Integrating Knowledge Graphs and Complex Network Analysis for Effective Team Formation in New Product Development Projects
提出一种结合知识图谱与复杂网络分析的方法，通过识别相似项目、计算成员依赖强度并改进社区发现算法，为新产品开发项目高效组建团队并划分子团队。
Forming an effective project team to rapidly develop new products is a significant challenge that involves selecting project members and assigning them to sub-teams. This study proposes an innovative method for team formation by combining digital approaches (i.e., knowledge graphs) with traditional modeling methods (i.e., complex network analysis). First, to search candidate members for a target new product development (NPD) project, we build the NPD project knowledge graph and identify completed projects with high similarity to the target project from a complex network perspective (i.e., structural similarity and attribute similarity). We select the members of these completed projects as the candidate team member set for the target project. Second, to calculate the dependency strength among members, we match members and activities based on expertise consistency and propose a method for calculating dependency strength under two scenarios by building a “activity-organization” network. Finally, to partition the team network into several sub-teams and identify overlapping members, we propose an improved Linkcomm algorithm to cluster the network. An industrial case validates the proposed method. To ensure research rigour, we conduct robustness checks (i.e., parameter test and added-variable test), sensitivity analysis and comparative experiments against the traditional model. This paper contributes to both theory and practice of project team formation.