Systems and methods for a cross media joint friend and item recommendation framework

Abstract

Various embodiments of systems and methods for cross media joint friend and item recommendations are disclosed herein.

What is claimed is: 
     
         1 . A method, comprising:
 developing an objective function, wherein the objective function comprises a shared dictionary, a set of sparse item representations, a set of latent user features, a projection matrix, and an interaction matrix;   executing a dictionary learning method, wherein the dictionary learning method updates the shared dictionary;   updating the set of sparse item representations using an alternating direction method of multiplier;   updating the set of latent user features and the interaction matrix using partial derivatives of the objective function with respect to the set of latent user features and the interaction matrix, respectively; and   updating the projection matrix using a gradient descent optimization procedure;   wherein the shared dictionary, the set of sparse item representations, the latent user features, the projection matrix, and the shared interaction matrix are iteratively updated until the objective function converges.   
     
     
         2 . The method of  claim 1 , further comprising:
 initializing the shared dictionary, the set of sparse item representations, the set of latent features, the projection matrix, and the shared interaction matrix.   
     
     
         3 . The method of  claim 1 , further comprising:
 pre-computing a graph Laplacian matrix and a maximum mean discrepancy matrix, wherein the graph Laplacian matrix and the maximum mean discrepancy matrix are included in the objective function.   
     
     
         4 . The method of  claim 1 , wherein updating the shared dictionary, the set of sparse item representations, the latent user features, the projection matrix, and the shared interaction matrix until the objective function converges produces a resultant latent user representation matrix and a resultant item representation matrix, and wherein the resultant latent user representation matrix and the resultant item representation matrix are respectively used to perform friend recommendation tasks and item recommendation tasks across a source social media site and a target social media site.