Code for CIKM-2016 paper "Collective Classification via Discriminative Matrix Factorization on Sparsely Labeled Networks"
The main program is start.m.
Cora contains 2,708 machine learning papers from seven classes and 5,429 links between them. The links are citation relationships between the documents. Each document is described by a binary vector of 1,433 dimensions indicating the presence/absence of the corresponding word.
Citeseer contains 3,312 publications from six classes and 4,732 links between them. The links are citation relationships between the documents and each paper is described by a binary vector of 3,703 dimensions indicating the presence/absence of the corresponding word.
PubMed contains 19,717 papers from 3 classes and 44,338 links between them. Each paper is describled by a 500-dimensional TF-IDF feature vector.
graph.txt: Each line contains two paper Ids which indicates the citation relationship between them. ID begins from 0.
group.txt: Each line contains two numbers: Paper Id and Group Id. For Cora and Citeseer, group Id begins from 0; For PubMed, group Id begins from 1.
feature.txt for Cora and Citeseer: This is the Paper-Word relationship matrix. Each line contains a binary vector of 1, 433 dimensions indicating the presence of the corresponding word.
tfidf.txt for PubMed: This is the sparse version of TFIDF matrix of PubMed dataset. Each line contains a paper Id, feature Id and the corresponding feature value.