ohdearaugustin / FoundationsOfCSS

Github Repository for Foundations of Computational Social Systems

Foundations of Computational Social Systems

David Garcia, 2021

Welcome to the online materials for Foundations of Computational Social Systems.

The widespread use of information and communication technologies in our digital society motivates the study of Computational Social Systems, where humans and machines interact in a way that generates new phenomena and data. These new phenomena require an interdisciplinary approach that builds on digital trace data to study Computational Social Systems at global scales, very high frequencies, and unprecedented levels of depth and resolution.

This course focuses on the fundamentals of a computational approach to study new social systems in our digital society. Students in this course learn how to plan, execute, and interpret complete Data Science projects to address questions about human behavior and emergent phenomena in Computational Social Systems. After this course, students will know how to gather data from social media, search trends, and other online and offline sources, how to process and store that data, and how to combine, analyze, and visualize data to address specific questions. The course makes a special emphasis in interpretation and critique of Data Science in the Social Sciences, aiming at an interdisciplinary approach rooted on the understanding of the power and limitations of online technologies for social interaction.

Who am I?

I am the Professor for Computational Behavioral and Social Sciences the Graz University of Technology, where I lead the Computational Social Science Lab. I am also group leader at the Medical University of Vienna and at the Complexity Science Hub Vienna. My background is Computer Science but I worked my whole career with psychologists, sociologists and physicists to learn new ways to understand human behavior. I got my PhD from ETH Zurich in 2012 and a habilitation in 2018, starting to work as full professor TU Graz in 2020. To learn more about my research, check my publications. I teach this course in collaboration with Dr. Jana Lasser, a postdoctoral reasearcher in the Computational Social Science Lab.

Place and time

The main lecture takes place on Thursdays at 14:15 (sharp) in lecture room i3 in the Inffeldgasse campus of TU Graz. The lecture is followed by one exercise session in the same room. An additional time slot for another exercise group is available on Tuesdays at 13:00 (also sharp) in room HS II at Rechbauerstrasse 12, where the same exercise as the previous week is discussed. Lectures will be streamed and recorded. When attending in person, proof of 3G status is required and seating space is limited depending on varying policies. Seating will be given in a first-come-first-served basis each session and students that do not fit will have to follow the lecture online in another place.

Course Contents

The course is organized in 13 sessions. Each session contains a practical part with exercises for you to apply what you learned. From the third week, the practice session will consist of a discussion of solutions to the exercise corresponding to the previous session. In exercises, you collect your own data and try to answer questions about human behavior and online phenomena. The online materials do not contain the solutions to the exercises, but if you are stuck or want to start from an easier point, in the github folder of the exercise you can find a version of the exercise with hints.

1. Introduction to Computational Social Systems (7.10.2021)
   1.1. Course administration and information
   1.2. Computational Social Systems
   1.3. Social Data Science
   1.4. The parable of Google Flu Trends

• Tutorial session: R crash course part 1 -- Tutorial resources
  (Please install RStudio before the tutorial takes place)

14.10.2021, 5pm: Supplementary lecture at Computer Science Faculty Day (registration required)

2. Search Behavior (21.10.2021)
   2.1. Measuring temporal orientation with Google Trends
   2.2. Google Trends data in R
   2.3. Accessing the World Development Indicators from R
   2.4. Measuring correlation

• Tutorial session: R crash course part 2 -- Tutorial resources

3. Social Trends (28.10.2021)
   3.1. Social Trends: The Simmel Effect
   3.2 Online Social Trends
   3.3. Old Big Data: Baby name trends
   3.4. Linear regression

• Tutorial session: Data wrangling with dplyr -- Tutorial resources
• Week exercise: Future orientation and economic development -- Exercise resources

4. Social Impact (4.11.2021)
   4.1. Social Impact Theory
   4.2. Online Social Influence
   4.3. Bootstrapping

• Week exercise: The Twitter API in R

5. Social Media Text Analysis (11.11.2021)
   5.1. Measuring emotions
   5.2. Unsupervised sentiment analysis
   5.3. Emotions in pagers after 9/11

• Week exercise: Division of impact on Twitter

6. Supervised Sentiment Analysis (18.11.2021)
   6.1. Evaluating sentiment analysis
   6.2. Supervised sentiment analysis
   6.3. Sentiment in social media

• Week exercise: Running unsupervised sentiment analysis in R

7. Social network analysis (25.11.2021)
   7.1. Introduction to social networks
   7.2. The Friendship paradox
   7.3. Social media data bias

• Week exercise: Evaluating sentiment analysis methods

8. Centrality in social networks (2.12.2021)
   8.1. Centrality and importance
   8.2. Limits to degree: Dunbar's number
   8.3. Twitter network data

• Week exercise: Handling network data in R

9. Social resilience of online communities (9.12.2021)
   9.1. The death of social networks
   9.2. Social resilience
   9.3. Coreness centrality

• Week exercise: Swiss politician data on Twitter

10. Online social network structures (16.12.2021)
    10.1. Structural holes and communities
    10.2. Assortativity
    10.3. Permutation tests

• Week exercise: Twitter network analysis

11. Societal issues (13.1.2022)
    11.1 Privacy in the digital society
    11.2 Personalization and discrimination
    11.3 Manipulation

• Solution session for last exercise and project feedback

12. Ethical and scientific issues (20.1.2022)
    12.1 The ethics of social media research
    12.2 Representation in digital traces 12.3 Measurement issues

• Project feedback

13. Project presentations (27.1.2022)

Where to access materials

• Handouts, codes, and data can be found on the Github repository of the course.
• Students at TU Graz can access the course at Teach Center to get additional information, watch videos, and to participate in the forums and quizzes.

Course grading

The assessment for the course is based on the final group research project (max. 4 students). The project grade is a combination of the presentation (50%) and the written report on the project (50%). Extra points (max 20%) can be achieved by delivering two optional exercise solutions: "Division of impact on Twitter" (deadline 17.11.2021) and "Assortativity among Swiss politicians on Twitter" (deadline 12.1.2022).