1. Social Networks as a Foundation for Computer Science
Jeffrey Forbes

2. Where are we going … Questions
What should our concerns be for those choosing to major in Computer Science?
courses, research, jobs, …
Should we be concerned by the precipitous decline in those taking our courses or majoring or …?
majors, technical students, non-technical …
What can we do to ensure the ongoing success of our academic discipline?
Look inward, look to others

3. Acknowledgements Social Networks/Broadening Participation group:
Owen Astrachan developed most of this talk
Casey Alt, Richard Lucic, Susan Rodger
Students: Ben Spain & Dametrious Peyton
Drawn from the work of:
Michael Kearns, UPenn
Eytan Adar, formerly of HP Labs
John Breese, David Heckerman, Microsoft Research
Jonathan Herlocker, Oregon State University
Thomas Hoffman, Brown University
Marti Hearst, UC Berkeley
Jennifer Golbeck, University of Maryland

5. Computer Science What is the foundation of computer science?
Historically, now, in the future
What changes are here, on the horizon?
From theory to practice to education
Can we relate to what and how students learn?
Is every generation different, the same, …

6. WWDD?
Edsger Dijstra is a famous computer scientist who won a Turing Award in 1972 and died in 2002
Computer scientist who Composed manuscripts with fountain pen
Wrote a paper titled th Cruelty of Really Teaching Computer Science

7. Questions and Answers
Judge a man by his questions rather than by his answers
Voltaire

8. History and Computer Science
Those who cannot remember the past are condemned to repeat it.
Don’t know much about history, don’t know much about biology, don’t know much about a science book
Who are these guys?
George Santayana
Sam Cook

9. Who, when?
“No stretching … is required to envision computer consoles installed in every home.
Everyone will have better access to the Library of Congress than the librarian himself now has.
Full reports on current events, whether baseball scores, the smog index in Los Angeles or the minutes of the 178th Korean Truce Commission will be available for the asking.”
Who said this? When?
John, McCarthy, Information, Chapter 1, 1966

10. You win some, you lose some
People will soon become discontented with the “canned” programs available; they will want to write their own. The ability to write a computer program will be as widespread as the ability to drive a car.
Not knowing how to program will be like living in a house full of servants and not speaking their language.
Many people can write simple programs after an hour or two of instruction. … Programming is far easier to learn than a foreign language or algebra.

11. kentlew.com
Then and Now
Bailey, SIGCSE 1972
It is remarkable that the majority of students can indeed handle fairly complex (Fortran) I/O by the end of the first six lessons, even though they have not actually been formally taught how to do it.
Roberts et al, SIGCSE 2006
The problem most often cited by those attempting to teach Java to novices is the lack of a simple input mechanism,

12. What has changed in 20 years?
Machines
Characteristics and Availability
Internet
Availability, IM, web, Google, …
Students
Comfort with technology, Expectations

13. The more things change…?
Assume I took your first course(s) in 1984 and understood the concepts so completely that I could still get a 100 on the final from 1984 if I took it today (e.g., I've been in a cryogenic chamber). How would I do on the 2004 final exam?

14. What has changed in Physics?
"You'd get a 100 plus or minus sigma. Intro classical physics hasn't really changed that much over the last 100 years. In graduate level e.g. E&M or quantum classes I think ditto, although sigma would be bigger (and might depend more on the instructor variation than on any real variation in the material). The main difference is, I think, that your chances of GETTING 100 now would be much higher."
Rob Brown,
Head of Physics Instruction

15. What has changed in Biology?
"The basic principles and concepts of biology haven't changed much in 20 years.  What has changed relates to specific content, and in this arena the changes have been enormous.  20 years ago, we barely knew how to sequence DNA; today information of this kind has had a major impact on just about every topic in the biological sciences.  Thus, some questions on an exam today would address topics that would be completely unfamiliar to a 1984 time-traveler. "
Greg Wray,
Director of Undergraduate Studies, Biology

16. What has changed in Economics?
"… we now cover material that was only introduced in an advanced or intermediate course in In 1984 we spent the bulk of the time dealing with the Keynesian model and virtually no dialogue about supply side policies. Now the Keynesian stuff is a small subset of a much broader exposure to Aggregate demand and supply… Also there is more international coverage now - as opposed to 20 years ago for obvious reasons."
Lori Leachman,
Director of Undergraduate Studies, Economics

