Research On Collaborative Session-based Recommendation With Memory Networks

With the rapid development of the Internet industry and big data technology,the network information resources have increased significantly.On the one hand,more and more people enjoy the convenience and efficiency brought by the information revolution.On the other hand,information overload becomes a new challenge in the big data era.For users,it is important and di:fficult to find the content they need in the growing digital information resources.For service providers,how to accurately provide users with goods and services to meet the needs of users and improve service quality is also a challenging task.The recommendation system can alleviate these problems.Based on the users'personal information and historical behaviors,the recommendation system can predict users'interest,and help users select goods or services that they are interested in more quickly.The recommendation system can satisfy users'needs with personalized services,and enhance the interaction between users and service providers.The more the information resources on the Internet,the higher the value of personalized recommendation systems.The traditional recommendation systems usually use the users'ratings,browsing history or the basic information of users and items.But they cannot perform well in many realistic scenarios that users are not logged in.To tackle this problem,session-based recommendation is proposed to predict the next item to recommend when the only available information consists of anonymous behavior sequences.In recent years,using recurrent neural networks(RNNs)for session-based recommendation has attracted wide attention and achieved encouraging results.RNNs use their hidden state to store the user's information.The current hidden state is determined by the previous hidden state and the current input.RNNs can model the sequential behavior patterns of the current session and help learn the user's short-term preferences.However,previous methods for session-based recommendation focus mostly on the current session,ignoring collaborative information in so-called neighborhood sessions,sessions that have been generated previously by other users and reflect similar user intents as the current session.We hypothesize that the collaborative information contained in such neighborhood sessions may help to improve recommendation performance for the current session.Therefore,in the research work of this thesis,we propose a Collaborative Session-based Recommendation Machine(CSRM),a novel hybrid framework to apply collaborative neighborhood information to session-based recommendations.CSRM consists of two parallel modules:an Inner Memory Encoder(IME)and an Outer Memory Encoder(OME).Specifically,the IME models a user's own information in the current session with the help of RNNs and an attention mechanism.The OME exploits collaborative information to better predict the intent of current sessions by investigating neighborhood sessions.Then,a fusion gating mechanism is used to selectively combine information from the IME and OME so as to obtain the final representation of the current session.Finally,CSRM obtains a recommendation score for each candidate item by computing a bilinear match with the final representation.To demonstrate the effecti-veness of CSRM,we conduct extensive experiments on three public datasets.Experimental results demonstrate significant improvements over state-of-the-art baselines in terms of Recall@20 and MRR@20.Our analysis of CSRM's recommendation process shows when and how collaborative neighborhood information and the fusion gating mechanism positively impact the performance of session-based recommendations.