Development of A Target Company Recommendation System for Account-Based Marketing

Abstract

In this paper, we developed a system for the new B2B marketing method ABM (account-based marketing). The system recommends target companies as future customers by analyzing current customers. There are two requirements in recommending target companies. (1) Users can grasp the reasons for the recommendation of target companies. (2) Users can update the model by modifying the importance of features without affecting that of other features. We propose a model that extends Naive Bayes classifier with a modified smoothing method. The experimental results show that the classification accuracy of our proposed model achieves AUC equal to or better than logistic regression and GBDT.