1.5.1. Making Maps Talk

Introduction

The final method of portrayal that we will consider will be how geographic information can be presented users of location-based services in the form of text or audio. There are a number of reasons for doing this. There are many situations where the user does not want to be distracted by looking at a map, this is most evident in navigation, where the user wants to focus on the task at hand, e.g. driving, with the minimum of interuption. (Streeter et al. 1985) for example has shown in such situations that navigation is more effective if verbal instructions are given than if a standard map is provided. Other examples might be in following a guided historical tour when the user wants to look at the interesting features of the landscape rather being distracted by a mobile device. (Bartie et al. 2006) describes such a system that provides audio descriptions of historic buildings in Edinburgh that can be seen by the user as they walk around the city.

Changing the medium of the map in such a way is a challenging problem because it requires the model of information to be shifted from one that is spatial, e.g. symbols on a two-dimensional planar map, into one that is based on narrative. As such natural language based representations tell a story which involves a sequences of events. We will look at how this story can be automatically developed for the example of in-car or pedestrian navigation. This starts with a geometric description of a route and ends up with a set of spoken instructions describing how the route is navigated.

We can think of the process of generating a route description as consisting of a translation between two models. On the one hand, there is a spatial model that represents information about geographic features geometrically. On the other, there is a language model that represents geographic features in terms of the roles they play within a natural language sentence.