a few hubs?

Take out 9 centres

Take out 7 centres – but target the hubs


Putting people into models social networks and bayesian networks

Epidemic spreading

Structure matters: Power law versus random networks

Structure matters: what do we know?

- Structural properties influence a system strengths and weaknesses.

- Structural properties influence diffusion processes such as viruses, pests, communication, information, migration and so on.

- There is no golden rule (the “perfect” structure for all systems does not exist)


Putting people into models social networks and bayesian networks

Australian fisheries example of network analysis:

Lease quota trade for lobsters

Industry structural change after tradeable quota introduced


Putting people into models social networks and bayesian networks

  • Mapping the lease quota trade

  • (each line is a trade between two individuals)

1999 (year after the introduction of quota)

2007 (8 years later)

New relationships – more brokers / hubs


Putting people into models social networks and bayesian networks

Income supplementer

Lease market network

Lease quota dependent fisher

Investor

Quota redistributor

Independent fisher

Active fishers


Putting people into models social networks and bayesian networks

25

Portfolio investors

D

C

Ownership characteristics (number of quota units owned by fisher)

Concentration of ownership

B

75

A

Fishing effort (number of quota units fished)

Income supplementers (A-C-D)

Lease dependent fishers (A-B-C)

Investors (A-D)

Quota redistributors (A-B-C-D)

Independent fishers (A-C)

Van Putten(2011)


Putting people into models social networks and bayesian networks

Putting people into models

Social networks and Bayesian networks

Ingrid van Putten

CSIRO – Marine and Atmospheric research (Hobart- Australia)

Jacopo A. Baggio

School of Human Evolution &Social Change,Arizona State University

Linking qualitative, networks and Bayesian networks

How do Bayesian networks work?


Putting people into models social networks and bayesian networks

Bayesian models

Network models

Qualitative models

Conditional probabilities

p2

p3

N1

N2

Cold

(1)

Flu

(2)

+

+

undirected

p1

N3

Fever

(3)

-

Road between power station 1-2, and 1-3, but not between 2-3

If you have a cold (1) there is a chance you have a fever (3), and if you have the flu (2) there is also a chance you have a fever

i2

i3

Animal 3 experiences external factors that limit it (self effect). Animal 1 has a positive effect on animal 2, and animal 2 also has a positive effect on animal 3, but animal 2 has no effect on animal 1 (commensalism)

directed

i1

Cold

Flu

Fever

True

False

Fever

True

False

0

0.6

0.1

True

0.7

True

Individuals 1-3 are friends with each other, and 1 is friends with 2, but 2 doesn’t feel like 1 is their friend and 2-3 are not friends at all

False

0.4

1

False

0.3

0.9


Putting people into models social networks and bayesian networks

Bayes' theorem gives the relationship between

the probabilities of Aand B

P(A) and P(B)

and the conditional probabilities of Agiven B and B given A

P(A| B) and P(B | A)

Thomas Bayes(1701-1761)

A Bayesian network is a directed graph

Each node represents a random variable.

Each node represents a variable A with parent nodes representing variables B1, B2,..., Bn

Each node is assigned a conditional probability table (CPT)


Putting people into models social networks and bayesian networks

Smoking

Visit to Asia

Tuberculosis

Lung Cancer

Bronchitis

Tuberculosis

or Cancer

XRay Result

Dyspnoea(SOB)

Example from Medical Diagnostics

Patient Information

Medical Difficulties

Diagnosis

Diagnostic Tests

Network represents a knowledge structure between medical difficulties, their causes and effects, patient information and diagnostic tests


Putting people into models social networks and bayesian networks

Smoking

Visit to Asia

Tuberculosis

Lung Cancer

Bronchitis

Tuberculosis

or Cancer

XRay Result

Dyspnoea (SOB)

Example from Medical Diagnostics

Dyspnea

Bronchitis

Medical Difficulties

Tub or Can

True

True

False

False

Bronchitis

Present

Absent

Present

Absent

Present

0.90

0.70

0.80

0.10

Absent

0.l0

0.30

0.20

0.90

Medical Difficulties

CSIRO.


