Imagine you witnessed an accident and needed to call an ambulance, but you did not know where you were. How would you describe your location to the operator? You might mention a nearby landmark, how many minutes drive you are from a main street or, if you are in an unfamiliar city, you may only be able to describe the colours of the buildings around you.
Associate Professor Stephan Winter, from the Melbourne School of Engineering’s Department of Infrastructure Engineering, is leading a research team of experts in linguistics, computer science and spatial information science to try and bridge the gap in the way we explain locations to our friends and how we type descriptions into mapping services such as Google Maps or in-car navigation services.
While engineering researchers are providing expertise in mapping, spatial information, information technology and programming, Associate Professor Lesley Stirling from the School of School of Languages and Linguistics, at the University of Melbourne, is providing her expertise in analysing language structures, and searching for similarities and differences in the way people describe places to help the team extract as much data as possible from even the vaguest place description.
“Every day we need to explain where people or objects are in the world, but we do not have any applications that can understand the varying, complicated and sometimes strange ways we do this. This project aims to develop new methods for capturing the way we describe locations and create automated systems for understanding place descriptions,” Associate Professor Stirling says.
The dilemma arises because while we often provide vague descriptions involving landmarks and our proximity to things when describing place, when we are asked to type a location into Google Maps or a car navigation system, we are required to provide exact addresses and very specific information. Computers struggle to translate the simple and natural language we use to describe locations, and we need to overcome this obstacle in order to enable smarter navigation and mapping services. Improving computer recognition of natural language will not only lead to more convenient and effective communication, but will also assist in emergency situations and could potentially save lives.
The research team was recently awarded over half a million dollars by the Australian Research Council and project partners the Emergency Services Telecommunications Authority (ESTA), Land and Property Management Authority NSW, PSMA Australia and the Surveyor-General of Victoria. To assist with the research, the team has devised a mobile phone game; users simply log in and confirm on a map, where their phone has located them, and type in a description of where they are. These descriptions are used by researchers to gather the words we use to describe a place.
Rod Tucker, Laureate Professor from the University of Melbourne’s Department of Electrical and Electronic Engineering and Director of the Institute for a Broadband-Enabled Society, which funded the development of the mobile game says crowd sourcing said that getting large amounts of data from the public is one of the exciting new ways to participate, engage and contribute to a larger good; and one reason to assume that broadband will change our current economy fundamentally.