Extract the core soul of any PDF — in seconds.
Aero Core scans every character of your PDF — even 500+ pages — and uses a hybrid TextRank + TF‑IDF + cosine MMR algorithm (PhD‑level NLP math) to surface only the sentences that truly matter. Your file never leaves your device.
Clarity of mind is not the absence of thought — it is the precise selection of which thoughts deserve your attention.
Awareness collapses noise into signal; the observer changes the observed simply by paying attention.
★ core_score: 0.94 topic: cognition
Drop a PDF. Get the essence.
Tune the depth, pick a focus topic (optional), and let the algorithm rank every sentence by importance, novelty and centrality.
Hybrid TextRank · TF‑IDF · Cosine MMR
A PhD‑grade pipeline that combines graph theory, information retrieval and convex optimization to identify the true “core” of a document.
Tokenize & Clean
Pages → sentences → tokens. Stopwords removed, lemmas normalised, headers/footers/page numbers stripped.
TF‑IDF Vectors
Each sentence becomes a sparse vector: w(t,s) = tf(t,s) · log(N / df(t)). Rare‑but‑topical terms get weight.
Similarity Graph
Cosine similarity between every sentence pair builds a weighted graph G = (V, E, w).
TextRank (PageRank)
Iterate PR(v) = (1‑d) + d · Σ PR(u)·w(u,v)/Σw(u,*) until convergence (ε < 1e‑5).
MMR Re‑ranking
Maximal Marginal Relevance: argmax [ λ·Sim(s,Q) − (1‑λ)·max Sim(s,S*) ] — kills redundancy.
Core Selection
Top‑K sentences ordered by original position → coherent, non‑repetitive, focus‑aware core content.
Clarity of Mind — what 1% of humans truly understand
Aero Core isn’t just a tool. It’s built on a simple, powerful idea: your attention shapes your reality. The mind that learns to select signal over noise — in PDFs, in life, in emotion — never falls behind.
The Observer Effect (Daily Life)
What you observe, you amplify. Anxiety watched grows; awareness watched grows too. Neuroscience: the prefrontal cortex strengthens whatever circuit you repeatedly attend to (Hebbian learning — “cells that fire together, wire together”).
Signal vs Noise
A 566‑page book has maybe 30 sentences that change you. The same is true of a day, a relationship, a career. Train your filter — the rest is noise. This app is a literal demonstration of that filter.
Emotion ↔ Attention Loop
Emotion narrows attention (amygdala hijack); narrowed attention reinforces emotion. Break the loop by widening focus — name 5 things you see, 4 you hear. The loop dissolves in < 90 seconds (Jill Bolte Taylor, neuroanatomist).
Predictive Brain
The brain is a prediction machine (Karl Friston, Free‑Energy Principle). When you know the pattern in advance, suffering drops. That’s why this tool surfaces the pattern before you read all 566 pages.
Detect Before It Hits
Recognising an emotional theme early = prefrontal activation before limbic flood. Practical drill: the moment you notice a familiar “loop”, label it (“this is the comparison loop”). Labelling alone reduces amygdala activity by ~50% (Lieberman, UCLA, 2007).
Living the Theory
Read less, extract more. Feel less, observe more. Speak less, signal more. The same algorithm that compresses a PDF can compress a life — keep only what carries weight, drop the rest with grace.
Built different. Built for you.
Everything runs in your browser. No server. No tracking. No upload. Just math, taste, and clarity.
Handles 500+ pages
Streaming page‑by‑page text extraction with progress feedback — works on huge books.
100% Private
PDF.js parses locally. Nothing ever leaves your device.
Focus‑aware
Type a topic — extractor biases toward sentences semantically close to your query.
MMR Diversity
Tune λ to balance importance vs novelty — kill repetitive paragraphs.
Export anywhere
One click → copy, .txt, or markdown. Drop straight into Notion, Obsidian, your blog.
Aero Aesthetic
Glassmorphism, 3D tilt, nebula parallax, neon accents — premium without noise.
Comments
Post a Comment