Computational Analysis of Digital Communication
Inhoudsopgave
Lecture 1 – introduction ................................................................................................................................. 3
1.1 What is computational social science? ....................................................................................................... 3
1.2 Computational communication Science ..................................................................................................... 5
1.3 The trouble with R ....................................................................................................................................... 7
Lecture 2 – Data Wrangling and Data Visualization ....................................................................................... 10
2.1 From Raw Data to Tidy Data ..................................................................................................................... 10
2.2 Why we should look at data ..................................................................................................................... 12
2.3 examples for different methods ................................................................................................................ 14
2.4 Data visualization ...................................................................................................................................... 19
2.5 Perception and visual decoding ................................................................................................................ 20
2.6 Principles of data visualization ................................................................................................................. 22
2.7 Creating visualization in R ......................................................................................................................... 26
Lecture 3 – Basics of Automatic Text Analysis and Dictionary Approaches .................................................... 28
3.1 Text as Data ............................................................................................................................................... 28
3.2 Automatic Text Analysis ............................................................................................................................ 29
3.3 Deductive approaches: dictionary/lexical analysis .................................................................................. 34
3.4 Reliability and validity............................................................................................................................... 37
3.5 Examples in the literature ......................................................................................................................... 39
Lecture 4 – Machine Learning: Supervised Text Classification ....................................................................... 44
4.1 What is machine learning? ....................................................................................................................... 44
4.2 Supervised text classification .................................................................................................................... 46
4.3 Principles of supervised text classification ............................................................................................... 48
4.4 Validation .................................................................................................................................................. 55
4.5 Examples from the literature .................................................................................................................... 60
4.6 Outlook and Conclusion ............................................................................................................................ 63
5. Required readings summary ..................................................................................................................... 67
5.1 Reading from lecture 1 - When Communication Meets Computation: Opportunities, Challenges, and
Pitfalls in Computational Communication Science - ....................................................................................... 67
5.2 Reading 1 from Lecture 2 - Political migration discourses on social media: a comparative perspective
on visibility and sentiment across political Facebook accounts in Europe .................................................... 71
5.3 Reading 2 from Lecture 2 - Sourcing Pandemic News: A Cross-National Computational Analysis of
Mainstream Media Coverage of COVID-19 on Facebook, Twitter, and Instagram ....................................... 76
1
,5.4 Reading 1 from Lecture 4 - The Validity of Sentiment Analysis: Comparing Manual Annotation, Crowd-
Coding, Dictionary Approaches, and Machine Learning Algorithms ............................................................. 80
5.5 Reading 2 from Lecture 4 - Uncivil and personal? Comparing patterns of incivility in comments on the
Facebook pages of news outlets ..................................................................................................................... 85
2
,Lecture 1 – introduction
1.1 What is computational social science?
Study Blumenstock about wealth and poverty in Rwanda.
Study with 1000 customers of the largest mobile phone provider
Collected demographics, social & economic characteristics + access to complete call records
from 1.5 million people
Combining both data sources they used the survey data to train a machine learning model to
predict a person’s wealth based on their call records
à this created a high-resolution map of the geographic distribution of wealth in Rwanda.
Side effect being that the results were hard to validate since there was no comparable
estimates for all geographic areas in Rwanda
Crime prediction by Thompson 2010
Police department have started using a system called CRUSH
It evaluates patterns of past crime reports, offender behavior profiles or weather forecasts
This combo of data is used to predict potential hot spots + allocate resources to areas where
particular crimes are most likely to occur
Computational social science:
- Field of Social Science that uses algorithmic tools & large data to understand human
behavior
- Computational methods as “microscope”: methods are not the goal, but contribute
to theoretical development
- Complements rather than replaces traditional methodologies
Why is this important now?:
- In the past collecting data was expensive
- In the digital age, the behaviors of billions of people were recorded, restored and
therefore analyzable
- A digital record of your behavior is created and stored any time you make a call/pay
for something with creditcard
- Data are byproduct of people’s everyday actions and are therefore called digital
traces
- Large scale records are called big data
3
, Typical computational research strategies
1. Counting things à researchers can “count” more than ever
2. Forecasting and nowcasting à big data allow for more accurate predictions both in
present & in future
3. Approximating experiments à computational methods provide opportunities to
conduct “natural experiments”
Advantages of Computational methods:
- Actual behavior vs self-report
- Social context vs lab setting
- Small N to large N
Disadvantages of Computational methods:
- Techniques often complicated
- Data often proprietary
- Samples often biased
