400011 SE Advanced quantitative text analysis (2024W)

Lehrende: Fabienne Lind, Petro Tolochko

Facing the massive volumes of text data that are available in digital format and valuing their potential, over recent years social scientists have increasingly turned to methods that rely on the support of computer power, so- called automated text analysis methods. The text-as-data methods are used to draw reproducible and valid inferences or meanings from documents. As an enhancement of the more classical manual methods of content analysis, automated methods of text analysis are becoming prevalent in disciplines that are overall increasingly computationally oriented.
This course is aimed at people with some knowledge of automated text analysis who want to use this method in their PhD and/or want to deepen their expertise in the matter.
The course covers topics related to data collection, data processing, quality control, and the critical interpretation of results.
We cover the following topics:
What kind of questions can be answered with automated text analysis
Recap Python
Text Data Collection and Selection
Pre-processing
Causal Inference in Text
Regular Expressions and Classification with Dictionaries
Machine Learning and Classification
Neural Networks and Transformer Models
Topic modeling/k-means
Text analysis and network analysis
Multilingual text analysis
Validation
Ethics and Data Security
Critical reflection on the methods
All topics are introduced with a lecture type approach and then illustrated with practical examples. In the lecture type part, we show applied cases, explain the motivation for specific methods, illustrate the mathematical background and discuss the pros and cons of different methods in respect to different research goals. The practical part consists of guided coding sessions, where we work together through prepared code, and small coding challenges (no grades), which are worked on alone or in groups.
The guided coding sessions during class will predominantly feature examples in Python. To fully engage with the course content, it's essential to possess the ability to create and manipulate vectors, data frames, and list objects, as well as to load tabular data files such as CSVs. Proficiency in performing fundamental operations like subsetting/filtering, indexing, and creating/changing columns within data frames is also crucial. If you require a refresher or a starting point to bolster your proficiency in these Python basics, please refer to the resources provided below. Additionally, most of our code examples include an equivalent version in R, which you can employ for your personal projects.
Furthermore, when it comes to submitting your final paper, you have the flexibility to choose between Python or R.

Please note: The prerequisite for participation in advanced seminars is the conclusion of the doctoral thesis agreement.