Extracting Navigation Hierarchies from Networks with Genetic Algorithms

Information networks are nowadays an important source of knowledge, indispensable for our daily tasks. Because of their size, however, efficient navigation can be a challenge. Following the idea to use network hierarchies as guidance in human as well as algorithmic search processes, this work focuses on the creation of optimized navigation hierarchies. Based on an established model of human navigation, decentralized search, we defined two quality criteria for network hierarchies and propose a genetic algorithm applying them. We conducted experiments on an information as well as a social network and analyzed the optimization effectivity of our approach. Furthermore, we investigated the structure of the resulting navigation hierarchies. We found our algorithm to be well-suited for the task of hierarchy optimization and found distinct structural properties influencing the quality of navigational hierarchies.