Topics Discovered by LDA
The LDA model found 4 distinct themes across all 126 articles. Each topic is a cluster of co-occurring words.
What Does Each Leader Focus On?
Topic weights show how much of each leader's coverage falls in each theme area (mean probability across all their documents).
Venn Diagram — Common & Disjoint Topics
The Venn diagram shows which thematic domains are shared by both leaders and which are exclusively owned by one. A topic counts as "shared" if both leaders have a meaningful weight (≥ 0.20) on it.
Reading the Venn Diagram
Yogi Only (Left)
- Governance & infrastructure
- Hindutva & religious identity
- Bulldozer law & order narrative
- Jewar Airport, UP mega projects
- State administration machinery
Shared (Overlap)
- Electoral & campaign rhetoric
- Anti-opposition criticism
- Both address state elections
- Both use party-political framing
Owaisi Only (Right)
- Social justice & rights (47.5%)
- Muslim minority advocacy
- Constitution & legal rights
- Anti-CAA / NRC positions
- Parliamentary floor debate
Which Domains Do They Operate In?
Domain classification maps coverage into recognisable policy/political areas using vocabulary-based scoring.
Discourse Fingerprint — Radar View
The radar chart shows each leader's topic "signature". The shape of the polygon reflects the overall character of their public discourse.
Common Ground & Key Differences
Where do both leaders converge in their messaging? Where do they diverge most sharply?
Data Sources & Methodology
Articles were fetched via NewsAPI over 30 days, covering recent speeches, reports and coverage of both leaders.
NewsAPI · 4 queries per leader · 126 articles total (91 Yogi, 35 Owaisi) · Sources: Indian Express, NDTV, Times of India, The Hindu, The Wire, ANI, PTI
Lowercasing · URL removal · Custom stopword list (140+ terms) · Min word length filter · CountVectorizer (vocab: 612 terms, n-gram 1–2)
sklearn LDA · 4 topics · 25 iterations · batch learning · α=0.1 · β=0.01 · Perplexity ≈ 1314.5 (lower = better fit)