Sequential recommendation as an emerging topic has attracted increasing attention due to its important practical significance. Models based on deep learning and attention mechanisms have achieved good performance in sequential recommendation. Recently, the generative models based on Variational Autoencoder (VAE) show the unique advantage in collaborative filtering. In particular, the Sequential VAE model as a recurrent version of VAE can effectively capture temporal dependencies among items in user sequence and perform sequential recommendation. However, VAE-based models suffer from a common limitation that the expression ability of the obtained approximate posterior distribution is limited, resulting in lower quality of generated samples. This is especially true when making sequence generation. To solve the above problem, in this work, we propose a novel method called Adversarial and Contrastive Variational Autoencoder (ACVAE) for sequential recommendation. Specifically, we first employ the contrastive loss. The latent variables will be able to learn more personalized characteristics for different users by maximizing the mutual information between input sequences and latent variables. Then, we introduce the adversarial training for sequence generation under the Adversarial Variational Bayes framework, which enables our model to generate high-quality latent variables. Besides, we apply a convolutional layer to capture local relationships between adjacent latent variables of items in the sequence. Finally, we conduct extensive experiments on three real-world datasets. The experimental results show that our proposed ACVAE model outperforms other state-of-the-art methods.

2021 THE WEB CONFERENCE NEWSLETTER
The Web Conference is announcing latest news and developments biweekly or on a monthly basis. We respect The General Data Protection Regulation 2016/679.