The construction and interaction procedure of phishing and user in the deception mode is presented. We analyses phishing behavior when tempting human in order to construct a phishing-deception human-based data model (PDHDM) based on frequent associated events. The proposed phishing-deception human-based data model is utilized to generate association rules and to accurately classify between phishing and legitimate websites. This approach can reduce false positive rates in phishing detection systems, including a lack of effective dataset. Classification algorithms is employed for training and validation of the model. The proposed approach performance and the existing work is compared. Our proposed method yielded a remarkable result. The finding demonstrates that phishing-deception human-based data model is a promising scheme to develop effective phishing detection systems.