This study focuses on the bipartite network structure that describes social interactions. We analyze the characteristicsof the projection of bipartite network by comparing them with randomized models (BiCM1 and BiCM2)with different constraints. By evaluating the number of motifs, such as triangles, triplets, and cycles, as well as globalclustering coefficients and degree correlation coefficients, for both real and randomized data, we explore the originsof structural characteristics in social networks. The results reveal that real data exhibit a heterogeneous structure ofmultiple overlap cliques, contributing to the strengthening of clustering and degree correlation. In addition, positivedegree correlations within cliques and the frequency of six-cycle appearances suggest the potential influence of socialand psychological factors that simple structural constraints cannot fully capture. These analyses provide a foundationfor understanding the structure of bipartite networks and quantifying preference effects underlying real-world data.
Fujiki et al. (Sat,) studied this question.