Skip to content

Latest commit

 

History

History
 
 

04_ecommerce_vectors

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
README.md
README.md
 
 

README.md

Make E-commerce Vector-powered

Hi everybody,

In this session I'll introduce you to Superlinked - a framework for building vector-powered systems!

We will build Search, Recommendations and Analytics on top of a multi-modal e-commerce dataset.

THE CATCH: In most demos like this, people just use product images and descriptions.. while in reality you have to deal with product ratings, prices, behavioral data and more. Today, we will fix that!

Session Overview

Instructor: Daniel Svonava, Founder at Superlinked, ex-ML TL in YouTube Ads

Duration: 30 minutes (1:30PM - Track 2)

Objectives:

  • Learn how to recognize problems solvable with vector embeddings
  • Learn to work with multi-modal e-commerce data in Google Colab
  • Learn how to integrate metadata into your vector search projects (and no we don't mean 'hybrid search')
  • Sign up to Part 2 of this workshop, focused on bringing your vector-powered systems to production

By the end of this session, you will have a deeper understanding of how to apply vector search to a range of use-cases in e-commerce and to similar multi-modal data.

Prerequisites

  • Basic knowledge of Python and ability to follow along in a Google Colab.
  • No external dependencies required (i.e. you don't need a local python environment or a vector database to get started).

Agenda

  1. Get the data

    • Use a Google Colab notebook to analyse an e-commerce product dataset of text, images, numbers and categorical properties

  2. Let's start building

    • Build a Similar Products feature that uses all the available product metadata
    • Build a simple Recommendation system that uses user data & behavioral events
    • Build a Search system 1.0 that uses the image and description data
    • Build a Search system 2.0 that takes into account all product metadata
    • Build a Search system 3.0 that adds per-user result personalization

  3. Part 2 Summary + Q&A

    • Get a sneak-peak at Superlinked Cloud CLI for running vector-powered systems reliably at scale
    • Ask your questions & connect with the speaker (LinkedIn, X) - 👋 I mean it!

Instructions

Step 1: Navigate to Google Colab & follow along

Additional Resources

Solutions

If you need help with any part of the session, refer to the solution file in the solutions folder.

Contact

If you have questions during the workshop, please reach out to [[email protected]] or open an issue in the repository.

Happy coding!