17. What has changed in Calculus?
We have two varieties of calculus courses, the lab courses and the traditional ...  The latter two have not changed significantly in decades, and I think that a student who fared well on the 1984 exam in those courses would do well today, and vice versa.
[In the lab courses] You would ace about half the exam.  The other half would be unfamiliar to you.  For example, you would probably not know how to answer a problem on modeling a set of data, creating an approximation using Euler's method, interpreting derivatives in the context of applications in other fields, or giving explanations of ideas …
Lewis Blake,
Supervisor of First-year Instruction

18. Changes in Computer Science?

19. Changing CS? Rock, Hard place
If Computer Science has changed drastically is it to keep up with fads and stylistic changes or because of fundamental changes in the discipline?
Are we leveraging the technological and intellectual resources at our disposal
If we haven’t changed, is it because of a solid bedrock of principles that endures? Or because we’re lazy, good-for-nothing, …

20. What is CS? Who wants to study it? Why do they want to?

22. NY Times in 1984

23. What is CS? Why study it? Do we have Physics (Math, …) Envy?
“It's hard for voice over Internet Protocol or e-commerce to compete with finding the age of the universe,” Peter Lee, CMU
Does familiarity breed contempt?
What was different in 1984 than today?

25. Occupational Distribution of Projected S&E Job Openings 2002-2012
Natural
Science
Managers
Information
Technology
Engineers
Physical
Scientists
Life
John Sargent, US Department of Commerce, 2004

26. Annual Degrees and Job Openings in Broad S&E Fields
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
Engineering
Physical Sciences
Computer Sciences
Biological /
Agricultural Sciences
PhD
Master's
Bachelor's
Projected Job Openings
John Sargent, US Department of Commerce, 2004

27. Demographics: 18 - 24 year olds
White
Asian/
Pacific
Islander
Hispanic
African American
Native American
2000
66% 4%
15%
14% 1%
2010
63% 5%
17%
2025
55% 7%
22%
US Census Bureau

28. Bachelor’s Degrees from Doctoral Institutions

29. How are Black students doing?

30. How are Black students doing? (2)

31. How are black students doing? (3)

32. Awarded Degrees in CS B.A./ B.S. M.S. Ph.D. Female 28% 34% 23% White
57%
53%
76%
African American
10% 8% 3%
Hispanic 5% 2%
Native American / Alaskan Native
.6%
.7%
.2%
Persons with Disabilities 1%
Women, Minorities, and Persons with Disabilities in S&E, 2004 (2001 data)

33. Tough stats Between 1970 and 2001, CS/CE doctorates 8,913 to whites
154 to blacks
8,915 tenure-track faculty at 177 departments
497 Women
63 Hispanic*
32 African American*
6 Native Americans*
2003 Taulbee Data

34. More local

37. COHFE
Amherst College, Barnard College, Brown University, Bryn Mawr College, Carleton College, Columbia University, Cornell University, Dartmouth College, Duke University, Georgetown University, Harvard University, Johns Hopkins University, Massachusetts Institute of Technology, Mount Holyoke College, Northwestern University, Oberlin College, Pomona College, Princeton University, Rice University, Smith College, Stanford University, Swarthmore College, Trinity College, University of Chicago, University of Pennsylvania, University of Rochester, Washington University in St. Louis, Wellesley College, Wesleyan University, Williams College, Yale University

39. If you don’t take a course in CS, you won’t major in it.

40. Who's going to College?

41. Who's going to College?

42. Who's going to College?

43. How do we get Students into the CompSci Tent?
Does anyone know the North Carolina motto?
esse quam videri (to be rather than to seem)
in some ways, we need to do the opposite as computer scientists

44. Interdisciplinary minors
At Duke it is difficult to double major in sciences
Too many requirements, 17 courses in biology
Students are interested in credentials
No business major/minor, certificate program (requires intro, capstone, six courses)
Minor requires five courses, double counting ok
Three courses in CS, two in econ or biology
From gene to social networks, data mining, …

45. A Future for Computer Science?

46. Is there a Science of Networks?
From Erdos numbers to random graphs to Internet
From FOAF to Selfish Routing: apparent similarities between many human and technological systems & organization
Modeling, simulation, and hypotheses
Compelling concepts
Metaphor of viral spread
Properties of connectivity has qualitative and quantitative effects
Computer Science?
From the facebook to tomogravity
How do we model networks, measure them, and reason about them?
What mathematics is necessary?
Will the real-world intrude?

