Nadezhda Ozornina researches the methods of unsupervised classification of multimodal social media posts using computational methods. With the help of natural language processing and computer vision, she investigates the application of clustering approaches such as topic modeling and possible biases in the international media context.
With the ongoing datafication of society, multimodal social media posts have become an essential source for automated analyses in communication science. The classification of such content, which combines various modalities such as text and image is gaining significance. This project focuses on the unsupervised classification of multimodal content using computational methods and examines the associated application possibilities and challenges in the communication science. The primary goal is to provide research guidelines for unsupervised classification of multimodal data using topic modeling, which will correspond to the theoretical concept of topic in communication science. Additionally, the analysis explores how these methods can be applied to social media data from different cultural and language contexts and what biases in this process may arise.
Multimodal Communication | Computational Communication Science | Topic Modeling
Academic Staff
Computational Communication Research • Multimodal Communication • Multilingual Text Analysis • Political Communication
The project includes several research projects within LMU. That's because it's great and very interdisciplinary.