Research Article

Rough-Set-Based Real-Time Interest Label Extraction over Large-Scale Social Networks

Algorithm 5

User real-time interest label extraction based upper approximation algorithm named RILE.
Input: subgraph stream .
Output: interest labels of all user at the time of .
(1)for to do
(2) for to do
(3)  ;
(4)  obtain subgraph using the followees of each user in ;
(5)  put users in into one of the user types called using and TALE algorithm;
(6)  if the user interest label has not been extracted at the time of then
(7)   crawl the nearly posts posted by all users in until the time of , named ;
(8)   topic modeling on to get candidate user interest label;
(9)   use formulas (9), (10), and (12) to update the interest label of users in according to the sequence of , , and ;
(10)  else
(11)   crawl the nearly posts posted by all users in between the time of and , named and topic modeling on ;
(12)   user interest label at the time of is used as candidate user interest label;
(13)  end
(14)  while new interest labels extracted by user are detected do
(15)   update interest labels of ;
(16)   use UNITE, formulas (9), (10), and (12) to update the interest label of users who follow according to ;
(17)  end
(18)  while new interest labels extracted by user are detected do
(19)   update interest labels of ;
(20)   use UNITE, formulas (9), (10), and (12) to update interest label of users who follow according to ;
(21)  end
(22)  while new interest labels extracted by user are detected do
(23)   update interest labels of ;
(24)   use UNITE, formulas (9), (10), and (12) to update interest label of users who follow according to ;
(25)  end
(26) end
(27)end
(28)return interest labels of all users and each interest of is formalized as .