Chengyi Xia

On Structural Properties of Large-Scale Software Systems: From the Perspective of Complex Networks

From the viewpoint of network, large-scale computer software system scan be regarded as complex networks composed of interacting units at different levels of granularity (such as functions, classes, packages, source files, etc.). In this paper, the collaboration relationships between header files in the source node of Linux kernels, which are representative examples of large-scaleopen-source software systems, are analyzed by constructing weighted network-Header File Collaboration Network (HFCN). Through using appropriate non-weighted and weighted quantities, the complex structural properties, the weight distribution and the impact between them of these networks are characterized and analyzed. These results can provide a better description of the organizational principles at the basis of the architecture of source codes in large computer software systems.

Mining Important Topological Properties in Large-Scale Computer Software Systems Based on Complex Networks

Recently, large-scale computer software systems have attached a great deal of attention when they are regarded as complex networks composed of interacting units. In this paper, the collaboration relationships between header files in the source code of Linux kernels are analyzed by constructing a weighted File Collaboration Network(FCN): each node represents a header file, two nodes are connected if corresponding header files are both included in the same source file at least once, also the link weight is assigned to evaluate the intensity of co inclusion of two header files. Through using appropriate non-weighted and weighted quantities, structural properties of FCNs are characterized and analyzed. The study of large-scale softwares from the viewpoint of complex networks can provide a better description of the organizational principles and evolving mechanism of complex software systems.

Generalized collaboration networks in software systems: a case study of Linux kernels

The collaboration relationships between header files in the source code of Linux kernels are analyzed by constructing a weighted Header File Collaboration Network (HFCN): each node represents a header file; two nodes are connected if corresponding header files are both included in the same source file at least once; also the link weight is assigned to evaluate the intensity of co-inclusion of two header files. Through using appropriate non-weighted and weighted quantities, structural properties of two kinds of HFCN networks(HFCN-I and HFCN-II) are characterized and analyzed. The study of Linux kernels from the viewpoint of complex networks can provide a better description of the organizational principles and evolving mechanism of complex software systems.