Network-based Recommendation Algorithm
Yiqin Cao · Research and Exploration in Laboratory · 2012
A novel network-based recommendation algorithm that considers the item attribute is proposed to solve cold start.The similarity of any two items was calculated according to the item attribute matrix.Then the final resource allocation vector of the target user was calculated using the resource allocation matrix with the item similarity and the first resource allocation vector.Finally the top-N items that the target user did not choose in the past were recommended to the user.This algorithm can solve the problem of new item recommendation effectively and provide better recommendation results for the system.