marc-recommend v1.0.5

@flike/recommend

More information about Flike can be found here.

Table of contents

Installation

Quick Start

Classes

• Recommender

Interfaces

• Recommendation
• RecommendationsResponse

Installation

Install the @flike/recommend package:

npm install @flike/recommend

Quick Guide

1. Install the module as described in Installation.

2. Import the module into your code

   import {Recommender} from '@flike/recommend'

3. Instantiate the Recommender with your API key.

   const recommender = new Recommender(<your API key>);

4. Call the corresponding methods whenever a user interacts with a content item.

   • start when a user starts interacting with a content item.
   • like when a user seems to like a content item. E.g. in the case of a video, call like when the user watched more than 80% of a video.
   • dislike when a user seems to dislike a content item. E.g., in the case of a video, call dislike when they stop watching after watching less than 50% of it.
5. Retrieve recommendations for a user by calling the recommend method.
6. Filter and sort the recommendations if any constraints need to be considered.
7. Display/Use the recommendation in your application in whatever way applicable.

Class: Recommender

Flike Recommender lets you easily log relevant user interactions with content items and recommend content items to users based on their interactions.

Constructors

constructor

• new Recommender(api_key, server_url?, version?)

Parameters

Methods

dislike

▸ dislike(user_id, item_id): Promise<boolean>

Registers a user-started item as 'disliked' by the user. 'Dislike' refers to any action indicating that a user dislikes the content item. E.g., for a video, this could be a user only watching 5% of the video and not finishing it.

Parameters

Returns

Promise<boolean>

Resolves to true if successful. Otherwise, it will throw an exception.

like

▸ like(user_id, item_id): Promise<boolean>

Registers a user-started item as 'liked' by the user. 'Like' refers to any action indicating that a user likes the content item. E.g. for a video, this could be a user watching more than 85% of the video.

Parameters

Returns

Promise<boolean>

Resolves to true if successful. Otherwise, it will throw an exception.

recommend

▸ recommend(user_id, num_items?): Promise<RecommendationsResponse>

Get an array of content items that a user is probable to consume/buy/subscribe/like or similar. Recommendations are sorted by descending probability of a user 'liking' them.

Parameters

Returns

Promise<RecommendationsResponse>

Resolves to a RecommendationResponse if successful. Otherwise, it will throw an exception.

start

▸ start(user_id, item_id, correlation_id?): Promise<boolean>

Registers a user starting to consume/interact with a content item.

Parameters

Returns

Promise<boolean>

Resolves to true if successful. Otherwise, it will throw an exception.

validate

▸ validate(): Promise<boolean>

Validates the connectivity to the API.

Returns

Promise<boolean>

Resolves to true if the connection is successful, false otherwise.

Interface: Recommendation

Recommendation of a content item for a user.

Table of contents

Properties

• item_id
• probability

Properties

item_id

• item_id: string

Unique identifier of the content item being recommended.

probability

• probability: number

Probability of a user 'liking' the recommended item.

Interface: RecommendationsResponse

Table of contents

Properties

• correlation_id
• items

Properties

correlation_id

• correlation_id: string

The unique identifier of this recommendation.

items

• items: Recommendation[]

Recommendations for a user.