Material for a Computational Social Science (CSS) Workshop hosted by the four Dutch technical universities.

Project maintained by vtraag Hosted on GitHub Pages — Theme by mattgraham

4TU Computational Social Science Seminar

7 April 2017 @ University of Twente

With increasing amounts of data the social sciences have the opportunity to become more computationally oriented, bringing together elements of machine learning and data science with substantive social theories. We will provide a brief introduction to the core of the computational social science. The main focus in this workshop is on two key elements of this new field: natural language processing (NLP) and social networks. We will take a hands-on approach, with interactive lectures that combine theoretical explanations with actual coding. In the group assignments, we will stimulate an integrated approach to both NLP and network analysis.


Time Topic
10:00 - 10:30 Computational Social Science introduction
10:30 - 12:30 Natural Language Processing (NLP) lab
The first part will focus on applying NLP tools. Topics that will
be discussed include sentiment analysis, named entity recognition,
part-of-speech tagging and feature representation.
The second part will focus on NLP to analyze and model
social dynamics. We will discuss computational approaches
to analyze language & social identity (gender, age, location)
and language in interaction (e.g., social relationships,
dynamics in online communities).
12:30 - 13:30 Lunch
13:30 - 15:30 Network lab
We will start with an overview of traditional network analysis,
covering theories of structural holes and weak ties, and relating
them to graph theoretical notions of centrality and paths. We
proceed with some more recent developments, covering diffusion
and community structure. Finally, we discuss social balance
theory, which relates to signed networks, making a connection
to NLP.
15:30 - 17:00 Group assignment
17:00 - 17:30 Wrap up & drinks

[Optionally: dinner]

Prerequisite knowledge

We intend the workhop to be suitable for graduate students from all disciplines. Because of this, we will not assume any particular skillset. Nonetheless, there will be some coding in python, because we aim to make the workshop practically relevant. It is therefore advisable to acquaint yourself with python if you have no experience yet. See also below for what software to install prior to the workshop.


The workshop will be interactive, and we plan to switch between theoretical explanations and practical coding during class. Please bring your own laptop, with all prerequisite software installed. If you encounter any issue installing any software, please do not hesitate to contact one of the instructors.

We will use python 2 during our workshop. We will depend on the scipy stack, and it is recommended to use the Anaconda distribution especially if you are on Windows.

Some of the other packages we will rely on during the workshop are:

All can be installed in Anaconda (or using pip if you prefer).

It is practical to use python in for example ipython or spyder, both are included in the Anaconda distribution. In particular, the exercises and the labs will use jupyter notebooks, so please ensure that you have it installed.

All presentations, notebooks and assignments can be downloaded here.


Participation is free, but registration is required to help us plan the event accordingly: registration form. Registration closed the 5th of April.


The seminar will be held at the INFORM space in the Designlab of the University of Twente.


4TU This is a 4TU Humans & Technology event. The lecturers:

Backgroud reading (optional)





Natural Language Processing

Social media




Natural Language Processing