Software Systems as Complex Networks

Abstract

networks [1][18]. Those networks include both those occurring in nature, for example, neuron activity, cellular metabolisms, protein folding and social As software systems become larger and more networks, and those created by people, for example, complex, in order to understand, manage and evolve electricity supply networks, the web and the internet. these systems, we need better ways of characterizing Even though they represent totally different types of and controlling their macroscopic properties. We systems, they share many common properties such as suggest complex network theory may be useful for power-law degree distribution [1], small world these purposes. In recent years, researchers have property [19] and high level of clustering [2][3]. The shown that many complex systems from different similarity between different complex networks inspires disciplines can be investigated as complex networks the idea that network growth mechanisms represent and most of them comply with a scale-free network some basic natural tendency to create order from model. We explore the view that a software system can chaos. be studied as a network with a number of components (classes) connected by dependency (integration) The recent progress in complex network theory relationships; we call this network the Component motivates us to study software systems as complex Dependency Network (CDN). The CDNs of several networks. A software system can be treated as a Java libraries and applications have been examined network of components connected by dependency and all of them exhibit some scale-free characteristics. relationships [7]. The importance of relationship is This result has some practical value including that it stressed by the Object-Attribute-Relation (OAR) model allows us to identify important components (classes) [4], which shows that human memory and knowledge and thereby assists software maintenance and are represented by connections of synapses between reengineering. We have built a tool to study software neurons, rather than by the neurons themselves [5]. systems as complex networks. In the paper we also Similarly, for software systems, some high level suggest ways of controlling and changing how systems functions are realized by the relationships of low level evolve in order to improve theirunderstandability and components. In this paper, the term component maintainability, indicates an abstract form of software or even hardware entities as in the behavior tree approach [6].

Investigación en Progreso: Estudio del Comportamiento Dinámico del Diseño de Sistemas de Información Basado en Redes Complejas

“…A medida que los sistemas software crecen en tamaño y complejidad se necesitan mejores métodos y herramientas para poder caracterizarlos y estudiarlos, de forma tal de tener la capacidad de entenderlos y gestionarlos eficientemente [33]. Las métricas de software cumplen un papel esencial a la hora de entender los sistemas y administrarlos a lo largo de todo su ciclo de vida [7].…”

“…Del cuerpo de conocimiento desarrollado en el área se concluye que los sistemas software se comportan como redes libres de escala (scale-free), de forma análoga a la mayoría de sistemas creados por el hombre [1] [33].…”

studio del Comportamiento Dinámico del Diseño de Sistemas de Información basado en Redes Complejas

“…El número arcos o aristas de cada nodo (el grado del mismo) seguirá por ende una distribución normal [98]. Este modelo se mantuvo vigente hasta el aumento en la capacidad de cómputo y el desarrollo de grandes bases de datos, avances que llevaron al desarrollo de nuevos conceptos [2].…”

“…Las redes que cumplen con una función de distribución potencial del tipo son llamadas redes libres de escala (scale-free) ya que en ellas existe un pequeño conjunto de nodos, llamados hub, con un número de conexiones (arcos o aristas) mucho mayor al promedio [98]. 1.…”

“…2) Sistemas software como redes complejas Un sistema software puede ser tratado como una red compleja de componentes conectados por relaciones de dependencia, red a la que se denomina CDN (Component Dependency Network o red de dependencia de componentes), y que representa una vista de arquitectura del sistema software [98].…”