Research And Realization Of Team Construction In Social Network

Team formation in social networks is finding a team of experts who can complete a given task effectively. In real life, the task will be influenced by many points, such as the time and the budget and the wok-flow. These points are very important, but previous research work doesn’t consider these points. In this paper, we add these points in team formation problem.1.Team formation with a time limit. To the team work effectively, the task will make a time limit, so, the time is very important. We add a time limit for every skill, also we only consider the experts who can do skills in the specified time. To complete a task experts should communicate with each other, which will cost some time too. This paper is the first to consider these two elements and define a new cost model TotalCost. Based on the new cost model, we propose two methods. The objective is to find a team not only covers the skills that the task requires as well as completes every skill in the specified time and minimize TotalCost. Experimental results show that the team built by our methods satisfies the given task and with a low TotalCost.2.Team formation with a budget. Team formation is an important research question in real life. The task in practice will not only have a time limit but also work-flow and the budget. So we consider the time and work-flow and the budget in team formation. The time is the experts complete the skill cost and the work-flow is the order the skills to be completed, the work-flow is different, the time cost will be different. The budget we define it as time cost budget and communication cost budget. The time budget is the team completes the task within the budget. The communication cost budget is the experts communicate each other within the budget. We propose a new model called TotalCost which contains both the time completing skills and the time communicating among experts. The objective is to find a team not only covers the skills with work-flow of the given task but also completes the task with the least TotalCost. Our experiments were on the real DBLP datasets and our algorithms on them get very good results.