About & FAQ
A living atlas of recorded human dreams
What is Dream Galaxy?
An explorable atlas of recorded human dreams. You write a dream and it finds, among 228 sourced historical dream records, the ones closest in imagery, feeling and story, and places your dream on a star map.
How are similar dreams matched?
The baseline matcher runs in your browser and compares the dream text with each record’s words and editor-assigned themes, images, feelings, settings and story patterns, while limiting over-representation from one source. If you opt into the local semantic model, it adds on-device semantic similarity and re-ranks the results.
Is my dream uploaded?
No. Matching runs entirely in your browser and the text stays in memory; it is only saved in this browser if you choose to.
Is this dream interpretation?
No. It shows similar dreams, where they come from, and how cultural traditions have talked about such images. These are cultural lenses, not a diagnosis.
What does “closer than 97% of the atlas” mean?
It is a rank: this record is closer to your dream than 97% of all records. It does not mean two dreams are 97% the same — the similarity score is not calibrated.
Where do the dreams come from?
Sources include classical Chinese anthologies, Tang and Song poetry, the Bible, Homer and Herodotus, Indian epics and Upanishads, early Freud and Jung, and world literature. Each work was checked as public domain in the jurisdiction listed on its record; every record carries its source, rights evidence and an uncertainty note.
What is local semantic matching and how big is it?
An optional feature: with your consent the browser downloads a multilingual language model once (about 150 MB) to match on meaning, not just words. It runs on your device and can be deleted any time.