Projects on(Social) Network Analysis
F.W. Takes
Leiden University, The Netherlands
Masterclass — September 16, 2015
Projects — Masterclass — September 16, 2015 1 / 7
Projects
Topics: social network analysis, network science, graphmining, data mining, data science, algorithms
Quantitative studies (so, with concrete algorithms, largereal-world datasets and experiments)
Project work in C++, Python or other Linux-friendly languages
Thesis in LATEX
Relevant courses
1 MSc: Social Network Analysis for Computer Scientists (orComplex Networks)
2 BSc: Algorithms, Data Structures and Data Mining
Interested? Contact me at [email protected] orwalk by Snellius room 157b
Projects — Masterclass — September 16, 2015 2 / 7
Project 1: Corporate Network Analysis
Analyze global network of 10 million firms across the globe
Firms are connected due to board interlocks or ownership ties
Interdisciplinary project: computer scientists and social scientists
Approach: (social) network analysis techniques: centralitymeasures, community detection algorithms, etc.
Collaboration with University of Amsterdam, CORPNET project
Interested? [email protected] or find me in 157b
Projects — Masterclass — September 16, 2015 3 / 7
Figure : Board interlock network (200 000 nodes and over 1 000 000 edges)
Project 2: Comparing Graph Frameworks
Many network analysis toolkits
Not all equally suitable for analyzing “big” network data
Compare toolkits: SNAP, NetworkX, Graph-Tool, . . .
Data: up to 100 social network datasets of different sizes
Literature review of suitable comparison measures
Comparison based on time and memory usage for certain criticaloperations (distance computation, component analysis, etc.), . . .
Interested? [email protected] or find me in 157b
Projects — Masterclass — September 16, 2015 5 / 7
Project 3: Patent Network Analysis
Data: over 1 million European patent applications and relatedscientific publications
Question (broad): How can we automatically mine thepatterns in the patent data and scientific publication data,and what do these patterns tell us about how scientificpapers and patents are related?
Network modelling: patent citation networks, scientific citationnetwork and co-authorship networks
Approach: bit of text mining + network analysis techniques:centrality measures, community detection algorithms, etc.
Collaboration with Kenedict Analytics and CWTS Leiden
Interested? [email protected] or find me in 157b
Projects — Masterclass — September 16, 2015 6 / 7