Teaching

Network Analysis

Exam Description

The exam consists of conducting a study project that applies the knowledge on network analysis acquired during the course.

The exam’s hand-in consists of a report, a 5–6-page PDF document (8–9 pages for projects with more students), detailing the context, the problem/motivation, the data, the measures, and the results of the project that the students performed.

The project is either individual or in a small group, and it is negotiated with the teacher through a project proposal. The project proposal is a 1-page PDF document (2 pages for proposals with more students) that briefly presents the content of the study (context, the problem/motivation, the proposed data, the designed measures, and the expected results of the research project that the student will perform).

To make sure students have enough time to work on their projects, there is a “minimal work period” between the agreement on the project proposal and the first upcoming project delivery deadline. This minimal work period is 20 days before the next deadline for delivering the report. Students can work on their proposals and projects notwithstanding the above constraint. The only effect of this rule is that projects whose proposals were approved after that deadline will be eligible for evaluation starting from the second upcoming project delivery deadline.

Once the student has received the approval on their project proposal, they can deliver their report within the first valid (as per rule above) deadline and the following ones—i.e., once a project proposal has been approved, the student can deliver the project at any valid following deadline.

The evaluation starts after each project delivery deadline and considers all projects delivered before that deadline.

The deadlines for project delivery are 8 in total within an A.Y. and spaced around 45 days apart (except for August’s break).

Project delivery deadlines (23:59 UTC+2) for the A.Y. 22–23:

The official exam sessions are published on AlmaEsami, where the students must register before the delivery of their project reports. Students registered for the exam on AlmaEsami should automatically receive a notification of the grading (which usually takes around two weeks after each deadline). Grades are officially registered one week after the notification from AlmaEsami. Within that week, students who passed the exam but want to re-take it (on the same project proposal), can notify the teacher, who will not register their grade. Students have 1 attempt to re-take the exam, after which the last passing grade will be registered automatically.

Delivery of Project Proposals and 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 student at each stage of the project. In the proposal (and, by extension, in the report) it is fundamental to indicate A) what phenomena will be investigated, B) what measures will be applied, and C) how each measure will be 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.

Moreover, besides “journaling” progress, the report is a useful tool for the students 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 listed below:

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

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