Evaluating automatically generated location-based stories for tourists

Tourism provides over six percent of the world's gross domestic product. As a result, there have been many efforts to use technology to improve the tourist's experience via mobile tour guide systems. One key bottleneck in such location-based systems is content development; existing systems either provide trivial information at a global scale or present quality narratives but at an extremely local scale. The primary reason for this dichotomy is that, although good narrative content is more educationally effective (and more entertaining) than a stream of simple, disconnected facts, it is time-intensive and expensive to develop. However, the WikEar system uses narrative theory-informed data mining methodologies in an effort to produce high-quality narrative content for any location on Earth. It allows tourists to interact with these narratives using their camera-enabled cell phones and an innovative interface designed around a magic lens and paper map metaphor. In this paper, we describe a first evaluation of these narratives and the WikEar interface, which reported promising, but not conclusive, results. We also present ideas for future work that will use this feedback to improve the narratives.