Stability of Topics in Automatic Clustering Algorithms

TODOs

• keep models "exclusive": religion, spirituality, religion AND spirituality
• add .py scripts and modules with final scripts for 10-topic models
• use a common preprocessing pipeline for all models
• create a requirements.txt for a working virtual environment
• write user instructions in the final README

Project Docs

• Project pitch / future presentation slides: https://docs.google.com/presentation/d/1xniCpG-9FYCbaGrvz2yerRiDV8IL5eRf/edit?usp=sharing&ouid=106479919216175443892&rtpof=true&sd=true
• Overleaf project (to be created)

Resources

• The ‘Spiritual’ and the ‘Religious’ in the Twittersphere: A Topic Model and Semantic Map (Fabian Winiger, Gerold Schneider, Janis Goldzycher, David Neuhold, Simon Peng-Keller)

Top2Vec

• Top2Vec: Distributed Representations of Topics: https://arxiv.org/abs/2008.09470

LDA

• Demystifying Topic Modeling Techniques in NLP: https://vijay-choubey.medium.com/demystifying-topic-modeling-techniques-in-nlp-c0d11616a287

Pachinko Allocation

• Mixtures of Hierarchical Topics with Pachinko Allocation: https://mimno.infosci.cornell.edu/papers/icml-hpam.pdf
• Pachinko Allocation: DAG-Structured Mixture Models of Topic Correlations: https://people.cs.umass.edu/~mccallum/papers/pam-icml06.pdf

Mallet

• Antoniak's wrapper: https://github.com/maria-antoniak/little-mallet-wrapper