Research On Multi-user Query And Concurrency Control Based On Group

More and more users want to share resources to enhance the cost under the premise of meeting their own needs,such as car-pooling,apartment-sharing,group tour and so on,which presents a new challenge to multiuser query and update based on group.When users query based on group,the group query algorithm needs to meet not only the user's preference for the query object,but also the user's preference for the members of the group,and to improve query efficiency as much as possible.The existing group query algorithms can not meet these requirements.When users perform update operations based on group,traditional methods can not solve the problem of concurrency within and between groups,therefore,the research on concurrency control based on group has important signif icance.Based on the study of current multi-user group query and concurrency control,we study multi-user preference query and concurrency control based on group.The main contributions of this paper summarized as follows:(1)The existing grouping algorithm does not consider the effect of user 's preference for other members in group.For this problem,we propose a multi-user preference query based on group algor ithm(PQBG)which considering the preference among members in the group.The query processes are divided into four steps: Firstly,generate the primary result set from all query users based on Euclidean distance.Secondly.calculate users' preference similarity of query objects and generate primary queue in descending order.Then using the group's attribute-set and preference-set to filter out the partners in the same group.Finally,generate a query result set for the group.We use the real-world datasets to compare the algorithm PQBG with the other three query algorithms.The results show that PQBG algor ithm has higher query performance in query time and user satisfaction.(2)As the existing concurrency control can not solve the problem of concurrency control between groups,we propose the GB-2PL locking mechanism and the concept of "group-lock".This locking mechanism can resolve the concurrency conflicts among groups and finally back to the user satisfing results.On this basis,we describe the rules of the GB-2PL locking mechanism and the locking manager 's data structure,and describe the algorithm of the GB-2PL mechanism.In addition,we prove the feasibility of GB-2PL mechanism.The simulation results show that the GB-2PL locking mechanism can solve the problem of concurrency control between groups effectively,and can improve the success rate of groups.(3)Based on the two preceding studies,we design and implement the carpool query and concurrency processing system: CQCPS,which can manage the users query and concurrency control based on group.We introduce the design idea,architecture of the system and the design of modules in detail.Compared with the system which does not divide users into groups and concurrency control by groups,the results show that CQCPS system has better performance.