Research Article
Detecting Shilling Attacks with Automatic Features from Multiple Views
Algorithm 1
Data preprocessing.
Input: DR | |
Output: R, P, M | |
1if (item sequence is not updated) then | |
2Calculate the novelty of items NIi according to formula (7) | |
3Construct the item popularity sequence SQI | |
4end if | |
8for each do | |
9for curi=1 to | |
10ui=indexU(curi) | |
11R(ui,i)=rate(DR,ui,i) | |
12P(ui,i)=NIi | |
13for curj= curi+1 to | |
14uj=indexU(curj) | |
15if ui!=uj then | |
16M(ui, uj)=M(ui, uj)+1 | |
17M(uj, ui)=M(ui, uj) | |
18end if | |
19end for | |
20end for | |
21end for | |
22Return R, P, M |