Dealing with incomplete and uncertain context data in geographic information systems

There are currently a growing number of people using smartphones or tablets, thus being potentially online at every moment. There are many useful applications using people's context data, to provide services to mobile telephony/internet subscribers. Location data is particularly interesting. These applications use location data assuming it is correct, which is sometimes not the case. In this work we propose a methodology for using incomplete/uncertain information to answer questions which include uncertainty like: “Which is the probability of finding exactly N persons within the geographic area A from time Tl to time T2?”, or “Which is the probability of having a traffic jam on street S between times Tl and T2?”. We also consider some logical constraints on the data. For instance: “Exclude counting people on the subway or inside buildings because the advertising will be on screens at open air”. Our approach uses Dempster-Shafer theory combined with an ontological definition of variable types sharing similar probabilistic behavior. The whole process and the results are explained using an example case based in one of the busiest areas of the world (the Shibuya Station in Tokyo, Japan), consisting of underground train lines, surface transportation, large avenues and shopping centers. A language to describe the fuzzy scenarios is also introduced along with an application which allows users to generate and visualize 2D and 3D suitability maps using this language.