La spatialité algorithmique : apports, limites et réductions de la personnalisation algorithmique dans l'assistance à la navigation et au wayfinding
|Advisor:||Roche, Stéphane; Lussault, Michel|
|Abstract:||This Ph.D. research focuses on navigational and wayfinding aid provided by tools such as GPS systems and webmapping platforms (e.g. Google Maps). Cognitive sciences, which constitute the theoretical and conceptual foundation of this thesis along with human geography, show that optimal navigational instructions necessarily include landmarks along decision points (i.e. intersections where people have to change direction according to the scheduled itinerary). In addition, wayfinding assistance can be considered as truly effective only if users have (or acquire) a minimal spatial knowledge of the area they plan to visit. The Big Data phenomenon is now on the rise so that algorithmic personalization is considered as a standard; a feature that « smart » systems should offer. In this context, some researchers advocate the personalization of navigational and wayfinding assistance. However, this statement must be empirically supported. Indeed, do GPS users necessarily rely on such tools when they travel within familiar areas? In which case, do they need some navigational instructions that systematically include a familiar landmark along each decision point? In the same vein, do they really need a customized route for planning their travels on webmapping platforms? Actually, assessing the impact of spatial familiarity on both the landmark selection (during navigation) and the route planning (through a webmapping platform) is the main objective of this research. Specifically, the thesis is organized into two main parts. The first one is quantitative per se and investigates through a virtual navigation experiment the significance of the relationship between spatial familiarity and semantic salience of urban places. The second component is more qualitative. On the one hand, it explores the uses of webmapping platforms from a set of semi-structured interviews. On the other hand, cognitive strategies of locals (inhabitants of the Province of Quebec) and strangers are obtained and compared through experiments based on the think-aloud method. In the end, this research indicates that places with a high semantic salience score are likely to grab locals’ attention. These people also tend to pick up personal and popular landmarks along the itinerary provided by the webmapping platform. Nevertheless, this approach restricts the acquisition of new spatial knowledge and affects users’ cartographic literacy at the same time. Those drawbacks lead me to argue that the improvement of navigational and wayfinding aid does not necessarily lie in the customization of the computed route and its directions. In my opinion, encouraging users to actively participate in the wayfinding process is a path that researchers in spatial cognitive engineering should explore.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||24 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.