##Compiler & OS

• g++
• Ubuntu 14.04

##Process

• 1.Use tf-idf get the most representative words from all of the documents
• 2.Use lda(Latent Dirichlet Allocation) topic model to draw realation between topics and words, documents and topics.
• 3.Use ctr(Collaborative Topic Recommendation) to draw realtion between users and topics, and modify realation between documents and topics using collaborative filtering

##Run

• ./lda est 8 200 settings.txt mult.txt random result/
• ./ctr --directory result/ --user user.txt --item item.txt --mult mult.txt --theta_init final.gamma --beta_init final.beta
• ./predict.out