> ## Documentation Index
> Fetch the complete documentation index at: https://www.meilisearch.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Building recommendations with similar documents

> Use the /similar endpoint to recommend documents that are semantically close to a given item, powering "More like this" and "Related items" features.

The `/similar` endpoint finds documents that are semantically close to a reference document. Once you have configured an embedder, you can use it to build recommendation features such as "More like this", "Related items", or "You might also like".

<Note>
  This guide requires a configured embedder. If you haven't set one up yet, see the [hybrid search getting started](/capabilities/hybrid_search/getting_started) guide.
</Note>

## Create an index with embeddings

Create an index called `movies` and add this <a href="/assets/datasets/movies.json" download="movies.json">`movies.json`</a> dataset to it. If necessary, consult the [getting started](/getting_started/first_project) for more instructions on index creation.

Then configure an OpenAI embedder:

<CodeGroup>
  ```bash theme={null}
  curl \
    -X PATCH 'MEILISEARCH_URL/indexes/movies/settings/embedders' \
    -H 'Content-Type: application/json' \
    -H 'Authorization: Bearer MEILISEARCH_KEY' \
    --data-binary '{
      "movies-text": {
        "source": "openAi",
        "apiKey": "OPENAI_API_KEY",
        "model": "text-embedding-3-small",
        "documentTemplate": "A movie titled {{doc.title}} whose plot is: {{doc.overview}}"
      }
    }'
  ```
</CodeGroup>

Replace `MEILISEARCH_URL`, `MEILISEARCH_KEY`, and `OPENAI_API_KEY` with the corresponding values in your application.

Meilisearch will start generating embeddings for all documents. Use the returned `taskUid` to [track the progress of this task](/capabilities/indexing/tasks_and_batches/async_operations).

## Find a reference document

To recommend similar items, you first need a reference document. This is typically the item a user is currently viewing. For this example, search for "batman":

<CodeGroup>
  ```bash theme={null}
  curl \
    -X POST 'MEILISEARCH_URL/indexes/movies/search' \
    -H 'Content-Type: application/json' \
    -H 'Authorization: Bearer MEILISEARCH_KEY' \
    --data-binary '{
      "q": "batman",
      "hybrid": {
        "semanticRatio": 0.5,
        "embedder": "movies-text"
      }
    }'
  ```
</CodeGroup>

The top result is the movie "Batman" with `id` 192. Use this as the reference document.

## Retrieve similar documents

Pass the reference document's `id` to the [`/similar` endpoint](/reference/api/similar-documents/get-similar-documents-with-post), specifying your embedder:

<CodeGroup>
  ```bash theme={null}
  curl \
    -X POST 'MEILISEARCH_URL/indexes/movies/similar' \
    -H 'Content-Type: application/json' \
    -H 'Authorization: Bearer MEILISEARCH_KEY' \
    --data-binary '{
      "id": 192,
      "embedder": "movies-text"
    }'
  ```
</CodeGroup>

Meilisearch returns the 20 documents most similar to the reference movie. Display these as recommendations to your users.

## Use cases

* **E-commerce**: "Customers also viewed" or "Similar products" on product detail pages
* **Content platforms**: "Related articles" or "More like this" alongside the current content
* **Media streaming**: "Because you watched X" recommendations based on the current title

## Next steps

<CardGroup cols={2}>
  <Card title="Choose an embedder" href="/capabilities/hybrid_search/how_to/choose_an_embedder">
    Compare embedding providers and pick the right one for your use case.
  </Card>

  <Card title="Personalized search" href="/capabilities/personalization/getting_started/personalized_search">
    Combine recommendations with personalized search for a tailored experience.
  </Card>

  <Card title="Similar documents API reference" href="/reference/api/similar-documents/get-similar-documents-with-post">
    Full API reference for the /similar endpoint.
  </Card>
</CardGroup>