The Research And Implementatio--N Of Activity Recommendation Model For Random Groups

With the rapid development of the Internet,many users choose to share their lives in social networks such as forums,post bars,and short video live broadcast platforms.In these network platforms,more users share information in the form of groups,and the demand for group recommendation is increasing day by day.How to accurately obtain the preference representation of members in the group and use a reasonable preference aggregation model to describe the group preference characteristics is the core problem of group recommendation;many scholars have proposed some effective solutions in the research of this problem,but for random The recommendation problem of group-type groups has not been studied in depth,but in interest forums and social platforms,random-type groups are more active.Such groups have the characteristics of an unfixed number of members in the group,large differences in interests,and unclear membership among members.The social connection and other characteristics of the platform,and the surge in the number of platform users has led to a gradual increase in the size of the group.This paper mainly focuses on the research of group member preference representation and group preference aggregation strategy in group recommendation problem;in addition,diversity recommendation is also one of the research contents of this paper.Diversified content can increase users’ satisfaction and experience,and at the same time The efficiency of information flow in the system.Therefore,this paper studies and analyzes the group preference aggregation strategy for two random groups with different characteristics,and conducts an in-depth study on the diversity recommendation combined with practical application scenarios.The main content of this article includes the following parts:(1)This paper proposed a group recommendation model based on membership relationship and comprehensive influence,which improved the traditional clustering algorithm,modeled the group feature representation based on the multiple relationships and comprehensive influence among the group members,and introduced the trust between members to improve the traditional recommendation algorithm.Experiments on two real data sets show that the proposed model has better performance than the benchmark model.(2)Double relation is proposed in this paper a kind of fusion of preference aggregation strategy group recommendation model,using the neural since note network mining the deep implicit relationship between the group members said that with the characteristics of group used at the same time attention mechanism by excavating the history of the team’s members interactive project and the correlation between the target project depicting the contribution weight parameters of members;Through the interactive relationship between learning groups and projects,the items in the recommendation results are sorted and optimized by the loss algorithm based on logistic regression.Finally,experiments on open data sets show that the model has better recommendation effect than other group recommendation models because of its innovative aggregation strategy.(3)This paper proposes a recommendation based on user preferences diversity distribution model,the model uses Epanechnikov kernel density estimate the user’s interest distribution of diversity,at the same time,the same user on different projects are analyzed the history of the score difference of user preferences to express the degree of the influence of the weight parameters of the users of the project study reorganizes the suggestion list;The results show that the proposed model achieves a balance between accuracy and diversity of recommendation results,and improves user satisfaction.(4)In this paper,a diversity group recommendation algorithm based on greedy strategy is proposed.The target problem is defined as a twocriteria optimization problem.After k iterations of greedy strategy,the final diversity recommendation list is solved,and a comparative experiment is conducted by simulating random groups in two scoring data sets.(5)This paper analyzes the characteristics of the proposed four models,and on this basis,designs and implements a group activity recommendation system for different group characteristics and individual users.The group recommends activities,which improves the accuracy of the recommendation,and at the same time,it can recommend a variety of activities for users,which improves the user’s sense of experience and satisfaction.