Heterogeneous Source Consensus Learning via Decision Propagation and Negotiation

Nowadays, enormous amounts of data are continuously gen- erated not only in massive scale, but also from di®erent, sometimes con°icting, views. Therefore, it is important to consolidate di®erent concepts for intelligent decision mak- ing. For example, to predict the research areas of some people, the best results are usually achieved by combining and consolidating predictions obtained from the publication network, co-authorship network and the textual content of their publications. Multiple supervised and unsupervised hypotheses can be drawn from these information sources, and negotiating their di®erences and consolidating decisions usually yields a much more accurate model due to the di- versity and heterogeneity of these models. In this paper, we address the problem of consensus learning" among compet- ing hypotheses, which either rely on outside knowledge (su- pervised learning) or internal structure (unsupervised clus- tering). We argue that consensus learning is an NP-hard problem and thus propose to solve it by an e±cient heuris- tic method. We construct a belief graph to ¯rst propagate predictions from supervised models to the unsupervised, and then negotiate and reach consensus among them. Their ¯nal decision is further consolidated by calculating each model's weight based on its degree of consistency with other models. Experiments are conducted on 20 Newsgroups data, Cora research papers, DBLP author-conference network, and Ya- hoo! Movies datasets, and the results show that the proposed method improves the classi¯cation accuracy and the cluster- ing quality measure (NMI) over the best base model by up to 10%. Furthermore, it runs in time proportional to the number of instances, which is very e±cient for large-scale data sets.