For JS devs: LightRec/lr-ibcf - Open-Source Item-Based Collaborative Filtering in JavaScript

Hello folks! I know we all have seen sections like "Recommended For you", "You may like it" in various platforms. As a developer, I always thought how they made it. After some research I found out that these are made using Machine Learning methods via python.... Wait a bit!

What about the js developers who write their whole code via JS? I am a developer who uses JS a lot and implementing a full machine learning model for every model is tough. That's why I've made "LightRec" (https://lightrec.vercel.app), a platform for recommendation engines and I'll keep publishing new models here.

The first model is "lr-ibcf".

And, its quite easy to use.

1. Installation:

npm install @lightrec/lr-ibcf

2. Import and Initialize

import { RecEngine } from 'lightrec';



// Create a new recommendation engine

const engine = new RecEngine();

3. Pre-Train (Optional)

const interactions = [

    { userId: 'u1', itemId: 'i1', points: 5 },

    { userId: 'u1', itemId: 'i2', points: 3 },

    { userId: 'u2', itemId: 'i1', points: 4 },

]; // Your interaction data



engine.feed(interactions); 

4. Train

const data = {itemId:'i1', userId:'u1', points:3}

engine.act(data.itemId, data.userId, data.points)

5. Recommend

const recommendations = engine.recommendForUser('u1', 5);

console.log(recommendations); // ['i3', 'i4', 'i5', ...]

And, that's it!

Only 5 lines of code and you get Hello folks! I know we all have seen sections like "Recommended For you", "You may like it" in various platforms. As a developer, I always thought how they made it. After some research I found out that these are made using Machine Learning methods via python.... Wait a bit!

What about the js developers who write their whole code via JS? I am a developer who uses JS a lot and implementing a machine learning for every model is tough. That's why I've made "LightRec", a platform for recommendation engines made via JS.

The first model is "lr-ibcf".

Its quite easy to use.

1. Installation:

npm install @lightrec/lr-ibcf

2. Import and Initialize

import { RecEngine } from 'lightrec';



// Create a new recommendation engine

const engine = new RecEngine();

3. Pre-Train (Optional)

const interactions = [

    { userId: 'u1', itemId: 'i1', points: 5 },

    { userId: 'u1', itemId: 'i2', points: 3 },

    { userId: 'u2', itemId: 'i1', points: 4 },

]; // Your interaction data



engine.feed(interactions); 

4. Train

const data = {itemId:'i1', userId:'u1', points:3}

engine.act(data.itemId, data.userId, data.points)

5. Recommend

const recommendations = engine.recommendForUser('u1', 5);

console.log(recommendations); // ['i3', 'i4', 'i5', ...]

And, that's it!

Only 5 lines of code and you get personalized recommendations!

But how this model works?

Its a Item Based Collaborative Filtering method. To know more stay tuned with me, and I'll write another article on how this works.

For now, you may check this github repo:

Github⭐

Have a visit to my portfolio!

Happy Coding!