Teaching

Network Analysis

Exam Description

The exam regards the conduction of a group study project that applies the knowledge acquired attending the course. The exam’s hand-in consists of a report in PDF format of at most 5000 words — approximately, these correspond to 30k characters and 9–10 pages. Reports that exceed the word count will be penalised. The report shall detail the context, the problem/motivation, the data, the measures applied, and the results obtained (along with their qualitative interpretation). Once a group assembled an outline of the project they want to conduct, they can contact the teacher for feedback with a short summary of the outline (using the address below).

The project is performed by a group of 3–5 students, who can deliver their report within any of the exam session’ deadlines reported below. The evaluation of the projects starts after each exam session’s deadline and considers all projects delivered before that deadline. The exam sessions are 7 within an academic year (AY) and spaced around 50 days apart (excluding the summer break).

Project delivery (exam session) deadlines for the AY 24–25 (at 23:59 UTC+2 of each date):

When delivering their project, all members must also register to the available exam session on AlmaEsami. After the evaluation, the students will receive the notification of the grading on AlmaEsami (usually around two weeks after each exam session’s deadline).

Grades are officially registered one week after the notification from AlmaEsami. Within that week, groups which passed the exam but want to re-take it (fixing possible shortcomings of the delivered project), can notify the teacher via email, who will exclude their grade from the registration. Groups have 1 attempt to re-take the exam, after which the last passing grade will be registered automatically.

Delivery of Project Reports

All deliveries happen via email at the address saverio.giallorenzo[at]gmail.com and must contain in the subject the prefix [Network Analysis].

Guidelines and Report Evaluation

The report is the artefact evaluated for the exam, but students are invited to see it as a support tool when conducting their projects. Indeed, if done step-by-step during the project, reporting is a helpful tool to clarify and document the decisions taken by the group at each stage of the project. In the report, it is fundamental to indicate A) what phenomena are investigated, B) what measures are applied, and C) how each measure is interpreted to explain the related phenomenon. Simplified example: A) we want to study which nodes are in the core of the studied network, B) we apply the k-core measure, C) k-core interprets core-periphery as the connected set of nodes where each is joined to at least k of the others, hence, we expect to find core nodes in the set with the highest k-value.

Besides “journaling” progress, the report is a useful tool for the group to check that they considered and reported on all the important aspects of their project study. The clear presentation/explanation of those aspects also forms the weighted partition that guides the grading (and indicates a ratio for distributing the contents of the report), as shown below:

Context Problem(s) Motivation	Dataset(s) and source(s) Validity and reliability Applied measure/s and its/their justification, possible hypotheses and tests, results	Interpretation of the overall results Critique (brief)
20%	60%	20%

Lecture Notes

Introduction to Network Analysis	Slides
Some Logistics (Reading Groups)	Slides
Research Design (also, How to read a research paper)	Slides
Paper Presentation: An Information Flow Model for Conflict and Fission in Small Groups	Slides
The Mathematics of Networks	Slides
Data Collection and Data Management	Slides
Measures and Metrics, Nodes	Slides
Measures and Metrics, Networks	Slides
Testing Hypotheses	Slides

Lecture Schedule

Bibliography

Main references:

• Lecture Notes, Research Papers, and datasets provided by the teacher.
• Newman, M. (2018). Networks. Oxford university press. Recommended chapters: 1 (+ one or more among 2, 3, 4, 5), 6, 7, 10, 11.
• Borgatti, S. P., Everett, M. G., & Johnson, J. C. (2018). Analyzing social networks. Sage. Recommended chapters: 1–8.

Additional references:

• Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications (Vol. 8). Cambridge university press.
• Watts, D. J. (2004). Six degrees: The science of a connected age. WW Norton & Company.
• Koch, R., & Lockwood, G. (2011). Superconnect: Harnessing the power of networks and the strength of weak links. Random House Digital, Inc..
• Barabasi, A. L. (2014). Linked-how Everything is Connected to Everything Else and what it Means F (pp. 1-1). Perseus Books Group.

Dataset examples (UCINET format)

• simple-graph matrix
• simple-graph nodelist
• simple-graph, weighted edgelist
• bimodal nodelist

NetworkX dataset and analysis examples

• adjacency list network
• edge list network
• analysis example in Jupyter Notebook format.
• analysis example in python format—needs the networkx, matplotlib, and scipy libraries installed, e.g., via the command pip install networkx matplotlib scipy ).

Resources and links

Data refinement

• OpenRefine is a powerful tool for working with messy data: cleaning it; transforming it from one format into another; and extending it with web services and external data.

Network Analysis

• UCINET is a comprehensive package for the analysis of social network data that includes many network measures and permutation-based statistical analysis (Windows only).
  • UCINET Getting Started Tutorial
  • UCINET Metrics and Visualisation Tutorial
  • UCINET Download link
  • UCINET Installation Guide
  • UCINET Guide
  • UCINET Hanneman & Riddle's Tutorial
  • UCINET Datasets
• Gephi is a network visualisation and exploration software for network analysis. Gephi equips a plugin system that lets users add/include new functionalities—transformations, measures, and layouts (Windows, macOS, Linux).
  • Gephi Download
  • Gephi Wiki
• NetworkX is a Python package for the creation, manipulation, and study of the structure, dynamics, and functions of complex networks (Windows, macOS, Linux).
• Memgraph is an in-memory graph database designed to maximise the performance of querying network data—users can connect several machines to increase both the available storage space and the performance of the queries. Memgraph natively provides some network measures and it also supports the creation and manipulation of network data with NetworkX.

Datasets

Dataset repositories available online, e.g., useful to draw inspiration for projects:

• UCINET Datasets
• Stanford Large Network Dataset Collection
• Alex Arenas Website
• Pajeck Datasets
• Statistical Analysis of Network Data
• Mark Newman's Network Data
• Community Resource for Archiving Wireless Data
• Foundations of Data and Visual Analytics
• NetWiki Datasets
• KDnuggets Datasets
• Awesome Public Datasets
• The Colorado Index of Complex Networks
• The Moviegalaxies Dataset

Reading Group Paper

An Information Flow Model for Conflict and Fission in Small Groups