SYNTHETIC STAGING · Eight fictional films. Try “Lantern” or “Bell”. No real movie dataset or images.

NightShelf

CURATED BY YOUR TASTE

Discover your
next
horror movie.

Find films with the stories and scares you enjoy. Choose one or more favorites, then let NightShelf explore eight original fictional movies.

Start discovering

01 · YOUR FAVORITES

What kind of night are you planning?

Search for films you already like. Add several to describe your taste more clearly.

02 · YOUR TASTE PROFILE

Selected favorites

03 · DISCOVERY STYLE

How should we explore?

Choose a simple discovery style. Technical details remain available when you want them.

Influence values are score parameters, not percentages or probabilities.

EXPLORE ANOTHER WAY

Browse by horror elements

Choose recognized elements to find movies containing all selected tags. These are synthetic keyword assignments for testing the 45-element taxonomy.

HOW RECOMMENDATIONS WORK

Helpful names, real algorithms

Story Similarity B1
TF-IDF and cosine similarity compare stored movie text.
Horror Element Match B0
Weighted Jaccard compares exact recognized synthetic keyword elements.
Balanced Discovery B2
Adds an alpha-scaled element score to story similarity. Alpha is not a probability.

No keyword match means UNKNOWN, not that an element is absent.