PushTrust: An Efficient Recommendation Algorithm by Leveraging - - PowerPoint PPT Presentation
PushTrust: An Efficient Recommendation Algorithm by Leveraging Trust and Distrust Relations Rana Forsati, Iman Barjasteh, Farzan Masrour, Abdol-Hossein Esfahanian, Hayder Radha Michigan State University RecSys15 Outline Some background
Rana Forsati, Iman Barjasteh, Farzan Masrour, Abdol-Hossein Esfahanian, Hayder Radha Michigan State University RecSys’15
2
Recommend ¡items to ¡users ¡to ¡maximize ¡some ¡objective(s)
Latent feature models (e.g., matrix factorization): The current most successful technique as demonstrated in KDD Cup and Netflix competition
3
Intuition: ratings are deeply influenced by a set of non-
Goal: Extract latent features from existing ratings for future (unknown) predictions Items: Users: Partially observed rating matrix:
4
How to find latent features for users and items? Assume there are k latent features for rating
5
Solve ¡the ¡following ¡ ¡optimization ¡problem: Model ¡each ¡user/item ¡as ¡a ¡vector ¡of ¡features ¡(learned ¡from ¡data)
: observed ratings
Training error on
Regularization to avoid overfitting
6
Huge ¡number ¡of ¡ unrated ¡movies
Handling cold-start users
who have rated only a few items
? ? ? ? ?
? ? ?
? ? ? ? ? ? ? ? ?
Sparsity of rating matrix
Example: Epinions only 0.02%
7
Side ¡ Information ¡ about ¡Items
Rating ¡Matrix
? ? ? ? ? ? ? ? ?
Goal: exploit other sources of information to cope with these challenges
Side ¡ Information ¡about ¡Items Side ¡ Information ¡ about ¡Users
Side ¡ Information ¡ about ¡Ratings
8
Social Recommender Systems Spatial Recommender Systems
Social Networking Services (e.g., Facebook, Twitter)
9
Adding side information to resolve sparsity and cold-start problems.
Our ¡focus: trust/distrust relations ¡between ¡users
How to effectively exploit social relations of users in recommendation to boost the accuracy?
Growth of social networks Trust ¡= agreement on ratings [Guo et al., 2014] ¡
Opinion Aware Recommendation Systems
Distrust ¡= disagreement on ratings [Forsati et al., 2014] ¡
10
11
propogate.
12
Rating Matrix Trust Network Distrust Network
Memory based methods: distrust is used to either filter out or debug propagated trust network [Victor et al., 2011] Factorization based methods: regularization by ranking of latent features [Forsati et al., 2014]
13
Trust+Distrust enhanced MF: the latent features of trusted users are closer and distrusted users are more distant
trust network Distrust network users arranged based on similarity of their latent features
14
where
Regularization with Social Ranking
15
Computationally expensive: in large social graphs,
number of users
.
Only consider the trusted and distrusted friends and ignores the neutral (users with no relation) users.
The neutral friends might appear before the trust friends and be negatively influential! Not scalable to large social networks!
The top portion of ranked list might include distrusted friends due to nature of pairwise ranking model
The latent features might affected by distusted friends who appear at top of the ranked list!
16
A more complete approach:
to the top of list as much as possible.
bottom of list as much as possible.
PUSHED to the middle of list.
17
Rewrite the social-regularized matrix factorization Here is social regularization of latent features of individual users.
18
Put neutral friends above distrusted ones Put trusted friends above neutral ones Put trusted friends above distrusted ones
where is indicator function.
non-convex indicator function
: latent features of trusted friends : latent features of neutral friends : latent features of distrusted friends
19
Indicator function:
20
Hinge loss:
21
Rewrite the social regularized matrix factorization where
22
The problem is non-convex jointly in both U and V. Solution: The standard gradient descent method
23
Set , where
24
Set , where
25
incorporated in ranking the latent features.
quadratically
26
27
Two evaluation metrics:
Mean Absolute Error: Root Mean Squared Error: where is the set of rating that should be predicted.
28
The only social rating network dataset publicly available. User trust and distrust information is included in this dataset The social network in Epinions is directed
Users Users Items Items Ratings Ratings Trust Relations Trust Relations Distrust Relations Distrust Relations Quantity Quantity 121,240 685,621 12,721,437 481,799 96,823
80% of the rating data was selected as the training set with 20% as the test data.
29
■ MF: Matrix Factorization ■ MF-T: Matrix Factorization with Trust information ■ MF-D: Matrix Factorization with Distrust information ■ MF-DT: Matrix Factorization with Trust and Distrust ■ PushTrust: The proposed algorithm
0. 0.25 0.5 0.75 1.
5 10
MAE Latent vector dimensions PushTrust MF MF-T MF-D MF-DT
0. 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1. 1.1 1.2 1.3
5 10
RMSE Latent vector dimensions
30
To evaluate different algorithms:
select 30%, 20%, and 10%
users. For cold-start users, no rating is included in the training data consider all ratings made by cold-start users as testing data.
0.75 0.8125 0.875 0.9375 1 10% 20% 30% Cold-start users
MAE
MF MF-T MF-D MF-TD PushTrust
1 1.1875 1.375 1.5625 1.75 10% 20% 30% Cold-start users
RMSE
31
Thank You for you attention! 32 Questions?
32