Prediction and Analysis of Amazon User Behavior Based on Long Short-Term Memory and Manual Feature

Abstract

The user preference information implicits that the user behavior sequence has an important impact on the click-through rate estimation. Current research on click-through rate estimation mostly designs models from the perspective of low- order and high-order interactions of modeling features, which is difficult to fully explore and utilize user preference information implied in the user’s historical behavior sequence. This article reviews and summarizes the results of previous research and exploration on the click-through rate estimation problem. On this basis, this article proposes a click-through rate prediction model based on user preference networks, modeling user preferences. Two Experiments on the two click-through rate prediction models proposed in this paper based on user preference networks are conducted based on the international public data set. The final experimental results show that the model proposed in this paper can fully mine user preference information hidden in user behavior sequences and, as a result, obtain more effective results. User preference indicates that compared to other comparative click-through rate estimation models, it has achieved better performance in the relative improvement of AUC and models.