Recombee API Client

A javascript library for easy use of the Recombee recommendation API.

It is intended for usage in browsers and other client side integrations (such as in React Native / NativeScript mobile apps). For Node.js SDK please see this repository.

Documentation of the API can be found at docs.recombee.com.

Installation

The library is UMD compatible.

<script> tag

You can download recombee-api-client.min.js and host it at your site, or use a CDN such as jsDelivr CDN:

1<script src="https://cdn.jsdelivr.net/gh/recombee/js-api-client@5.0.2/dist/recombee-api-client.min.js"></script>

and use the global recombee object.

Package managers

1npm install recombee-js-api-client
2# or
3yarn add recombee-js-api-client
4# or
5pnpm add recombee-js-api-client
6# or
7bun add recombee-js-api-client

and import into your code using

1import recombee from 'recombee-js-api-client'
2// or
3const recombee = require('recombee-js-api-client');

The library ships with types, so you should get autocomplete in your IDE out of the box. If you're using TypeScript, it should recognize these correctly and warn you about any type errors.

How to use

This library allows you to request recommendations and send interactions between users and items (views, bookmarks, purchases ...) to Recombee. It uses the public token for authentication.

It is intentionally not possible to change the item catalog (properties of items) with the public token, so you should use one of the following ways to send it to Recombee:

• Use one of the server-side SDKs (Node.js, PHP, Java...). The synchronization can done for example by a peridodically run script. See this section for more details.
• Set a catalog feed at Recombee web admin.

Sending interactions

1// Initialize client with name of your database and PUBLIC token
2const client = new recombee.ApiClient('name-of-your-db', '...db-public-token...', {region: 'us-west'});
3
4// Interactions take the ID of the user and the ID of the item
5client.send(new recombee.AddBookmark('user-13434', 'item-256'));
6client.send(new recombee.AddCartAddition('user-4395', 'item-129'));
7client.send(new recombee.AddDetailView('user-9318', 'item-108'));
8client.send(new recombee.AddPurchase('user-7499', 'item-750'));
9client.send(new recombee.AddRating('user-3967', 'item-365', 0.5));
10client.send(new recombee.SetViewPortion('user-4289', 'item-487', 0.3));

Requesting recommendations

You can recommend items to user, recommend items to item or even recommend Item Segments such as categories, genres or artists.

It is possible to use Promises or callbacks.

Promise

1// Get 5 recommendations related to 'item-365' viewed by 'user-13434'
2const response = await client.send(
3	new recombee.RecommendItemsToItem("item-356", "user-13434", 5)
4);
5// or
6client
7	.send(new recombee.RecommendItemsToItem("item-356", "user-13434", 5))
8	.then(function (res) {
9		console.log(res.recomms);
10	})
11	.catch(function (error) {
12		console.log(error);
13		// use fallback...
14	});

Callback

Callback function take two parameters:

• err - null if request succeeds or Error object
• res - object containg reply from Recombee

1const callback = function (err, res) {
2	if (err) {
3		console.log(err);
4		// use fallback ...
5		return;
6	}
7	console.log(res.recomms);
8};
9
10// Get 5 recommendations for user-13434
11client.send(new recombee.RecommendItemsToUser("user-13434", 5), callback);

Personalized search

Personalized full-text search is requested in the same way as recommendations.

Promise

1const searchQuery = " ... search query from search field ....";
2const response = await client.send(
3	new recombee.SearchItems("user-13434", searchQuery, 5)
4);
5// or
6client
7	.send(new recombee.SearchItems("user-13434", searchQuery, 5))
8	.then(function (res) {
9		console.log(res.recomms);
10	})
11	.catch(function (error) {
12		console.log(error);
13		// use fallback...
14	});

Callback

1const searchQuery = " ... search query from search field ....";
2client.send(new recombee.SearchItems("user-13434", searchQuery, 5), callback);

Recommend Next Items

Recombee can return items that shall be shown to a user as next recommendations when the user e.g. scrolls the page down (infinite scroll) or goes to the next page. See Recommend next items for more info.

1const initialRecomms = await client.send(
2	new recombee.RecommendItemsToUser("user-13434", 5)
3);
4// Get next 3 recommended items as user-13434 is scrolling the page down
5const nextRecomms = await client.send(
6	new recombee.RecommendNextItems(initialRecomms.recommId, 3)
7	// notice we're using recommId from previous request ^
8);

Optional parameters

Recommendation requests accept various optional parameters (see the docs). Following example shows some of them:

1client.send(new recombee.RecommendItemsToUser('user-13434', 5,
2  {
3    scenario: 'homepage', // Label particular usage. You can assign various settings
4                          // for each scenario in the Admin UI (https://admin.recombee.com/).
5    returnProperties: true, // Return properties of the recommended items
6    includedProperties: ['title', 'img_url', 'url', 'price'], // Use these properties to show
7                                                              // the recommended items to user
8    filter: "'title' != null AND 'availability' == \"in stock\""
9                                                     // Recommend only items with filled title
10                                                     // which are in stock
11  }
12), callback);
13
14

Integration Example

1. Create instant account at recombee.com

2. Upload items catalog

You can use a script or set a product feed at Recombee web admin. We will set following sample Google Merchant product feed: product_feed_sample.xml. You will see the items in web interface after the feed is processed.

