This course deals with computer science (CS) aspects of social network analysis (SNA), and is part of the master computer science programme at Leiden University.
- The textual comments from the 2014 evaluation forms are available here.
- All papers have been graded. You should have received an e-mail containing your team's grade. Grades have also been submitted to the student administration.
- Looking for a skilled and ambitious student to do an 18 or 42 ECTS master project on Corporate Social Network Analysis. Contact me if you are interested!
Time and place: Fridays from 11:15 to 13:00 in the fall semester of 2014 in
Snellius room 403.
Prerequisites: basic knowledge of Algorithms and Data Mining
Literature: provided papers (no book)
Examination: based on presentation, paper, programming, peer review and participation (no exam). Specifically, an individual homework assignment (20% of the grade) and a course project in teams, consisting of a presentation (30%), course project paper (50%) and small peer review assignment as main deliverables.
Study points: 6 ECTS
Lecturer: Frank Takes - firstname.lastname@example.org, Snellius room 156a
|1.||Sep 5, 2014||Lecture 1: Introduction and small world phenomenon|
|2.||Sep 12, 2014||Lecture 2: Centrality, distance measures and densest subgraphs
Example presentation: Determining the Diameter of Small World Networks
|Sep 19, 2014||Lab session from 9.00 to 10.45 in room 302|
|3.||Sep 19, 2014||Lecture 3: Structure of the web and propagation-based centrality algorithms
Deadline for choosing a course project topic
|4.||Sep 26, 2014||Deadline for handing in the homework assignment
Lecture 4: Temporal social network analysis
|5.||Oct 10, 2014||Student presentation 1: Link prediction
Homework evaluation and course project progress discussion
|6.||Oct 17, 2014||Student presentation 2: Diameter computation
Student presentation 3: Visualization
|7.||Oct 24, 2014||
Student presentation 4: Graph compression/representation
Student presentation 5: Community detection
|8.||Oct. 31, 2014||Deadline for first three paper sections (conditionally extended from Oct. 24)
Peer review session
On-site peer review in team pairs
|9.||Nov 7, 2014||Student presentation 6: Graph partitioning
Student presentation 7: Virality and outbreak detection (2)
|10.||Nov 14, 2014||Student presentation 8: Virality and outbreak detection (1)
Student presentation 9: Sampling methods
|11.||Nov 21, 2014||Optional deadline for intermediary paper "check"
Student presentation 12: Shortest path computation
Student presentation 11: Neighborhoods
|12.||Nov 28, 2014||Student presentation 12: Closeness centrality
Course evaluation and feedback
|13.||Dec. 4, 2014
Dec. 5, 2014
|Evaluation sessions cf. schedule
Evaluation sessions cf. schedule
|Dec. 12, 2014
||23:59:59: Deadline for full course project paper (conditionally extended from Dec. 5)|
Network with 1458 nodes and 1948 edges.
We will use the UNIX workstations in room 302 for this lab session.
If you really must, you can attempt to use Windows, fully at your own risk (you will at least need to edit gephi.conf so that it points to the correct Windows JRE version 6 or 7 (and not 8)).
Some information on the workstations is provided by the ISSC, see their pages on the
SSH access and
remote access to UNIX systems with a configuration similar to the workstations.
There is no deliverable for this lab session. The main goal of this lab session is to become familiar with Gephi (experimental beta-software to visualize networks for research purposes) and its input format.
Teams work on a course project, consisting of:
Suggestions for topics are listed below, and you are encouraged to pick one of these topics (as a team). You are also allowed to suggest your own problem (topic) in the field of social network analysis, assuming you are have found at least three scientific papers that suggest different techniques or algorithms for solving this problem (so, addressing this same topic). Please choose a topic together with your team mate before September 26, and inform the lecturer via e-mail of your choice. Also feel free to suggest a free date slot for your presentation. Your topic choices will be processed on a first-come-first-serve basis.
If you choose a topic, then you are also encouraged to look up additional papers in this problem domain to study and compare to the paper that you presented, as the papers to study are merely "simple" suggestions. For some topics, there exists an overview/review paper that discusses many more papers that study the same (type of) problem.
Scientific papers (ACM, Elsevier, etc.) can often only be opened from within the university domain (or from home via SSH/nxClient/VPN/RDP).
IEEE Explore papers can be opened by looking them up via computer.org. Alternative links and preprints of papers can often be found through Google Scholar.
This course deals with the computer science aspects of social network analysis. With topics such as big data and data science becoming more and more popular, the study of large datasets of networks (or graphs), is becoming increasingly important. Examples of such networks include webgraphs, communication and collaboration networks and perhaps most notably (online) social networks (such as Facebook and Twitter). With millions of nodes and possible billions of links, traditional graph algorithms are often too complex and unable to solve trivial algorithmic and data mining related problems. Typical tasks in this field include community detection, clustering, outlier detection, link prediction but also more fundamental problems such as efficient retrieval, storage, and compression of graph data and computational problems such as computing shortest paths and other descriptive graph properties. Also see the description in the e-Studyguide.