Research And Implementation Of The Online Team Formation With Skill Level And Work Time Constraints

Team formation has long been acknowledged as a problem of finding a group of experts from a candidate set and gathering them in a cooperative way to complete the given project or task.Through two decades of research,team formation problem has been fully developed.The basic model of team formation has been extended into a variety of variants that are closer to practical scenes,making this study more complete and more flourishing.With the business requirement in real word being more elaborate,a team covering all the required skills has not been satisfied by a project,which makes further demands for the completion quality of a team.At present,the team-formation has large application scenarios in online labor markets,paper cooperation network,paper review and other fields.Thus,the further study on it is of theoretical significance in science,along with great application value in practical businesses.In this thesis,we consider such a setting where a given project P is described by a set of required skills,each of which is specified by a skill level and work amount,indicating the minimum skill grade and the needed work hours the project requires for that skill module respectively.In the meantime,each expert in the candidate set is specified by an available work time and a set of skills that he possesses,each of which is associated with a level value to show his skillfulness on this skill.Under aforementioned circumstance,this thesis proposes a novel team formation model where the skill level and work hours are constrained.To the best of our knowledge,this is the first work that takes both the skillgrading and the work amount into consideration,which means that our work has certain innovation.In our work,we aim to find a team along with a corresponding division scheme for the above described project P to make the completion quality of the team towards this project P as high as possible.We refer to this as Quality-Team Formation problem or QTF for short.To specifically describe the concept of completion quality,this thesis defines a function of team score Tscore()to measure to what degree the given project is completed by a formed team.In addition,the proposed QTF problem is proven to be NP-hard,and the detailed analysis of its computational complexity is presented in this thesis.To solve the QTF problem,two heuristic approximation algorithms Skill First and Expert Assign are provided.This two approaches start from the objective function Tscore(),centering on skill and expert respectively,to select every member one by one to join in the team.In this thesis,extensive experiments are conducted on two real-world datasets to demonstrate the performance of our algorithms.Experimental results show that this two algorithms are both effective and practical to automatically build teams for various projects,each with its strength.By comparison,Skill First could form teams with higher team score and greatly outperform Expert Assign on time efficiency.But Expert Assign has big advantage in producing tidy and compact teams with smaller size than Skill First,which can obtains the preference from employers.