Putting people into models social networks and bayesian networks

Smoking

Visit to Asia

Tuberculosis

Lung Cancer

Bronchitis

Tuberculosis

or Cancer

XRay Result

Dyspnoea (SOB)

Example from Medical Diagnostics

CSIRO.


Putting people into models social networks and bayesian networks

We have some information about the patient

We know the person has been to Asia

From P=1.04

From P=6.48

From P=43.6

From P=11.0

Given evidence about a cause, what are the predicted effects (e.g. you know the person has been to Asia what is the probability that they have tuberculosis?)

Predictive reasoning


Putting people into models social networks and bayesian networks

We also can now see the x ray results are normal

Increases the probability that it’s not tuberculosis or cancer

X-ray results are normal ….

Given evidence about an effect (symptom) how does this change our beliefs in the causes? (e.g. I observe there is nothing abnormal about the x-ray– how does that the affect the probability that it’s tuberculosis or cancer?)

Diagnostic


Putting people into models social networks and bayesian networks

Australian fisheries example of BBN: Torres Strait

(between Papua New Guinea and far northern Australia)


Putting people into models social networks and bayesian networks

Australian Examples: Torres Strait

Non-indigenous commercial catch

SEC fishery

Pre-season survey

Lobster abundance

Cost related drivers

Papuan

Price related drivers

Hookah ownership

Private freezer

Season

Exchange rate

Fuel costs

Functional Island freezer

Price live

Weather

Fishing costs

Regional Authority ($)

Price tails

Ease of catching lobster

Other lobster available

Community business knowledge

Socio-cultural drivers

Returns from fishing

Full time alternative income

Community role models

Government employment scheme

Crew availability

Working age men

Incidental household payments

Social capital

Tradition & culture

Profit drivers

Casual fisher

Part time fisher

Full time fisher


Putting people into models social networks and bayesian networks

Australian Examples: Torres Strait

Non-indigenous commercial catch

SEC fishery

Pre-season survey

Lobster abundance

  • Objective: more full time indigenous fishers (use olympic quota, ITQ, community quota ?)

  • Assumed: economic drivers = key

  • Actually: socio-cultural & infrastructure

Cost related drivers

Papuan

Price related drivers

Hookah ownership

Private freezer

Season

Exchange rate

Fuel costs

Functional Island freezer

Price live

Weather

Fishing costs

Regional Authority ($)

Price tails

Ease of catching lobster

Other lobster available

Community business knowledge

Socio-cultural drivers

Returns from fishing

Full time alternative income

Community role models

Government employment scheme

Crew availability

Working age men

Incidental household payments

Social capital

Tradition & culture

Profit drivers

Casual fisher

Part time fisher

Full time fisher


Putting people into models social networks and bayesian networks

Australian Examples: Torres Strait

Non-indigenous commercial catch

SEC fishery

Pre-season survey

Lobster abundance

  • Full time fishers

    • Economics (profit) is a driver

    • Social capital important too (crew, freezers)

Cost related drivers

Papuan

Price related drivers

Hookah ownership

Private freezer

Season

Exchange rate

Fuel costs

Functional Island freezer

Price live

Weather

Fishing costs

Regional Authority ($)

Price tails

Ease of catching lobster

Other lobster available

Community business knowledge

Socio-cultural drivers

Returns from fishing

Full time alternative income

Community role models

Government employment scheme

Crew availability

Working age men

Incidental household payments

Social capital

Tradition & culture

Profit drivers

Casual fisher

Part time fisher

Full time fisher


Putting people into models social networks and bayesian networks

Australian Examples: Torres Strait

Non-indigenous commercial catch

SEC fishery

Pre-season survey

Lobster abundance

  • Part time fishers

    • Socio-cultural is key

    • Ease of access vs other income

Cost related drivers

Papuan

Price related drivers

Hookah ownership

Private freezer

Season

Exchange rate

Fuel costs

Functional Island freezer

Weather

Price live

Fishing costs

Ease of catching lobster

Regional Authority ($)

Price tails

Socio-cultural drivers

Other lobster available

Community business knowledge

Returns from fishing

Government employment scheme

Full time alternative income

Community role models

Crew availability

Incidental household payments

Working age men

Social capital

Tradition & culture

Profit drivers

Casual fisher

Part time fisher

Full time fisher


  • Login