Tigris Vector Database Client Library for TypeScript

Getting Started

Tigris makes it easy to build AI applications with vector embeddings. It is a fully managed cloud-native database that allows you store and index documents and vector embeddings for fast and scalable vector search.

1. Install the client library

npm install @tigrisdata/vector

2. Fetch Tigris API credentials

You can sign up for a free Tigris account here.

Once you have signed up for the Tigris account, create a new project called vectordemo. Next, make a note of the clientId and clientSecret, which you can get from the Application Keys section of the project.

3. Create the Vector Database client

import { VectorDocumentStore } from "@tigrisdata/vector";

const vectorDocStore = new VectorDocumentStore({
  connection: {
    serverUrl: "api.preview.tigrisdata.cloud",
    projectName: "vectordemo",
    clientId: "clientId_here",
    clientSecret: "clientSecret_here",
  },
  indexName: "my_index",
  numDimensions: 3,
});

Here, we have created a new VectorDocumentStore instance that connects to the Tigris Vector Database. The indexName is the name of the index that will store your embeddings, documents, and any additional metadata. You can use any name you like for the index. The numDimensions is the number of dimensions of the vector embeddings.

4. Add documents to the index

await vectorDocStore.addDocumentsWithVectors({
  ids: ["id1", "id2"],
  embeddings: [
    [1.2, 2.3, 4.5],
    [6.7, 8.2, 9.2],
  ],
  documents: [
    {
      content: "This is a document",
      metadata: {
        title: "Document 1",
      },
    },
    {
      content: "This is another document",
      metadata: {
        title: "Document 2",
      },
    },
  ],
});

Here, we have added two documents to the index. The ids are the unique identifiers for the documents. The embeddings are the vector embeddings for the documents. The documents are the actual documents that you want to store in the index. The content is the text content of the document. The metadata is any additional metadata that you want to store for the document.

5. Query the index

You can query the index for the top k most similar documents to a given vector. It's that simple!

const results = await vectorDocStore.similaritySearchVectorWithScore({
  query: [1.0, 2.1, 3.2],
  k: 10,
});
console.log(JSON.stringify(results, null, 2));

6. Enhance the query with attribute filtering

You can enhance the query with attribute filtering. For example, you can filter the results by a given metadata field.

const results2 = await vectorDocStore.similaritySearchVectorWithScore({
  query: [1.0, 2.1, 3.2],
  k: 10,
  filter: {
    "metadata.title": "Document 1",
  },
});
console.log(JSON.stringify(results2, null, 2));

Next Steps

Tigris Vector Database is simple enough to get started with, but powerful enough to build real-world applications.

Tigris provides the following features:

• High performance: Tigris is built from the ground up to provide fast vector search. It is optimized for low-latency vector search and high throughput.
• Real-time: Tigris provides real-time updates to the index. You can add, update, and delete documents and embeddings in real-time.
• Hybrid search: Tigris supports hybrid search, which allows you to combine vector search with attribute filtering for more relevant search results.
• Fully managed: Tigris is a fully managed cloud-native database. You can get started with Tigris in minutes and scale as needed with ease.

Check out the docs at https://www.tigrisdata.com/docs/concepts/vector-search/.

Contributing

Building

# clean the dev env

npm run clean

# build

npm run build

# test

npm run test

# lint

npm run lint

Code Quality

1. Linting

The coding style rules are defined by Prettier and enforced by Eslint

2. Git Hooks

We use pre-commit to automatically setup and run git hooks.

Install the pre-commit hooks as follows:

pre-commit install

On every git commit we check the code quality using prettier and eslint.

License

This software is licensed under the Apache 2.0.

Contributors

Thanks to all the people who contributed!