Researching for GIS annotation: The design of a conversational annotation interface

Using maps has become a daily activity for virtually everyone. However, in terms of interactive maps are limited. Rich interaction, such as issuing a complex query: “Which route has more greenery for my evening run?” is currently impossible to process for most Geographic Information Systems (GIS). To enable such rich interaction, data in GIS needs to be extended with high-level semantic annotations. Such annotations, due to their intrinsic semantic nature, can currently only be supplied by humans, who might moreover also need to have some expert knowledge about GIS in order to produce useful annotations. Given the cognitive resources required, the number of annotations to be made, and the tediousness of the task, there is a need to design systems that can support experts in completing such annotation tasks. We report on a study comparing two alternative systems for performing GIS annotations, a Conversational User Interface (CUI), designed for the purpose of this study, and a Web interface with NN GIS, which reflects the current state-of-the-art used by experts. We find that CUIs can indeed help to improve the accuracy and efficiency of the annotations. We conclude the paper with some tentative design principles for annotation systems coupled to complex user interfaces.