摘要
科技项目培育或评审过程中,往往同时有多个项目和多位专家.每个项目都有限制被匹配的专家数量,每个专家又有最多容许匹配的项目数量.目前大多数的匹配过程中,专家多由人为确定或随机选取产生,往往导致专家与所评项目内容不匹配或相关度较低的现象.对于项目与专家网络具有的复杂网络特性,本论文着重考虑到项目与专家网络中同时存在的高聚类和小世界现象,将项目与专家进行抽象,从网络节点的关联性出发,利用二分图网络流模型,提出两种贪心匹配策略与一种传统最大流匹配策略的组合策略,设计出项目与专家的多重匹配算法.本文最后采用电力行业数据集进行多次实验,验证该策略可以有效应对项目专家网络,在计算耗时和匹配结果上都较传统网络流算法高效.
In the procedure of breeding or evaluating technology projects,there are always many projects and experts. Each project can only be matched by constant number of experts while each experthas their maximum matching capability. Nowadays the experts are selected manually or randomly in most time which may cause mismatching or lowcorrelation between projects and experts. Sincethis projects and experts networkhas the characteristic of the complex network,this paper focuses on the phenomenon of high-level clustering and small world,abstracts the projects and experts into network nodes,takes usage of bipartite graph network flowmodel based on the correlation between network nodes,and puts forward a combined multiple matching strategy of two greedy strategies and one traditional max-flowmatching strategy. We use the power industry database to preform several experiments in order to evaluate the efficiency when dealing with this kind of project and expert networks. The results showthat this strategy is more efficient than the conventional network flowstrategy both in time and results.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第3期545-550,共6页
Journal of Chinese Computer Systems
关键词
复杂网络
多重匹配策略
二分图
网络流算法
complex network
multiple matching strategy
bipartite graph
network flow algorithm
