Fork me on GitHub

Distributed and collaborative algorithms

Pervasive Applications course review - part 3

Distributed algorithms

Collaboration Filtering (CF): Process of filtering for information or patterns using the know preferences of a group of users to make recommendations or predictions of the unknown preferences for others users.

Common insight: personal tastes are correlated

Recommender systems assist and augment the natural social process to help people to find the most interesting and valuable information for them. Tapestry, first recommender system (1992)

CF fundamental assumption: if users X and Y rate n items similarly, or have similar behaviors, and hence will rate or act on other items similarly.


CF Challenges

The Netfix Prize Challenge: open competition for the best CF algo. Team “BellKor’s Pragmatic Chaos” 10.07% improvement of RMSE in 2009


CF Techniques


Memory-Based CF algos:




Model-Based CF Algos:


Hybrid CF Algo

To combine the memory-based and the model-based algos, overcome the limitation of native CF approaches, improves the prediction performance, overcomes the CF problems such as sparsity and loss information. Disadvantage: increased complexity and expensive to implement.