Exploiting efficient and effective lazy Semi-Bayesian strategies for text classification.

•A thorough study on the aforementioned combinations of NB strategies and feature weighting approaches.•We proposed a methodology that enables a deeper study of the impact of these NB strategies.•The proposal of new GPU-based lazy Semi-NB approach, called LSPTAN.•The introduced of an efficient GPU-based parallel implementation.•We investigate the real impact of the NB attribute independence assumption and the proposed Semi-NB approaches.