search

meilisearch / devrel

Anything Developer Relations at Meili
The Unlicense
26 stars 8 forks source link

Creating an e-commerce demo #134

Closed fharper closed 1 year ago

bidoubiwa commented 3 years ago

Just out of curiosity, which feature does the movie dataset not demonstrate? I would think that the following are not really compatible with the dataset

But I'm curious of the other features that might be not suited for this dataset

CaroFG commented 3 years ago

We were thinking of the distinct attribute one

fharper commented 3 years ago

I don't remember our discussion about it, but yeah, we wanted to have another great example.

CaroFG commented 3 years ago

Review of the datasets proposed in this article. I've only reviewed the first 14, because the other are customer reviews and sales data, not really products.

Fashion-MNIST Greyscale training images for Machine Learning (idx files)
Innerwear Data from Victoria’s Secret and Others Underwear. Obsolete image URLs
Electronic Products and pricing data Products from many different websites, no description,
image URLs different sizes, quality or broken
Men's shoes A lot of duplicates. No colors to showcase distinct attribute
Women's shoes A lot of duplicates. No colors to showcase distinct attribute
Item data Unique values. Only 2 fields: id and description.
Fashion products on Amazon.com Not fashion, but toys.
E-commerce tagging for clothing Broken link
Online Retail Dataset (UCI Machine Learning Repository) No picture, nor distinct attribute
Brazilian E-Commerce Public Dataset It's in portuguese. No picture.
Online auctions dataset Auctions + 55% of the items are the same product.
Retailrocket Recommender System Dataset Just numeric data
ECommerce Search Relevance Data from e-bay, very heterogeneous. Missing product description.
Best Buy Search Queries NER Dataset Broken link
CaroFG commented 3 years ago

Finding a suitable e-commerce dataset with images, that we can legally use has revealed itself to be quite tricky. Any help would be much appreciated 🙏

fharper commented 3 years ago

@CaroFG can you define what a perfect dataset should look like please? Which fields, images, copyright...

CaroFG commented 3 years ago

Ideally, the dataset should have the following characteristics:

  1. Free / open license so we can use the dataset
  2. Several fields including :
    • a numerical one (to showcase the custom ranking rule and numeric filters)
    • a long enough field like a "description" to showcase attributesToCrop
  3. Some documents should only differ in one attribute value (size or color) to showcase the distinct attribute
  4. Good quality pictures
  5. Several million documents
bidoubiwa commented 3 years ago

I'm just putting that here in case it might be a possibly dataset. Not sure at all. It's the UK website of oxfam:

Oxfam is a confederation of 20 independent charitable organizations focusing on the alleviation of global poverty,

They have an online shop that uses this API: https://docs.oracle.com/en/cloud/saas/commerce-cloud/cxocc/op-ccstore-v1-search-get.html

I tried several routes and I got some results. It might be a bit tricky to fetch the data the way we want to so it might be not a good solution but so you know.

Their website: https://onlineshop.oxfam.org.uk/

Example of usage of their public API: https://onlineshop.oxfam.org.uk/ccstoreui/v1/search

shivaylamba commented 2 years ago

We are building an e-commerce storefront demo to showcase the functionality of using Meilisearch in an e-commerce website.

The following tech stack will be used for creating the demo :

  1. Frontend Storefront : Medusa NextJS Storefront Starter
  2. MedusaJS : Headless E-Commerce Backend
  3. TailwindCSS
  4. Dataset: Amazon Dataset(private link)

The following features will be showcased :

  1. Searchbar to quickly search for products
  2. Filter on the basis of brand/category/tag
  3. Price Range filter
  4. Rating Menu Filter
  5. Sort on the basis of price/most reviews

Link : https://medusajs-storefront.vercel.app/

shivaylamba commented 2 years ago

The CSS has been updated for the demo.

shivaylamba commented 2 years ago

  1. Currently creating another demo with smaller data showing capabilities of Medusa backend and creating a script to help map data such as price, tags, categories from Meilisearch to enable items such as filters, sorting etc.

  2. Started with the outline for a blog post.

shivaylamba commented 2 years ago

Update:

  1. Successfully created another demo with smaller data showing capabilities of Medusa backend and creating a script to help map data such as price, tags, categories from Meilisearch to enable items such as filters, sorting etc.

  2. Updated the outline for the blog post with the new demo.

shivaylamba commented 2 years ago

Update: Demo in the final stages of getting live: Currently a PR has been created: https://github.com/meilisearch/demos/pull/65

Strift commented 1 year ago

Hello there,

It's been a while since we got an update on this issue. I've started working on a new ecommerce demo. ✨

Context & Objectives

Demo website Showcase how Meilisearch can be used to build an ecommerce website.

Demo repository Offer a starting point or reference for how to implement Meilisearch for ecommerce with Nuxt 3

Tutorial An article will also be created to complement to the other deliverables, and offer a more guided approach to the implementation.

🔗 (WIP) repository: meilisearch/ecommerce-demo

Current Progress

⚙️ A note on SSR

🚀 What's next?

After a round of feedback on the demo, I will start working on the tutorial—the article counterpart that will go with this demo.

(The following might be moved to a more relevant issue)

The current idea for it is:

The tutorial would roughly follow these steps:

Cheers,

shivaylamba commented 1 year ago

Looking forward to seeing the new demo @Strift!

Strift commented 1 year ago

Happy to announced we've published the new demo 🚀

shivaylamba commented 1 year ago

super awesome!