This repository contains a subset of the code and scripts developed during my Master's thesis, "Topic Dynamics in the DFG Research Grants Program." The analysis focuses on identifying and exploring research themes funded by the Deutsche Forschungsgemeinschaft (DFG) using advanced topic modeling, visualization techniques, and preprocessing methods.
Methods Implemented:
Non-Negative Matrix Factorization (NMF)
Term Frequency-Inverse Document Frequency (TF-IDF)
Coherence Score Calculation for Topic Optimization
Topic Visualization (Histograms and Coherence Score Plots)
Data: Extracted from the GEPRIS platform, containing research project descriptions.
Languages & Tools: Python, Pandas, NumPy, Matplotlib, Gensim, Scikit-learn, and NLTK.
data/: Contains sample datasets (e.g., GEPRIS exports).
scripts/: Python scripts for preprocessing, modeling, and evaluation.
notebooks/: Jupyter notebooks for interactive data analysis and visualization.
models/: Serialized models (e.g., NMF, TF-IDF vectorizer) for reuse.