Website categorization: A formal approach and robustness analysis in the case of e-commerce detection.

•Robust formal approach to website categorization based on web mining and classification.•Entirely automated procedure using a computationally viable pipeline.•Application to an important case: the detection of e-commerce in corporate websites.•Uses machine learning and dictionaries, hence applicable in other contexts or languages.•Analysis of the robustness w.r.t. the presence of misclassified training records.