DFIELDSOLUTIONS

Free software

LabKithra: People like you liked these

Item-item CF, the approach Amazon scaled: cosine/Pearson similarity over a ratings matrix, then a k-nearest similarity-weighted prediction. No training - just similarity and a weighted mean. 13 tests, including the full “fan of Inception → Interstellar” scenario and an honest cold-start.

Free software1 sections · 2 min readBy Dezső Mező.mdFree softwareKithra

Kithra is a recommendation engine built on item-item collaborative filtering, the approach Amazon scaled to production. The signature technique computes cosine or Pearson similarity over a ratings matrix and then makes a k-nearest, similarity-weighted prediction - no training, just similarity and a weighted mean - tested with the full "fan of Inception leads to Interstellar" scenario and an honest cold-start case. It is a from-scratch, dependency-light build you can download and run locally.

01About the app

Collaborative filtering

What to take away

  • Item-item CF, the approach Amazon scaled: cosine/Pearson similarity over a ratings matrix, then a k-nearest similarity-weighted prediction. No training - just similarity and a weighted mean. 13 tests, including the full “fan of Inception → Interstellar” scenario and an honest cold-start.
  • Download: Kithra (ZIP)
We build this for clientsFull-stack

Read the full write-up

More from the lab

Browse all entries

Want this looked at on your own system?Start a conversation

5.0
“New shop, more orders in the first week.”Müller Márk · GeekCornerManaging Director · Hungary