Home / Companies / Algolia / Blog / Post Details
Content Deep Dive

AI vs non-AI: Building recos for ecommerce & media | Algolia

Blog post from Algolia

Post Details
Company
Date Published
Author
Julia Seidman
Word Count
1,790
Language
English
Hacker News Points
-
Summary

The article discusses the evolution of recommendation tools used by online companies. Traditional heuristic-based filters, which categorize items and track co-occurrence in users' carts or watch lists, have limitations. Modern solutions based on machine learning (ML) offer significant improvements. Amazon is a leader in AI-powered recommender systems, with recommendations accounting for 35% of their sales. Two widely-used models are "Related Items" and "those-who-bought-this-also-bought," both collaborative filtering algorithms. ML-based solutions help with the problem of discovery by noticing patterns in shopper behavior that could be difficult to predict or describe, leading to more personalized recommendations. Algolia Recommend is an API-level recommendation engine that delivers high-performance AI-powered recommendations and iterative improvements.