Validation of the city metaphor in software visualization

Abstract

The rapid developments in computer technology has made it possible to handle a large amount of data. New algorithms have been invented to process data and new ways have emerged to store their results. However, the final recipients of these are still the users themselves, so we have to present the information in such a way that it can be easily understood. One of the many possibilities is to express that data in a graphical form. This conversion is called visualization. Various kinds of method exist, beginning with simple charts through compound curves and splines to complex three-dimensional scene rendering. However, they all have one point in common; namely, all of these methods use some underlying model, a language to express its content. The improved performance of graphical units and processors have made it possible and the data-processing technologies have made it necessary to renew and to reinvent these visualization methods. In this study, we focus on the so-called city metaphor which represents information as buildings, districts, and streets. Our main goal is to find a way to map the data to the entities in the fictional city. To allow the users to navigate freely in the artificial environment and to understand the meaning of the objects, we have to learn the difference between a realistic and an unrealistic city. To do this, we have to measure how similar it is to reality or the city-likeness of our virtual creations. Here, we present three computable metrics which express various features of a city. These metrics are compactness for measuring space consumption, connectivity for showing the low-level coherence among the buildings, and homogeneity for expressing the smoothness of the landscape. These metrics will be defined in a formal and an informal way and illustrated by examples. The connections among the high-level city-likeness and these low-level metrics will be analyzed. Our preliminary assumptions about these metrics will be compared to the opinions of users collected by an on-line survey. Lastly, we will summarize our results and propose a way to compute the city-likeness metric.