Skip to content

Latest commit

 

History

History
10 lines (7 loc) · 632 Bytes

README.md

File metadata and controls

10 lines (7 loc) · 632 Bytes
Raw

A clustering based social matrix factorization technique for personalized recommender systems

This is the implementation of the CREPE algorithm mentioned in LR D, Tamhane A, Pervin N. A Clustering Based Social Matrix Factorization Technique for Personalized Recommender Systems.

CREPE is a matrix factorization based algorithm for generating restaurant recommendations.

There are 3 main parts to the algorithm:

  1. Clustering users and businesses based on their ratings patterns.
  2. Generating the user-user preference network and item-item similarity network.
  3. Training the recommender system basem to generate predictions.