Collaborative Filtering 

- It is a technique used by some recommender systems. 

- has two seneses 

         - narrow one 

         - more general one 

- Collaborative filtering explores techniques for matching people with similar interests and making recommendations on this basis.

- Collaborative filtering algorithms often require 
   (1) users’ active participation, 
   (2) an easy way to represent users’ interests to the system, and 
   (3) algorithms that are able to match people with similar interests.

- Methodology
      - User-based collaborative filtering 
      - Item-based collaborative filtering 

- Types 
      - Memory based 

              This mechanism uses user rating data to compute similarity between users or items

               Pearson correlation or Vector cosine 

              A popular method to find the similar users is the Locality sensitive hashing, which implements the  nearest neighbor mechanism in linear time.

     - Model-based 
             Models are developed using data mining, machine learning algorithms to find patterns based on training data
             Model based CF Algorithms. 
                   These include Bayesian Networks, 
                                        clustering models, 
                                        latent semantic models such as 
                                                              singular value decomposition, 
                                                              probabilistic latent semantic analysis, 
                                                              Multiple Multiplicative Factor, 
                                                              Latent Dirichlet allocation and 
                                                              markov decision process based models