NBBR: A Baseline Method for the Evaluation of Bayesian Multi-label Classification Algorithms

Multi-label classification (MLC) is the task of automatically assigning an object to multiple categories. There are many important and modern applications of MLC such as text categorization (associating documents to various subjects) and functional genomics (determining the multiple biological functions of genes and proteins). MLC problems typically involve datasets that are both very large in size and highly complex in structure (e.g., text data, multimedia data, biological data, etc.), which drives the need for scalable algorithms. In this paper we propose the NBBR method, a simple and fast adaptation of the Naive Bayes algorithm for use in the context of MLC.