Week 6 Deliverable: Topic Modeling & LDA

Course: Topical Reading: Digital Humanities (BA3 Korean Studies)
Due: Friday, November 28, 2025 by 17:00

Objective

Build a simple LDA model in Orange, explore topics with LDAvis, and interpret what topic modeling reveals about your corpus.

Tasks

1. Build an LDA Model in Orange

• Load nikh_corpus.csv (in /data/nikh)
• Preprocess (tokenize, stopwords; noun-only script optional)
• Add Topic Modeling
• Choose a reasonable number of topics (4-6 should be fine)
• Connect to Data Table to view topic–document probabilities

Export: week06_flow.svg

2. Explore Topics with LDAvis

• Connect Topic Modeling → LDAvis
• Select two topics
• Use the λ slider (0.2–0.35)
• Take screenshots of the word relevance view

Export: topic_X_ldavis.png

3. Interpret Topic–Document Mixtures

• Pick three documents in the Data Table
• For each:
  • Which topic has highest probability?
  • Any topics with probability 0?
  • Briefly interpret what this suggests

Write 2–4 sentences per document in your README.md for this week (in addition the the short relections below).

4. Short Reflection

In your README.md (5–7 sentences):

• What themes did your topics represent?
• How did λ change interpretation?
• One strength and one limitation of LDA.
• What might you do next in your analysis pipeline?

5. Submit Your .ows File

Upload your Orange workflow: week06.ows

Deliverables (folder structure)

/week06/

├── week06_flow.svg
├── topic_1_ldavis.png
├── topic_2_ldavis.png
├── week06.ows
└── README.md