Cross-Modal Search for Social Networks via Adversarial Learning
Algorithm 1
The matching algorithm of CMSAL.
Input: Query set Q = {q1, q2, …, qT} about the target topic d; cross-modal presentative data features from social networks generated by the networks with optimized parameters and :
Output: Search result list.
(1)
For q in the query set Q:
(2)
Distinguish modality type of q
(3)
Preprocessing q into corresponding features blocks as for q in text or for q in images.
(4)
Extracting representation features: based on the optimized parameters for q in text or based on the optimized parameters .
(5)
For s in the cross-modal representation features set Sdoc:
(6)
Computing the similarity according to the query similarity = sim (Sq, s)