|
Instructors: M.A. Oliver Wieczorek
Event type:
Workshop
Org-unit: Studentische Forschung
Displayed in timetable as:
Methoden-Wks. | A
Hours per week:
1,5
Credits:
1,0
Location:
Campus der Zeppelin Universität
Language of instruction:
German
Min. | Max. participants:
10 | 24
Registration group: Methodenworkshop 2. Semester
Course content:
Techniques from the field of computational linguistics (Natural Language Processing) are increasingly used in the social sciences. Exemplary fields of application are the analysis of political debates, discussions and group phenomena on social media platforms, changes in cultural (interaction) forms, everyday discrimination or development of paradigms in science. Within this relevance framework, two types of methodological approaches have emerged: Topic Modeling and Sentiment Analysis. The former is used to extract topics from large text corpora, whereas the latter is applied to extract sentiment (e.g. fear, anger, joy) from texts.
The course focuses on three methods of analysis, demonstrated with common exercises and programming examples in R and RStudio, and cemented by three homework assignments. More specifically, these are correspondence analysis, lexicon-based sentiment analysis, and latent dirichlet allocation.
Educational objective:
The course aims at preparing you to carry out at least one of the three analysis methods in the course description independently and to apply it to answer your research questions. There are four guiding questions that will be answered throughout the course and will serve as your guide for conducting the analyses:
- How do I prepare my data so that I can use it for quantitative text analysis?
- How do I construct a topic space and how many topics can I (meaningfully) interpret?
- How can I identify and measure positive / negative attitudes and emotions in texts?
- How can I compare these results between groups and different points in time?
- How do I visualize the results and prepare them for my readership?
Further information about the exams:
A total of three homework assignments are to be completed: one on correspondence analysis, one on sentiment analysis, and one on latent dirichlet allocation. These are divided into subtasks based on the questions defined under learning objectives (data collection, data preparation, analysis, interpretation and visualization).
|
