网络内的派系团体比普通的联盟具有更高的联系紧密度，有助于进一步共享知识和合作研发。目前的研究较少关注派系团体与对团体创新能力的影响。本文基于中国9个高科技行业的联盟数据，并计算派系团体的派系度。派系团体内的构成越复杂，派系度越大。我们使用派系过滤算法和负二项回归模型，研究了最基本的3-派系-团体对于派系团体的创新能力的影响。研究结果表明，派系度与派系团体的创新能力之间呈倒U型关系，而派系团体外的企业数量会对这影响过程产生抑制作用。本研究对于企业联盟结派时选择派系团体梯队时，有一定的借鉴作用。

Communities in the network have closer ties than general alliances, which contribute to share of knowledge and collaborative research and development. Current studies are less concerned with clique communities and their impact on communities’ innovation. Based on the data of 9 high-tech industries in China, the clique-degrees are calculated in this paper. The more complex the cliques are, the larger the clique-degrees are. The clique percolation algorithm and the negative binomial regression model are applied to study the effect of the basic 3-cliques communities on the innovation capability. The result shows that there is an inverted U-type between clique-degrees and the innovative ability of communities, while the external enterprise number will inhibit the process. The study can be used as a reference for the choice of cliques’ communities in the enterprise alliance.