Dual-aspect Item Attention Network For Item-based Recommendation

With the rapid development of Internet and multimedia technologies,many customercentric websites and online services as the core of the websites or applications have emerged,such as e-commerce websites(Taobao,Jingdong,etc.),social media applications(Meituan,Tik Tok,Netease news,etc.)and so on.In order to improve the operaton efficiency and profit of websites or applications,recommendation systems have been widely used.However,how to extract each user's interests from the huge number of items on the website and personalize recommendation items for each user are great challenges to the recommendation systems.Attention mechanism is a neural network algorithm that has been developed in recent years and has attracted widespread attention of researchers.It has been successfully applied in many fields such as computer vision,natural language processing and machine translation.Therefore,this paper introduces attention mechanism and proposes a dual-aspect item attention network(DIAN).Thesis proposes dual-aspect item attention network(DIAN)for item-based recommendation,takes into account the aspects of importance of historical items in a user profile to the target items and the underlying relations among these items jointly.DIAN consists of two main modules,an attention-based item similarity model for item similarity between historical and target items(ABIS),and a dual normalization self-attention item similarity model for item similarity underlying historical items(SAIS).This paper designs an attention network to distinguish the different contribution of the historical items in a user profile to the perdition on the target item.In order to learn better representations for users' interests,this paper introduces a self-attention mechanism to infer the item-item relationship from user's historical interactions,and can estimate the relative weights of each item in user interaction trajectories.This paper also proposes a self-attention network that uses a dual normalization mechanism,consisting of a layer focusing on extracting user's representation from history items,and a layer making it unaffected by the number of user's historical items.In this paper,we conduct extensive experiments on two public datasets to verify the effectiveness of our proposed models.Experiments show that the proposed method can achieve accurate personalized recommendations and outperforms the state-of-the-art recommendation models.