Using Graph Partitioning Techniques for Neighbour Selection in User-Based Collaborative Filtering
0 likes706 views
This document proposes using spectral clustering techniques like normalized cuts for identifying neighbors in user-based collaborative filtering. It shows that this approach works better than k-means clustering and standard user-based collaborative filtering, providing higher prediction accuracy and coverage for recommendations. The approach identifies neighbors based on clustering users based on their ratings rather than selecting the nearest users.
Using Graph Partitioning Techniques for Neighbour Selection in User-Based Collaborative Filtering
- 1. Improvement of user-based CF by using spectral clustering techniques ACM Conference on Recommender Systems 2012 Poster slam September 11, Dublin, Ireland
- 2. Context: cluster-based CF neighbours identified based on a clustering method Spectral Clustering: Normalised Cut ACM Conference on Recommender Systems 2012 Poster slam September 11, Dublin, Ireland
- 3. Context: cluster-based CF neighbours identified based on a clustering method Spectral Clustering: Normalised Cut Better than k-Means: performance k-Means NCut 0.12 0.10 0.08 P @ 0.06 5 0.04 0.02 0.00 50 100 150 200 250 300 350 400 450 500 550 600 k ACM Conference on Recommender Systems 2012 Poster slam September 11, Dublin, Ireland
- 4. Context: cluster-based CF neighbours identified based on a clustering method Spectral Clustering: Normalised Cut Better than k-Means: coverage cov(k-Means) cov(Ncut) 100 90 C 80 o 70 v 60 e 50 r a 40 g 30 e 20 10 0 50 100 150 200 250 300 350 400 450 500 550 600 k ACM Conference on Recommender Systems 2012 Poster slam September 11, Dublin, Ireland
- 5. Context: cluster-based CF neighbours identified based on a clustering method Spectral Clustering: Normalised Cut Better than k-Means Also better than MF and standard UB UB50 MF NCut 0.12 0.10 0.08 P @ 0.06 5 0.04 0.02 0.00 50 100 150 200 250 300 350 400 450 500 550 600 k ACM Conference on Recommender Systems 2012 Poster slam September 11, Dublin, Ireland
- 6. Using Graph Partitioning Techniques for Neighbour Selection in User-Based Collaborative Filtering Alejandro Bellog鱈n Javier Parapar Information Retrieval Group Information Retrieval Lab Universidad Aut坦noma de Madrid University of A Coru単a ACM Conference on Recommender Systems 2012 Poster slam September 11, Dublin, Ireland