Abstract: In crowdsourcing design projects of complex products, designers are often required to form teams to continuously interact and collaborate to complete relevant tasks. In order to solve the problem of different preferences among team members during team formation, this paper proposes a team discovery algorithm called S_Louvain based on the combination of member preference and structure similarities, which considers the preferences among team members and improves the modularity index. The preference attribute similarity and the topology structure similarity of nodes are first calculated. The candidate node set of the target team is then expanded combining the nodes given by users and their neighbor nodes with consideration of their preference and structure similarities. With the candidate node set as the core, the interests and preferences of a design team are mined to calculate the improved modularity and update the optimal team division. The experimental results on public datasets and crowdsourced engineering instance datasets show that the modularity index of team division is improved, which verifies the feasibility and practicability of the algorithm proposed in this paper.

Key words: crowdsourcing design, user preference, similarity, modularity