AI marketing connecting different services¶

Research Background¶

AI marketing is "the provision of high-quality persona services through the prediction of user behavior patterns using AI technology. This research aims to promote the AI transformation of marketing. For example, by utilizing user (persona) models such as predicting purchases of e-commerce products or visits to real-world stores, it enhances user experience and increases profits for businesses. However, in the real world, sharing user (persona) models built by each service provider is challenging due to issues such as privacy and rights. Therefore, our research group is aiming to achieve AI marketing technology considering privacy protection that connects different services. Specifically, to address issues such as privacy and rights, we incorporate the premise of "no data sharing" into the research. Unlike existing cross-domain studies, we envision "completely different heterogeneous domains" and undertake the development of transfer learning techniques for behavior prediction across domains.

Research Overview¶

This research group is engaged in the development of techniques utilizing AI technologies such as collaborative filtering, deep learning, adversarial learning, transfer learning, and meta-learning to learn distributed representations (vectors) of users and construct models to capture user behavior patterns.

Learning user distributed representations from heterogeneous features and developing cross-domain recommendation systems.
Constructing and developing persona recommendation models optimized for domain-specific features.
Developing visit prediction systems to address the long-tail problem using clustering.
Developing ensemble methods for sales forecasting focusing on label imbalance.
Analysis of the impact of the COVID-19 pandemic on user behavior (transition of behavioral patterns).