Discovering the right place to check-in using web-based proximate selection

With information technology becoming increasingly embedded in our everyday physical world, there is a growing set of mobile applications that involve a connection with the digital representation of physical places. This association is normally initiated with a check-in procedure, through which a person asserts her presence at a particular place and determines the context for subsequent interactions. The common assumption is that a mobile application will be able to search the surrounding environment and present the user with the intended check-in target; however, in a world of ubiquitous place-based services, this assumption may no longer hold. A person in an urban environment would, at any moment, be surrounded by a large number of places, all of which could be regarded as possible interaction contexts for that person. In this work, we investigate the real-word challenges associated with wide-scale place selection and how the process can be affected by the place environment, by the position of the person in relation to the target place and by positioning errors. To study this reality, we used Google Places as a directory of georeferenced places. We conducted 14,400 nearby place queries structured around different combinations of our three independent variables. The results suggest that the overall performance is poor, except for low-density scenarios, and that this discovery process, albeit relevant, should always be combined with other place discovery approaches. The results also help to understand how this performance is affected by check-in positions and by the properties of the place environment.