SocialNetwork

SocialAutoRec:

See ./AERec

Data Format:

douban_train.txt and douban_test.txt:

For each line: userid /t itemid /t rating

social_info.txt:

For each line: follower /t followee

(follower:关注者，followee:被关注者)

douban_emb:

first line: number of users, embedding dimensions

for each line:

userid dimension1 dimension2 ... dimension128

\ \

Evaluation:

Rooted Mean Square Error(RMSE) and Mean Avarage Error(MAE).

社交网络的可视化：

见social_visual中，graph-npm.html文件，效果如下图，记得将三个.js文件放到同一文件夹下。

Baseline recommendation model结果

models	MAE	RMSE
MostPopular	1.564	1.849
pmf	0.650	0.848
socialmf	0.639	0.809
sorec	0.640	0.810
soreg	0.733	0.961
trustmf	0.718	0.912
trustsvd	0.858	1.146
AutoRec	0.645	0.826
SocialAutoRec	0.635	0.814
SocialAutoRec-ClusterReg	0.617	0.796