zantetsu-vecdb

Semantic title mapping via embedded vector search.

Features

• HNSW Index: Approximate nearest neighbor search for fast semantic matching
• Hybrid Search: Combines semantic similarity with lexical matching
• SQLite Storage: Persistent vector index with embedded database
• Title Normalization: Maps extracted titles to canonical anime names

Usage

use zantetsu_vecdb::VecDb;

let db = VecDb::open("anime_vectors.db").unwrap();

// Search by semantic similarity
let results = db.search("spy x family", 5).unwrap();
for (title, score) in &results {
    println!("{}: {:.2}", title, score);
}

License

MIT