*Title: Joint Nonnegative Matrix Factorization for Hybrid Clustering based on Content and Connection Structure
*Speaker: Haesun Park
School of Computational Science and Engineering
Georgia Institute of Technology, Atlanta, GA, U.S.A.
*Place/Date: AORC Seminar room/ May 15,2018(Tue) pm 4:30-5:30
* ABSTRACT:
A hybrid method called JointNMF is presented for latent information discovery from data sets that contain both text content and connection structure information. The method jointly optimizes an integrated objective function, which is a combination the Nonnegative Matrix Factorization (NMF) objective function for handling text content and the Symmetric NMF (SymNMF) objective function for handling relation/connection information. An effective algorithm for the joint NMF objective function is proposed utilizing the block coordinate descent (BCD) framework. The proposed hybrid method simultaneously discovers content associations and related latent connections without any need for post-processing or additional clustering. It is shown that JointNMF can also be applied when the text content is associated with hypergraph edges. An additional capability is prediction of unknown connection information which is illustrated using some real world problems such as citation recommendations of papers and leader detection in organizations.