A Cross-Distribution User Intention Identification Based on Topic Transfer

Abstract

At present, in the field of financial science, so as to make intelligent recommendation, the potential financial intention is usually analyzed by the behavior patterns of users. However, due to the diversity of user’ behavior and distribution on different platforms, the feature of users’ financial behavior are also composed of weak features. In this paper, a cross-distributed user’ behavior recognition method based on topic transfer model is proposed. To training a transfer model with APP data in source domain, the model can weaken the influence of different distributions and recognize the intention of the target user. Finally, based on the user’s APP online data, we prove that the model based on topic transfer has better performance on cross-distributed data.