Skip to content

Multi-Model Data Model

Uni combines a property graph with vectors and JSON documents. You can model entities and relationships, attach embeddings, and query structured or semi-structured data in one place.

What It Provides

  • Labeled vertices and typed, directed edges.
  • Typed properties plus flexible JSON overflow for ad-hoc fields.
  • Vector properties for similarity search.

Example

use uni_db::{DataType, Uni};

# async fn demo() -> Result<(), uni_db::UniError> {
let db = Uni::open("./my_db").build().await?;

db.schema()
    .label("Person")
        .property("name", DataType::String)
        .property("age", DataType::Int64)
        .vector("embedding", 384)
    .edge_type("KNOWS", &["Person"], &["Person"])
        .property("since", DataType::Date)
    .apply()
    .await?;

db.execute("CREATE (:Person {name: 'Ada', age: 31, city: 'London'})")
    .await?;
# Ok(())
# }

import uni_db

db = uni_db.Database("./my_db")

db.schema() \
    .label("Person") \
        .property("name", "string") \
        .property("age", "int64") \
        .vector("embedding", 384) \
        .done() \
    .edge_type("KNOWS", ["Person"], ["Person"]) \
        .property("since", "date") \
        .done() \
    .apply()

db.execute("CREATE (:Person {name: 'Ada', age: 31, city: 'London'})")

Use Cases

  • Knowledge graphs with semantic search.
  • Product graphs with text metadata and embeddings.
  • Mixed graph + document workloads in one engine.

When To Use

Use Uni's multi-model design when your application needs relationships, embeddings, and document-style properties in one consistent database.