Research And Implementation Of Group Recommendation Model Based On Neural Network

As group activities are common in people's daily life,recommending content to a group of users has become an important task in many information systems.Different from single person recommendation system,an important problem in group recommendation is how to aggregate group preferences and infer group decisions.Group preference is influenced by many factors,which is composed of individual preference,but it is not a simple accumulation of individual preference.There are also mutual influences among group members.Therefore,the traditional group recommendation system fuses the preferences of group members through static method,which is difficult to model the complex group decision-making process,resulting in poor recommendation effect.A good group recommendation model should be able to dynamically integrate member preferences from the data to meet the preference needs of more members as much as possible.In view of the above analysis,based on the latest development of attention network and neural collaborative filtering(NCF)in the field of recommendation,this paper proposes a new solution ncf-agree to solve the preference aggregation problem.Specifically,the main research contents of this paper are as follows:(1)Research and analyze the related technologies of group recommendation system,including attention mechanism,deep learning,collaborative filtering,etc.This paper summarizes the research status of group recommendation and the basic principles of related technologies.(2)In this paper,we propose a group recommendation model(ncf-agree).This model can dynamically learn the mutual influence among group members under different candidate topics,that is to say,the mutual influence among members is not invariable.Through NCF-AGREE model,we can recommend more suitable items for the group to meet the preferences of more members.In this paper,attention mechanism and NCF are integrated,mainly through two serial attention neural network modules for group preference modeling,which can learn the potential interests of users and the interaction of users in the group on specific items.First,the attention network is used to mine the influence of members on different candidate items,and the user item feature embedding vector is formed.On this basis,the second layer attention network is used to dynamically learn the influence of each user on other members in the group.Finally,under the NCF framework,the interaction between groups and projects is learned from the data,and the group prediction score is obtained.(3)Experiments on two datasets verify the effectiveness of the model.The experimental results show that,compared with the baseline model,the NCF-AGREE model proposed in this paper achieves better results in three evaluation indexes.(4)Design and implement the general group recommendation system,and integrate the algorithm model into the system.