Strategies in the Use of Referring Expressions to Describe Things Urban

In the context of wayfinding technologies, there is increasing interest in dialogue based systems that use description of landmarks as a way of guiding people through cities. In the absence of maps and photographs, the challenge for automated systems is the production of descriptions of things in the field of view that are unambiguous and easily interpreted. We are therefore interested in the mechanisms used by humans to create and interpret descriptions of things in the urban vista. Here we report on a web based experiment in which we explored the veracity of human generated referring expressions in order to better understand the most successful strategies for directing people’s gaze.