47. Physical Networks US Power Grid
The Internet
Vertices: Routers
Edges: Physical connections
Another layer of abstraction
Vertices: Autonomous systems
Edges: peering agreements
Both a physical and business network
Other examples
US Power Grid
Interdependence and August 2003 blackout

48. What does the Internet look like?

50. US Power Grid

51. Business & Economic Networks
Example: eBay bidding
vertices: eBay users
links: represent bidder-seller or buyer-seller
fraud detection: bidding rings
Example: corporate boards
vertices: corporations
links: between companies that share a board member
Example: corporate partnerships
links: represent formal joint ventures
Example: goods exchange networks
vertices: buyers and sellers of commodities
links: represent “permissible” transactions

52. Content Networks Example: Document similarity
Vertices: documents on web
Edges: Weights defined by similarity
See TouchGraph GoogleBrowser
Conceptual network: thesaurus
Vertices: words
Edges: synonym relationships

53. Enron

54. Social networks Example: Acquaintanceship networks
vertices: people in the world
links: have met in person and know last names
hard to measure
Example: scientific collaboration
vertices: math and computer science researchers
links: between coauthors on a published paper
Erdos numbers : distance to Paul Erdos
Erdos was definitely a hub or connector; had 507 coauthors
How do we navigate in such networks?

56. Acquaintanceship & more

57. Network Models (Barabasi)
Differences between Internet, Kazaa, Chord
Building, modeling, predicting
Static networks, Dynamic networks
Modeling and simulation
Random and Scale-free
Implications?
Structure and Evolution
Modeling via Touchgraph

58. Web-based social networks
Myspace 73,000,000
Passion.com 23,000,000
Friendster 21,000,000
Black Planet 17,000,000
Facebook 8,000,000
Who’s using these, what are they doing, how often are they doing it, why are they doing it?

59. Golbeck’s Criteria Accessible over the web via a browser
Users explicitly state relationships
Not mined or inferred
Relationships visible and browsable by others
Reasons?
Support for users to make connections
Simple HTML pages don’t suffice

60. CSE 112, Networked Life (UPenn)
Find the person in Facebook with the most friends
Document your process
Find the person with the fewest friends
What does this mean?
Search for profiles with some phrase that yields matches
Graph degrees/friends, what is distribution?

61. CompSci 1: Overview CS0 Audioscrobbler and last.fm
Collaborative filtering
What is a neighbor?
What is the network?

62. What can we do with real data?
How do we find a graph’s diameter?
This is the maximal shortest path between any pair of vertices
Can we do this in big graphs?
What is the center of a graph?
From rumor mills to terrorists
How is this related to diameter?
Demo GUESS (as augmented at Duke)
IM data, Audioscrobbler data

63. My recommendations at Amazon

64. And again…

65. Collaborative Filtering
Goal: predict the utility of an item to a particular user based on a database of user profiles
User profiles contain user preference information
Preference may be explicit or implicit
Explicit means that a user votes explicitly on some scale
Implicit means that the system interprets user behavior or selections to impute a vote
Problems
Missing data: voting is neither complete nor uniform
Preferences may change over time
Interface issues

66. Memory-based methods
Store all user votes and generalize from them to predict vote for new item
Predicted vote of active user a for item j:
where there are n users with non-zero weights, vi,j is the vote of user i and item j,  is a normalizing factor,
w() is a weighting function between users
Distance metric
Correlation or similarity

67. Computing weights - Cosine Correlation
In information retrieval, documents are represented as vectors of word frequencies
For CF, we treat preferences as vector
Documents -> users
Word frequencies -> votes
Similarity is then the cosine between two vectors
Dot product of the vectors divided by the product of their magnitudes

68. Computing weights - Pearson & Spearman correlation
Pearson Correlation
First used for CF in GroupLens project [Resnick et al., 1994]
Relatively efficient to calculate incrementally
Spearman Correlation
same as Pearson but calculations are done on rank of va,j and vi,j

69. Model-based methods
Really what we want is the expected value of the user’s vote
Cluster Models
Users belong to certain classes in C with common tastes
Naive Bayes Formulation
Calculate Pr(vi|C=c) from training set
Bayesian Network Models