320 likes | 523 Views
协同过滤中两个问题探讨. Related papers : Modeling Evolutionary Behaviors for Community-based Dynamic Recommendation --SIAM 2. Learning Preferences of New Users in Recommender System : An Information Theoretic Approach --SIGKDD. 演讲人:余海洋 2011:12:10. Will user U like item X?.
E N D
协同过滤中两个问题探讨 Related papers : Modeling Evolutionary Behaviors for Community-based Dynamic Recommendation --SIAM 2. Learning Preferences of New Users in Recommender System : An Information Theoretic Approach --SIGKDD 演讲人:余海洋 2011:12:10
1.Content-based filtering • 2. Collaborative filtering(协同过滤) • 3.Hybrid approaches
Cold start Dynamic Recommendation
How to choose items? Popularity Entropy Entropy0 HELF : harmonic mean of Entropy and logarithm of Frequency IGCN : Information Gain through Clustered Neighbors
1. PopularityIndicates how frequently users rated the item Uninformative Prefix bias
2. EntropyRepresents the dispersion of opinion of users on the item It often selects very obscure items Suggests “garbage” patterns to be useful as well. Do not considers the frequency .
3. Entropy0 Bias frequently-rated too much
5. IGCNInformation Gain through Clustered Neighbors IGCN works by repeatedly computing information gain of items where the necessary rating data is considered only from those users who match best with the target users profile so far
5. IGCNInformation Gain through Clustered Neighbors Missing values Inadequate goals
5. 2 Inadequate goals Find the final best k neighbors
Dynamic Recommendation 用户兴趣的变化 物品流行度的变化 季节效应
CBDB Community-based Dynamic Recommendation
Formal community construction People from the same department of same company tend to have similar interests , since they tend to have similar background and are working on similar projects.
Dynamic pattern analysis Freshness of documents Short-term of long-term type of document Popularity of documents User intention
Adaptive community model Time-Sensitive Adaboost