Design And Implementation Of Movie Recommender System Based On Dialogue System

With the development of deep learning technology and the advent of big data era,the dialogue system and recommender system have gradually penetrated into our lives.Such as voice assistant 'Siri' and movie recommendation website'Douban'.The combination of dialog system and recommender system called conversational recommendation system has become a novel method of recommendation.The system can complete the recommendation when talking with users.The traditional conversational recommender system needs a lot of manual rules.At the same time,most of the recommendation system also uses the retrieval-based method.Based on the above background,this paper designs and implements a conversational recommender system for movie recommendation task.After a lot of literature reading and user research,the implementation of the system is divided into three parts:the construction of conversational recommender system model,the requirement analysis and outline design of the system,and the detailed design and implementation of the system.Model building is the core of the whole system.In this thesis,according to the RedDial data set,the sentiment analysis dataset is created,and the BiGRU and BERT word embedding are used to get the user's movie preferences in the dialogue,the model obtains the F1 score of 0.8362.The whole model includes sentiment analysis,AutoRec based recommender model and dialogue model.The system has realized six functional modules:user management,authority management,film statistical display,random movie recommendation,movie data acquisition and dialogue movie recommendation.The flask framework and Vue framework are used to complete the development of the system's backstage and foreground.Finally,the functional requirements and non-functional requirements of the system are tested.Test cases are designed for the main functional modules,and the test system can achieve the functions in the requirements;through JMeter,the system performance is tested from three aspects of throughput,exception rate and average corresponding time,and the system performance can meet the user requirements.In this thesis,a novel movie recommendation system is implemented,which has certain reference value and use value.