Click-Through Rate (CTR) prediction plays an important role in many industrial applications, such as online advertising and recommender systems. How to capture users' dynamic and evolving interests from their behavior sequences remains a continuous research topic in the CTR prediction. However, most existing studies overlook the intrinsic structure of the sequences: the sequences are composed of sessions, where sessions are user behaviors separated by their occurring time. We observe that user behaviors are highly homogeneous in each session, and heterogeneous cross sessions. Based on this observation, we propose a novel CTR model named Deep Session Interest Network (DSIN) that leverages users' multiple historical sessions in their behavior sequences. We first use self-attention mechanism with bias encoding to extract users' interests in each session. Then we apply Bi-LSTM to model how users' interests evolve and interact among sessions. Finally, we employ the local activation unit to adaptively learn the influences of various session interests on the target item. Experiments are conducted on both advertising and production recommender datasets and DSIN outperforms other state-of-the-art models on both datasets.

Recommendation systems and computing advertisements are of great value for commercial applications. Click-through rate (CTR) prediction is a critical issue because the prediction accuracy affects the user experience and the revenue of merchants and platforms. Feature engineering is usually used to improve the click-through rate prediction; however, it heavily relies on user experience. It is difficult to construct a feature combination that can describe the complex patterns implied in the data. This paper combines the traditional feature combination methods and the deep neural networks to automate the feature combinations to improve the accuracy of the click-through rate prediction. We propose a mechanism named Field-aware Neural Factorization Machine (FNFM). This paper can have strong second-order feature interactive learning ability, such as Field-aware Factorization Machine; on this basis, a deep neural network is used for higher order feature combination learning. This experiment shows that the model has stronger expression ability than previous deep learning feature combination models, such as the DeepFM, DCN, and NFM.