Iterative Multi-Label Multi-Relational Classification Algorithm for Complex Social Networks

Abstract : We consider here the task of multi-label classification for data organized in a multi-relational graph. We propose the IMMCA model—Iterative Multi-label Multi-relational Classification Algorithm—a general algorithm for solving the inference and learning problems for this task. Inference is performed iteratively by propagating scores according to the multi-relational structure of the data. We detail two instances of this general model, implementing two different label propagation schemes on the multigraph. To the best of our knowledge, this is the first collective classification method able to handle multiple relations and to perform multi-label classification in multigraphs. Tests are performed for two generic applications, image annotation and document classification, on different social datasets. For image annotation, we have been using Flickr datasets of different sizes and with different configurations, with multiple relations such as authorship, friendship, or textual similarities. For document classification, we used the Cora classical benchmark plus an Email corpus. Additional experiments on artificial data allow us to analyze further the behavior of the model.
Document type :
Journal articles
Complete list of metadatas
Contributor : Lip6 Publications <>
Submitted on : Tuesday, July 7, 2015 - 2:16:20 PM
Last modification on : Friday, May 24, 2019 - 5:24:00 PM



Stephane Peters, Yann Jacob, Ludovic Denoyer, Patrick Gallinari. Iterative Multi-Label Multi-Relational Classification Algorithm for Complex Social Networks. Social Network Analysis and Mining, Springer, 2012, 2 (1), pp.17-29. ⟨10.1007/s13278-011-0034-8⟩. ⟨hal-01172456⟩



Record views