- Insufficient metadata
4
Inhoudsopgave
Lecture 1 – introduction ................................................................................................................................. 3
1.1 What is computational social science? ....................................................................................................... 3
1.2 Computational communication Science ..................................................................................................... 5
1.3 The trouble with R ....................................................................................................................................... 7
Lecture 2 – Data Wrangling and Data Visualization ....................................................................................... 10
2.1 From Raw Data to Tidy Data ..................................................................................................................... 10
2.2 Why we should look at data ..................................................................................................................... 12
2.3 examples for different methods ................................................................................................................ 14
2.4 Data visualization ...................................................................................................................................... 19
2.5 Perception and visual decoding ................................................................................................................ 20
2.6 Principles of data visualization ................................................................................................................. 22
2.7 Creating visualization in R ......................................................................................................................... 26
Lecture 3 – Basics of Automatic Text Analysis and Dictionary Approaches .................................................... 28
3.1 Text as Data ............................................................................................................................................... 28
3.2 Automatic Text Analysis ............................................................................................................................ 29
3.3 Deductive approaches: dictionary/lexical analysis .................................................................................. 34
3.4 Reliability and validity............................................................................................................................... 37
3.5 Examples in the literature ......................................................................................................................... 39
Lecture 4 – Machine Learning: Supervised Text Classification ....................................................................... 44
4.1 What is machine learning? ....................................................................................................................... 44
4.2 Supervised text classification .................................................................................................................... 46
4.3 Principles of supervised text classification ............................................................................................... 48
4.4 Validation .................................................................................................................................................. 55
4.5 Examples from the literature .................................................................................................................... 60
4.6 Outlook and Conclusion ............................................................................................................................ 63
5. Required readings summary ..................................................................................................................... 67
5.1 Reading from lecture 1 - When Communication Meets Computation: Opportunities, Challenges, and
Pitfalls in Computational Communication Science - ....................................................................................... 67
5.2 Reading 1 from Lecture 2 - Political migration discourses on social media: a comparative perspective
on visibility and sentiment across political Facebook accounts in Europe .................................................... 71
5.3 Reading 2 from Lecture 2 - Sourcing Pandemic News: A Cross-National Computational Analysis of
Mainstream Media Coverage of COVID-19 on Facebook, Twitter, and Instagram ....................................... 76
1
,5.4 Reading 1 from Lecture 4 - The Validity of Sentiment Analysis: Comparing Manual Annotation, Crowd-
Coding, Dictionary Approaches, and Machine Learning Algorithms ............................................................. 80
5.5 Reading 2 from Lecture 4 - Uncivil and personal? Comparing patterns of incivility in comments on the
Facebook pages of news outlets ..................................................................................................................... 85
2
,Lecture 1 – introduction
1.1 What is computational social science?
Study Blumenstock about wealth and poverty in Rwanda.
Study with 1000 customers of the largest mobile phone provider
Collected demographics, social & economic characteristics + access to complete call records
from 1.5 million people
Combining both data sources they used the survey data to train a machine learning model to
predict a person’s wealth based on their call records
à this created a high-resolution map of the geographic distribution of wealth in Rwanda.
Side effect being that the results were hard to validate since there was no comparable
estimates for all geographic areas in Rwanda
Crime prediction by Thompson 2010
Police department have started using a system called CRUSH
It evaluates patterns of past crime reports, offender behavior profiles or weather forecasts
This combo of data is used to predict potential hot spots + allocate resources to areas where
particular crimes are most likely to occur
Computational social science:
- Field of Social Science that uses algorithmic tools & large data to understand human
behavior
- Computational methods as “microscope”: methods are not the goal, but contribute
to theoretical development
- Complements rather than replaces traditional methodologies
Why is this important now?:
- In the past collecting data was expensive
- In the digital age, the behaviors of billions of people were recorded, restored and
therefore analyzable
- A digital record of your behavior is created and stored any time you make a call/pay
for something with creditcard
- Data are byproduct of people’s everyday actions and are therefore called digital
traces
- Large scale records are called big data
3
, Typical computational research strategies
1. Counting things à researchers can “count” more than ever
2. Forecasting and nowcasting à big data allow for more accurate predictions both in
present & in future
3. Approximating experiments à computational methods provide opportunities to
conduct “natural experiments”
Advantages of Computational methods:
- Actual behavior vs self-report
- Social context vs lab setting
- Small N to large N
Disadvantages of Computational methods:
- Techniques often complicated
- Data often proprietary
- Samples often biased
- Insufficient metadata
4