3. Use client to send interaction and get recommendations

Let's assume we want to show recommendations at product page of pants product-270 to user with id user-1539. The following HTML+js sample send the detail view of the product by the user and request 3 related items from Recombee:

1<!DOCTYPE html>
2<html lang="en">
3<head>
4    <link rel="stylesheet" href="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.7/css/bootstrap.min.css">
5</head>
6<body>
7    <div class="container">
8        <div class="row">
9            <h1>Related products</h1>
10            <div class="col-md-12">
11                <div class="row" id='relatedProducts'>
12                </div>
13            </div>
14        </div>
15    </div>
16
17    <script src="https://cdn.jsdelivr.net/gh/recombee/js-api-client@5.0.2/dist/recombee-api-client.min.js"></script>
18
19    <script type="text/javascript">
20
21    // A simple function for rendering a box with recommended product
22    function showProduct(title, description, link, imageLink, price){
23        return  [
24          '<div class="col-md-4 text-center col-sm-6 col-xs-6">',
25          '    <div class="thumbnail product-box" style="min-height:300px">',
26          '        <img src="' + imageLink +'" alt="" />',
27          '        <div class="caption">',
28          '            <h3><a href="' + link +'">' + title + '</a></h3>',
29          '            <p>Price : <strong>$ ' + price + '</strong>  </p>',
30          '            <p>' + description+ '</p>',
31          '            <a href="' + link +'" class="btn btn-primary" role="button">See Details</a></p>',
32          '        </div>',
33          '    </div>',
34          '</div>'
35        ].join("\n")
36    }
37
38    // Initialize client
39    const client = new recombee.ApiClient('js-client-example', 'dXx2Jw4VkkYQP1XU4JwBAqGezs8BNzwhogGIRjDHJi39Yj3i0tWyIZ0IhKKw5Ln7', {region: 'eu-west'});
40
41    const itemId = 'product-270';
42    const userId = 'user-1539'
43
44    // Send detail view
45    client.send(new recombee.AddDetailView(userId, itemId));
46
47    // Request recommended items
48    client.send(new recombee.RecommendItemsToItem(itemId, userId, 3,
49      {
50        returnProperties: true,
51        includedProperties: ['title', 'description', 'link', 'image_link', 'price'],
52        filter: "'title' != null AND 'availability' == \"in stock\"",
53        scenario: 'related_items'
54      }),
55      (err, resp) => {
56        if(err) {
57          console.log("Could not load recomms: ", err);
58          return;
59        }
60        // Show recommendations
61        const recomms_html = resp.recomms.map(r => r.values).
62                    map(vals => showProduct(vals['title'], vals['description'],
63                        vals['link'], vals['image_link'], vals['price']));
64        document.getElementById("relatedProducts").innerHTML = recomms_html.join("\n");
65      }
66    );
67    </script>
68
69</body>
70</html>

You should see something like this:

Please notice how the properties returned by returnProperties & includedProperties were used to show titles, images, descriptions and URLs.

Remark on user identification

In order to achieve personalization, you need a unique identifier for each user. An easy way can be using Google Analytics for this purpose. The example then becomes:

1ga('create', 'UA-XXXXX-Y', 'auto'); // Create a tracker if you don't have one
2                                    // Replace the UA-XXXXX-Y with your UA code from Google Analytics.
3
4const client = new recombee.ApiClient('js-client-example', 'dXx2Jw4VkkYQP1XU4JwBAqGezs8BNzwhogGIRjDHJi39Yj3i0tWyIZ0IhKKw5Ln7');
5
6ga(function(tracker) {
7  const userId = tracker.get('clientId'); // Get id from GA
8
9  client.send(new recombee.RecommendItemsToUser(userId, 3,
10    {
11      returnProperties: true,
12      includedProperties: ['title', 'description', 'link', 'image_link', 'price'],
13      filter: "'title' != null AND 'availability' == \"in stock\"",
14      scenario: 'homepage'
15    }),
16    (err, resp) => { ... }
17  );
18});
19

This time RecommendItemsToUser is used - it can be used for example at your homepage.