Extracting Places from Location Data Streams

Current wayfinding research usually addresses the question of how people navigate, orientate and how they can be supported in unfamiliar environments. This scenario is important to understand the underlying concepts of wayfinding and to identify general principles applicable in wayfinding assistance. However, in every day live we usually perform wayfinding tasks in partial (not every place and path is known) familiar environments (e.g., if we look for the address of a particular doctor, shop or agency). Recently, as the location awareness of mobile devices constantly increases, people get interested in analyzing location data to extract spatial user profiles for location based services (e.g., [1,2,3,10,11]) or diary applications [4,5,8]. But only very few contributions attack the question of how familiarity with an environment and its mental representation can be captured, represented and used for wayfinding assistance (e.g., [12]) and, to the knowledge of the authors, no available wayfinding assistance system is able to integrate previous personal environmental knowledge. All systems implicitly assume the user to be completely unfamiliar with the present environment. This assumption does not lead to wrong results, but it disregards cognitive and representational benefits for the user. If a system knows about the “spatial signature” (a unique set of places like a user's home, his work, his grocery, his cafés where he meets his friends, the kindergarten of his kids, ...) it can use this previous knowledge as a reference frame for personalized assistance. This spatial signature can be used as a personalized configuration for navigation assistance systems, mixed reality applications (like ubiquitous gaming), profile matching and scheduling applications. Integrated in a mobile device, like a mobile phone, such an assistance system can generate location sensitive route directions and maps based on the individual reference frame: a user's meaningful places and paths between them. In the following example we assume a person working at the university of Bremen (the black dot on the map in Fig. 3). While being at home, he is looking for an unknown address close to the university. A query using Google Maps [13] results in route directions (Fig. 1) and a map displaying the route (Fig. 2). If a system integrates the previous knowledge of the user it can identify the destination to be close to his work place (the red dot in Fig. 3). The system also knows from where the user usually approaches his work (the blue line in Fig. 3). This way, the user can be presented with “lightweight” but meaningful maps (Fig. 3) and route directions referring to the user's personal landmark “workplace”, like “pass your workplace and turn left into EnriqueSchmidt-Straße”.