π Data analysis
What this page shows & how it works
chanalyse scrapes 4chan board catalogues and the full posts of active threads, then an LLM groups each thread into a story (a specific, news-like development) under a theme (a durable bucket), extracting named entities, a one-line claim, a stance word, and a novelty score (0=evergreen chatter, 1=breaking).
- Story volume = how many threads mention a story per hour. A sharp rise vs its own baseline is what triggers an "article trigger".
- Sentiment = VADER compound score (β1β¦+1) averaged over posts; the LLM stance word adds nuance VADER misses.
- Novelty distinguishes a fresh development from a daily "general" thread.
- Velocity = total threads seen per hour on the board.
- Click any bar, pie slice, theme or story to see the underlying threads.
Recording window loadingβ¦
Story volume over time
What this tells you: how much story-tagged discussion each board carries per day β the overall "how loud is the board" baseline.
Thread velocity
What this tells you: new threads appearing per day, per board β a leading indicator of attention before it shows up as story volume.
Topic velocity (which topics are heating up)
Posts/hour for each top topic in the recent part of the window vs earlier β a concrete per-TOPIC heat signal. β² accelerating, βΌ cooling.
| Trend | Topic | Board | Now (posts/hr) | Accel | Total |
|---|
Average sentiment over time
What this tells you: mood (β1 bearish β¦ +1 bullish) of the conversation per board over time. Watch for boards diverging or a sharp mood flip.
Theme share (click a slice)
What this tells you: what proportion of conversation each broad theme occupies right now. Click a slice to read the underlying threads.
Story mentions over time (top stories)
What this tells you: which specific stories are driving volume and how they rise and fall against each other β the granular view under the theme share.
Novelty distribution
What this tells you: how breaking vs evergreen the conversation is. A right-shift (toward 1.0) means fresh developments are dominating.
Stance words (LLM tone)
What this tells you: the LLM's one-word read on how posters frame stories β nuance VADER sentiment misses (e.g. "skeptical", "hyped").
Top themes (click β threads)
What this tells you: the most-discussed durable themes by thread count. Click a bar to open its threads.
Top stories (click a row β open the story page Β· shows first seen β last seen)
| Story | Board | Mentions | Novelty | Sentiment | First seen | Last seen |
|---|
Board comparison
| Board | Threads | Stories | Avg sentiment | Avg novelty | Deletions |
|---|
π° Article threshold β why a topic becomes an article
Every detected spike (story that broke out above its own baseline) gets an article_score from 0 to 1. When that score reaches the 0.55 threshold, chanalyse auto-writes an article about it. Each story below is a point on the score axis; the red line is the article threshold. Points on/right of the line cleared the bar (β article written / β eligible β pending); points left of it did not (β below threshold). Click a point or row to see its score breakdown β and, if an article exists, to read it.
Scoring formula loadingβ¦
All scored stories (click a row for the breakdown)
| Story | Board | Score | z | Novelty | Entities | Status | Article |
|---|
π Movers & Shakers β what suddenly changed
Entities whose mention rate accelerated most recently vs their own baseline β like a "what's breaking out" index. Brand-new entities are listed separately. Click any row to see the underlying threads.
π Rising (accelerating)
| Entity | Cat | Now | Was | Accel |
|---|
β¨ Newcomers (first appeared)
| Entity | Cat | Mentions | Boards |
|---|
π Trend Index β compare entities over time (0β100)
Mentions per day, normalised to 0β100 (100 = the busiest bucket across the selected entities), the same relative scale Google Trends uses. Add up to 6 entities to overlay them.
| Entity | Cat | Total | Boards | Momentum |
|---|
π Entity Explorer β search companies, cryptos, tickers, political mentions
Every named thing the LLM has tagged. Search by name, filter by category, click a row to chart it in the Trend Index above or open its threads.
| Entity | Category | Mentions | Boards | Last seen |
|---|